i

PRAISE FOR
DATA-DRIVEN HR

Without doubt human capability (talent, leadership, organization and HR)

increasingly delivers value to all stakeholders. This excellent book provides
business and HR leaders the information required to improve decision making.
Its insights on analytics and AI will be the keys for progress.
Dave Ulrich, Rensis Likert Professor, Ross School of Business, University of
Michigan, and Partner, The RBL Group

If anyone was going to publish a book about the impact of the latest technology
developments such as AI on the field of HR and people analytics my bets were
on Bernard Marr. And you won't be disappointed. The book offers a deep dive
into the world of data of every kind, every possible use case, honest overview of
technology and important considerations. It has never been more critical to
educate ourselves about it.
Maja Luckos, VP, Employee Success, Salesforce

This book propelled me into a world of possibilities for HR leaders in embracing

the ‘intelligence revolution’ to shape people strategies that add value to their
organisations and their people. It’s enlightened me to the power of AI-enabled
HR and how I might use it, and it’s made me want to learn more. This is a
must-read for all HR leaders.
Linda Sleath, Group HR Director, Topps Tiles

Data-Driven HR strikes a nice balance between exploring emerging trends in

people analytics while primarily serving as a practical guide to HR professionals
at any stage of their data journey. This second edition seamlessly weaves AI into
a narrative that’s easy to engage with and is packed full of examples that bring
the theories to life.
Mark Ferrie, People Analytics Director, Meta

Data-Driven HR is a terrific overview of the enormous world of people analytics

and AI. For people trying to understand this important space, this book shows
you the way.
Josh Bersin, Global Industry Analyst and CEO of The Josh Bersin Company
ii

Data, analytics and AI provides ways to elevate HR from its traditional role as a
support function to one of a strategic partner creating value for the enterprise,
its customers and its employees. There’s a well-thumbed copy of the first edition
of Data Driven HR on my bookshelf, and in this timely update Marr, one of the
most knowledgeable people on the topic, explains how data and AI can enable
HR to drive better decision making about people, deliver an enhanced service to
employees and make HR processes more efficient.
David Green, Managing Partner at Insight222, co-author of Excellence in
People Analytics, and host of the Digital HR Leaders podcast.

Bernard Marr has once again delivered an indispensable guide to harnessing the
power of data, analytics, and AI in HR. This updated edition thoroughly captures
the latest innovations shaping human resources while still being accessible for
HR professionals at any level. Through compelling examples and clear
frameworks, Marr demonstrates how to drive business value through evidence-
based talent practices. This is a must-read playbook for any HR leader looking to
build capabilities in data-driven decision-making.
Professor Max Blumberg, University of Leeds

This is a great guide for HR professionals who are grappling with the transition
to becoming data led. It's easy to read, and with real examples and case studies
across the employee lifecycle, it's also a pragmatic resource to have in your HR
toolkit.
Matthew Mee, Talent Innovation, Vice President, Lightcast

Artificial Intelligence is here, and this is just the beginning. As always, Marr takes
a highly complex topic and decodes it in a way that it is understandable for
people at all levels – be it ‘people for people (HR)’ or the more technical
proficient experts who want to understand the people aspect of what AI brings
to table. As Marr explains with great simplicity it’s not all about the technology,
but about how we as humans learn to coexist, embed it in our day-to-day work
lives and more, answering critical questions and innovatively thinking about the
impact it will have on the future of our workforce, acceleration in building the
analytics capability and maturity across organisation is a MUST HAVE. Data-
Driven HR is an excellent build from his previous edition of this title.
Ashish Sinha Korn Ferry Head of People Analytics, AI & Strategy EMEA
Practice Leader
 iii

AI is transforming the world of work and our personal lives. With a people-
centric approach, Bernard Marr demystifies data driven AI enabled HR with
context, thought provoking insights and examples of AI at the time this book
was written. We all have a role to play when it comes to this rapidly evolving
space as the output of AI will be a reflection of our culture and values. Staying
on top of leading practices, lessons learned, emerging regulations and standards
is critical so we can unlock AI’s potential and value add to the business, our
customers and employees while minimizing risk. This book sets the foundation
so we can do just that.
Terilyn Juarez Monroe

Data-Driven HR is an indispensable resource for career services professionals

looking to equip their students with cutting-edge strategies in today's
competitive job market. This comprehensive book offers invaluable insights into
recruitment and candidate selection, employer branding, pinpointing the most
effective recruitment channels, and harnessing AI-enhanced automation to
identify and assess the best candidates for businesses. It's a game-changer for
career advisers committed to empowering their students with the knowledge
and skills needed to excel in the evolving world of talent acquisition and HR.
Amber Wigmore Álvarez, Associate Professor, IE Business School and IE
University, Spain
iv

THIS PAGE IS INTENTIONALLY LEFT BLANK

 v

Second Edition

Data-Driven HR
How to use AI, analytics and
data to drive performance

Bernard Marr
vi

Publisher’s note
Every possible effort has been made to ensure that the information contained in this book is
accurate at the time of going to press, and the publishers and authors cannot accept respon-
sibility for any errors or omissions, however caused. No responsibility for loss or damage
occasioned to any person acting, or refraining from action, as a result of the material in this
publication can be accepted by the editor, the publisher or the author.

First published in Great Britain and the United States in 2018 by Kogan Page Limited
Second edition 2024

Apart from any fair dealing for the purposes of research or private study, or criticism or review, as
permitted under the Copyright, Designs and Patents Act 1988, this publication may only be repro-
duced, stored or transmitted, in any form or by any means, with the prior permission in writing of
the publishers, or in the case of reprographic reproduction in accordance with the terms and li-
cences issued by the CLA. Enquiries concerning reproduction outside these terms should be sent
to the publishers at the undermentioned addresses:

2nd Floor, 45 Gee Street 8 W 38th Street, Suite 902 4737/23 Ansari Road
London New York, NY 10018 Daryaganj
EC1V 3RS USA New Delhi 110002
United Kingdom India

www.koganpage.com

Kogan Page books are printed on paper from sustainable forests.

© Bernard Marr 2018, 2024

The right of Bernard Marr to be identified as the author of this work has been asserted by him in
accordance with the Copyright, Designs and Patents Act 1988.

All trademarks, service marks, and company names are the property of their respective owners.

ISBNs

Hardback 978 1 3986 1458 1

Paperback 978 1 3986 1456 7
Ebook 978 1 3986 1457 4

British Library Cataloguing-in-Publication Data

A CIP record for this book is available from the British Library.

Library of Congress Control Number

2023948898

Typeset by Integra Software Services, Pondicherry

Print production managed by Jellyfish
Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY
 vii

To my wife Claire, the heart of our family, and to my children

Sophia, James, and Oliver, its soul.

You embody the essence of this book: that at the core of any
successful endeavour, even in a world driven by data and
technology, are people.
viii

THIS PAGE IS INTENTIONALLY LEFT BLANK

 ix

CONTENTS

01 How data and AI are transforming HR 1

HR and the ‘intelligence revolution’ 1
Defining intelligent HR 2
A (brief) word on automation 4
How to use this book 5
Key takeaways 6
Notes 7

PART ONE 
Laying the groundwork for
data-driven, AI-enabled HR

02 How data and AI have come to revolutionize HR 11

The emergence of HR and its early evolution 11
HR in the 21st century: The rise of people analytics 13
The changing technology landscape – and its impact on HR 15
The changing nature of work 17
New skills for a new era of work 19
So what lies ahead for data-driven HR? 19
Key takeaways 21
Notes 22

03 The data, analytics and AI tools available to HR 23

The amazing capabilities of analytics and AI tools
(and how to use them in HR) 23
Tapping into other data-driven and AI-enabled tech 34
The secret sauce: Data, data and more data 37
Key takeaways 47
Notes 48

04 Challenges and pitfalls of AI and data-driven HR 49

Understanding the value of data 49
The biggest data-related challenges 51
AI pitfalls 56
x Contents

Key takeaways 62
Notes 63

05 Finding the right ways to use data and AI in HR 65

Why it all starts with strategy 65
Begin with your objectives 66
Creating your strategy 69
Making the business case for data- and AI-driven HR 73
Returning to your strategy in the future 74
Key takeaways 75
Notes 75

PART TWO 
Data-driven and AI-enabled
HR in practice

06 Better HR insights and decision making 79

The role of HR analytics 79
The evolution of HR analytics 81
So, what can be measured and analysed with HR analytics? 83
Using HR analytics to inform decision making: Reporting
and dashboards 85
Key takeaways 87
Notes 88

07 Recruitment and candidate selection 90

Boosting your employer brand 90
Identifying the most effective recruitment channels for you 93
Identifying and assessing the best candidates for your business 94
AI-enhanced automation in recruitment 98
Recruitment in the metaverse era 103
Key takeaways 104
Notes 105

08 Improving employee onboarding 107

Onboarding in the AI era 107
Onboarding new hires in the metaverse 111
Key takeaways 113
Notes 113
 Contents xi

09 Performance monitoring and management 115

Lessons from the world of sports 116
Intelligently measuring employee performance 118
Intelligently reviewing employee performance (and not on an
annual basis) 120
Lessons from Amazon: How not to handle people monitoring
and reviews 125
Finding the right balance between trusting and tracking 128
Key takeaways 132
Notes 133

10 Data-driven employee training and development 135

How data and AI are positively disrupting the education
sector 136
What skills does your organization need? 138
The major trends in data-driven workplace learning 141
The cutting edge: Incorporating VR, AR, digital twins and the
metaverse 146
A quick look at the downsides of data-driven training and
development 151
Key takeaways 152
Notes 153

11 Employee safety and wellbeing 155

Improving employee safety with data and analytics 155
The cutting edge of workplace safety: VR and AR 161
Improving employee wellbeing and wellness 162
The cutting edge of wellness: Tapping into VR and the
metaverse 166
The potential downsides of data-driven employee safety and
wellness 167
Key takeaways 169
Notes 170
xii Contents

PART THREE 
Making data-driven and AI-enabled
HR happen

12 Identifying the use cases for your organization 175

Why talk about use cases now? 175
Identifying potential use cases 176
Fleshing out your use cases 178
Linking your use cases to your wider data/AI strategy 180
Learning from other businesses that have been on the data and AI
journey 181
Key takeaways 182

13 Building skills and aligning culture 184

What skills do HR professionals need? 184
Data literacy and HR 186
Addressing the data skills shortage 188
Preparing for a cultural shift in HR and the organization 193
Building an AI and data culture – practical steps 195
Key takeaways 197
Notes 198

14 Creating the technology and data infrastructure 200

Your HR technology strategy – and how it fits with the wider
business strategy 200
Putting in place the right data and AI infrastructure 203
Key takeaways 211
Notes 212

15 The future of HR 213

Future trends that will shape HR 213
Decentralized and fully remote work structures 214
Building the workplaces we want 216
Tell me what you think 217

Index 219
 1

How data and AI 01

are transforming
HR
We’re in the midst of a new industrial revolution – the fourth industrial
revolution, driven by an explosion in data, automation, intelligent machines
and hyper-connectivity. Like the previous industrial revolutions before it
(the most recent being the rise of computerization in the late 20th century),
this new wave of transformation will utterly change the way we live. And,
of course, the way we work.
The last few years have seen dramatic change for organizations across
almost all sectors. No doubt you’ve seen signs of this in your own job, as the
organization begins to take advantage of new technologies. That change is
only going to accelerate over the next five to ten years.
And yet, at the heart of our rapidly changing organizations are… people.
Companies are nothing without the right people. Even in this technology-
driven age, those organizations that are able to attract (and keep) the right
people are most likely to succeed now and in the future. As businesses navi-
gate rapid change, this need to attract and nurture talent is arguably more
important, not less. Now more than ever, people are the central driver of
success.
It’s therefore vital that human resources (HR) teams put in place the
intelligent systems and processes to find, recruit and retain the right peo-
ple for their organization. Which brings us to the need for data-driven,
AI-enabled HR.

HR and the ‘intelligence revolution’

I call this fourth industrial revolution the ‘intelligence revolution’ because
that’s exactly what it is – a revolution that’s creating an increasingly
2 How data and AI are transforming HR

s­ uper-intelligent, data-driven world. Artificial intelligence (AI) and the in-

telligence revolution will transform so much of our everyday lives, just as
the rise of computers did in the previous industrial revolution. (Maybe
even more so. Sergey Brin, co-founder of Google, has described AI as ‘the
most significant development in computing in my lifetime.’1)
Every business – and every function within the business – is going to have
to get smarter. This is equally true whether your organization is operating in
one of the more obviously contemporary industries, like the tech sector, or a
more traditional industry, like manufacturing. And it’s true whether you
work for a global corporation or a small start-up business.
The good news for HR professionals is that you have more data at your
fingertips than ever before. Now, almost everything we do at work can be
measured: from employees’ day-to-day actions, performance, happiness and
wellbeing to wider business operations. This, coupled with advances in ana-
lytics – largely driven by AI technologies – means the HR function has the
potential to access incredible insights. This is the age of data-driven, AI-
enabled HR – or as I’ll sometimes call it in this book, intelligent HR.

Defining intelligent HR
Data-driven, AI-enabled HR is about harnessing this data explosion and
new analytics tools to drive performance – within both the HR team itself
and the wider organization.

What does this look like in practice?

There are many ways HR teams can make good use of data and AI, but,
broadly speaking, they boil down to three main categories:

●● Unearthing the insights needed to drive better decision making (by which
I mean HR decision making and decision making across the organization).
●● Delivering a better HR service to employees (for example, through
improved recruitment and onboarding processes, or an enhanced
employee wellness offering).
●● And making HR processes more efficient (for example, through
automation).

This idea of data-driven, AI-enabled HR has gathered enormous pace in the

last couple of years as new software tools have emerged to serve the HR
 How data and AI are transforming HR 3

function. There’s a whole market of off-the-shelf HR analytics tools, for

example, that help HR professionals get the most out of data without any
knowledge of data science or AI algorithms. These new technologies are
making HR more data-driven, agile and responsive, while providing em-
ployees with more personalized and engaging experiences.
Ultimately, this business function that has traditionally focused on softer
elements like people, culture, learning and development and employee en-
gagement is becoming increasingly driven by data and analytics. That’s not
to say people aren’t still at the centre of everything HR does – of course they
are. Rather, data and analytics are providing exciting, innovative ways for
HR teams to serve those people.

It’s all about adding value

With data-driven, AI-enabled people management, the top priority is always
to add value to the organization, using all the tools at the HR team’s dis-
posal, including data, sensors, analytics and AI. What you want to avoid is
technology for technology’s sake. Because, yes, you can do some really cool
stuff with data and AI, but if it doesn’t add value to the organization, what’s
the point?
Everything you do with data and AI must therefore help the HR function
and the wider organization deliver its strategic objectives. Whether you
want to address employee turnover issues, build a winning employer brand,
boost performance in certain parts of the business or whatever, data and AI
can help you achieve your goals and solve your biggest challenges. Sounds
grand, doesn’t it? But as the examples in this book show, it is absolutely pos-
sible. HR teams are already gaining tantalizing benefits by using data and AI
in strategic ways to unearth business-critical insights.
Take Google’s often-touted approach to people management as an exam-
ple. Google gives staff free meals, generously paid holiday allowances, ac-
cess to ‘nap pods’ for snoozing during the day and space to grow their own
fruit and vegetables at work. Now, I’m sure Google’s leadership team is full
of lovely, generous people, but that’s not why they implemented these poli-
cies – or, at least, it’s not the only reason. These decisions were based on
what the data told them would increase employee satisfaction. Google’s ap-
proach to boosting staff satisfaction thoroughly disrupted the tech world,
dramatically changing the way Silicon Valley employers think about em-
ployee perks, and now tech companies of all sizes, from the big players to
small start-ups, seek to emulate the Google approach. And, while staff
4 How data and AI are transforming HR

t­urnover is consistently high in the tech world, Google remains one of the
most desirable tech brands to work for.

A (brief) word on automation

There’s been a lot of talk about the rise of automation and the threat this
brings to jobs. From factory line jobs to professions like accounting and
architecture, AI technologies like machine learning – where a computer
‘learns’ from what the data is telling it and adapts its decision making and
actions according to what it’s learned – mean that more and more tasks can
now be automated and completed by machines or algorithms. As we’ll see
throughout this book, a lot of HR tasks can now be automated. In many
cases, machines can perform a task to a much more accurate degree than a
human can. Algorithms can predict employee churn better than a human
ever could, for instance.
AI and automation will therefore have a major impact on HR over the
next five years or so. One Oxford University study looked at the types of
jobs that would be affected by automation, and concluded that, by 2035,
HR administrative jobs had a 90 per cent chance of being automated.2 HR
officers, managers and directors, however, were much less likely to be re-
placed by machines.
What does such automation look like? One good example is virtual help-
desk agents – chatbots, essentially – that could answer simple employee
questions such as ‘When is the company closed over the Christmas break?’
or ‘How much of my annual leave have I used already this year?’ AI technol-
ogy is now so sophisticated that it can respond to natural, spoken language,
rather than typed questions, and even detect the underlying sentiment be-
hind the words themselves. Call centres, for example, are using this technol-
ogy to analyse whether a caller is satisfied, frustrated or angry during the
course of their call.
So, it’s clear that HR will be affected by automation over the next few
years. However, in the context of intelligent HR, this should be seen as a
positive development. Automating the simpler, administrative-type tasks
frees up HR professionals to focus on more important tasks that align with
the company’s strategic vision and help deliver performance improvements.
Of course, there’s also a wider issue around automation – and that is the
impact on jobs elsewhere in the organization. As we’ll see in this book, au-
tomation doesn’t necessarily mean that people will lose their jobs (indeed,
 How data and AI are transforming HR 5

more jobs will be created). But it does mean that people’s jobs are likely to
evolve somewhat, as machines take on more and more tasks. HR will play a
critical role in preparing the organization for this change, for example by
upskilling and reskilling the workforce, and by building a culture of adapt-
ability and flexibility. But we’ll talk more about that later in the book.

How to use this book

My goal in this book is to explore the key ways in which data and AI can
drive performance, both in terms of the HR team’s own performance and
value within the organization, and in terms of how data-driven, AI-enabled
HR can help drive performance right across the business. I think of this
book as a journey, looking at the developments that have brought us to this
point and identifying a path forward for HR professionals.
With this in mind, Part One lines up the building blocks of intelligent HR,
including creating a data strategy, and identifying the different data and
analytics options. We’ll also look at some of the potential pitfalls and con-
cerns around using data and AI, including privacy issues, bias and the need
for transparency. As we’ll see in Chapter 4, the way a company uses data
and AI, and how that is communicated to staff, has a big impact on how
people react. Ill-considered, poorly communicated or discriminatory uses of
data and AI erode trust, and can be extremely harmful to morale. Thankfully,
there are plenty of ways HR teams can mitigate these potential issues and
gain employee buy-in, and we’ll look at some of those in Chapter 4.
Part Two looks at intelligent HR in practice, and how data and AI can
drive operational improvements and better decision making across all the
core HR functions: people analytics, recruitment, onboarding, performance
monitoring and management, training and development, and employee
safety and wellbeing. I cannot stress enough how all of these functions are
already being transformed by data and analytics. In these chapters, I’m not
making wild predictions about potential future developments. The future is
already here. The challenge for HR teams is to ensure they’re continually
adding maximum value to the organization through data and AI – while
maintaining a people-centric approach.
Finally, in Part Three we’ll put what we’ve learned into action, exploring
how to identify (and prioritize) data and AI use cases for your own organi-
zation, how to build the right skills and culture for data-driven, AI-enabled
success and what to consider in terms of technology and infrastructure.
6 How data and AI are transforming HR

And throughout the book you’ll find plenty of real-world examples that
showcase how companies across all sectors are using data and AI in incred-
ible ways. I hope these examples inspire you to tackle this exciting new
world head-on.
At the end of each chapter, I’ll summarize the critical learning points.
Even if you only manage to skim over certain parts of the book, these key
takeaways will give you the absolutely must-have info in one simple list.

Key takeaways

Here’s what we’ve learned in this chapter:

●● Almost everything we do at work can now be measured, from

employees’ day-to-day actions, performance and wellbeing, to wider
business operations. This explosion in data is all part of the fourth
industrial revolution – the ‘intelligence revolution’, which is being driven
by data, AI, automation and connectivity.
●● This explosion in data, coupled with AI-driven advances in analytics,
mean that HR teams can extract insights that improve the performance
of people within the company (including its HR team), and contribute to
the organization’s overall success.
●● With data-driven, AI-enabled HR, the top priority is to add value to the
organization through technology – while, at the same time, maintaining a
people-centric approach.
●● There are many ways HR teams can make good use of data and AI, but,
in their most basic sense, they boil down to three main categories:
unearthing insights that drive better decision making across the
organization; delivering an enhanced HR service to employees; and
making HR processes more efficient.

Now, let’s start our journey into data-driven, AI-enabled HR by exploring

how we got to this point. In Chapter 2 I look at the evolution of intelligent
HR, and how the explosion in data and analytics technologies, including AI
and the Internet of Things (IoT), is making HR more intelligent than ever
before.
 How data and AI are transforming HR 7

Notes
1 J Vincent. Google’s Sergey Brin warns of the threat from AI in today’s ‘technol-
ogy renaissance’, The Verge, 28 April. www.theverge.com/2018/4/28/17295064/
google-ai-threat-sergey-brin-founders-letter-technology-renaissance (archived at
https://perma.cc/3JH5-9XQH)
2 S Shah. Will AI augment or replace HR? HR Magazine, 1 December 2016.
www.hrmagazine.co.uk/article-details/will-ai-augment-or-replace-hr (archived at
https://perma.cc/QS42-V76V)
8

THIS PAGE IS INTENTIONALLY LEFT BLANK

 9

PART ONE
Laying the
groundwork
for data-driven,
AI-enabled HR
Organizations have seen enormous transformation in recent years, driven
by new technologies and automation. And the HR function is not immune
from this change. Far from it – the way in which organizations attract, re-
tain and nurture talent is becoming increasingly shaped by new data and
artificial intelligence tools.
In this first part of the book we’ll explore how data and AI are changing
the HR function, and how this transformation came about. We’ll also ex-
plore the foundations of ‘intelligent HR’, including the types of data avail-
able today, the different analytics options, the challenges that surround data
and AI, and how to use data and AI strategically in your organization.
10

THIS PAGE IS INTENTIONALLY LEFT BLANK

 11

How data and 02

AI have come to
revolutionize HR
Within the wider context of work, human resources is a relatively new func-
tion. Yet, in its short – roughly century-long – history HR has evolved sig-
nificantly. And that change has been particularly pronounced over the last
20 years, driven by an explosion in data and the evolution of AI technolo-
gies. Today, technology plays a vital role in the work of HR professionals.
But how did we get to this point?
In this chapter, we take a quick tour through the history of HR and ex-
plore how data and AI (as well as other technologies) have found their way
into the HR function.

The emergence of HR and its early evolution

We can trace the history of HR back to the late 19th century, when the con-
cept of ‘personnel management’ first emerged.

The Industrial Revolution took employment

from ‘personal’ to ‘personnel’
Coming in the wake of the Industrial Revolution, the late 19th century was
a time of huge technological and societal change. Processes that had tradi-
tionally been done manually were now mechanized, thereby changing the
work that humans did. In general, work shifted from agricultural labour
and small-scale manufacturing towards work in large factories. And this, in
turn, changed the nature of employment. Someone who had previously
worked with their family, or for a local, small-scale business might now be
one of potentially hundreds of employees in a large company – with no
12 Laying the groundwork for data-driven, AI-enabled HR

­ irect relationship with the owner of that company. In other words, the
d
closely connected relationship between employer and employee was, by and
large, a thing of the past. And this prompted a need for business owners to
hire people to manage the employee relationship – hence, personnel man-
agement. In these early days of personnel management, the focus was largely
on administrative tasks such as hiring and payroll, as opposed to looking
after people. Working conditions were often not great (to put it mildly), and
employee concerns often went unheard – prompting a period of unrest and
clashes between employees and employers. In America, this was known as
the ‘Progressive Era’, a period of social activism and reform aimed at defeat-
ing (among other things), corruption, poverty and the exploitation of ­labour.

Moving into the 20th century

Amidst this societal change, in the early 20th century the focus of HR shifted
more to managing employee relations and grievances, as well as making
sure the company complied with labour laws. It’s fair to say the work of HR
professionals during this time was largely reactive, rather than proactive –
dealing with employee concerns, for example, or negotiating with employ-
ees over working conditions.
This began to change in the 1960s and 1970s, as HR started to focus
more on strategic planning and organizational development, with HR pro-
fessionals working to align HR practices with the organization’s wider
goals.
But even during this period, as HR began to noticeably evolve, it was still
highly reactive. Within organizations, HR was often seen as the ‘corporate
police’ – making sure employers and employees complied with new legisla-
tion on discrimination and harassment, and dealing with the fallout when
rules were broken. Ask the average manager in the 1970s what HR did and
their answer would likely be along those lines: policing people’s words and
actions. You know, compliance stuff.
By the 1980s and 1990s the ‘personnel’ departments of old had evolved
into the ‘human resources’ function that we recognize today – one that em-
phasizes the importance of employee development, performance manage-
ment and talent management. This is really when HR professionals began to
work more closely with managers and executives to ensure that the organi-
zation had the right people in the right positions to achieve its goals.
And this evolution was matched by a technological revolution. To some
extent, HR has always been rich in data – employee records and so on – but
there was no widespread effort to analyse and harness that data. But in the
 How data and AI have come to revolutionize HR 13

1990s, HR departments began to adopt HR information systems (HRIS),

which allowed for the automation of routine HR tasks such as employee
record keeping and benefits administration. These systems also provided
HR professionals with access to data that could be used to monitor and
analyse HR trends and performance.
It was this digital revolution of the late 20th century that paved the way
for a new era of data-driven HR.

The turn of the 21st century and the beginnings

of data-driven HR
The early 2000s saw the emergence of talent management systems, which
allowed organizations to better track and manage employee performance
and development. These systems provided HR professionals with more so-
phisticated data analysis tools, allowing them to make more informed deci-
sions about talent management. This empowered HR professionals to work
in a much more strategic and forward-thinking capacity.
Of course, hiring, compliance and dealing with employee grievances re-
main a core part of the HR function, but from this point onwards HR is
recognized as having a critical impact on an organization’s success. People
are a huge part of what distinguishes a business from its competitors. Talent
is a key differentiator. And the HR function over the last 20–30 years has
evolved to help organizations get the most out of their talent.

HR in the 21st century: The rise of people

analytics
Building on the emerging data-driven approach of the early 2000s, in recent
years there has been an even greater shift towards people analytics, or using
data to gain insights into the workforce and to inform HR decision making.
This includes analysing data on employee engagement, turnover, productiv-
ity and diversity and inclusion – and then using insights from that data to
inform decisions and action.

The role of AI in HR
Increasingly, this data-driven approach is becoming more proactive and pre-
dictive. Rather than simply looking at what has been, with AI tools HR
professionals can predict future trends and identify potential areas of risk.
14 Laying the groundwork for data-driven, AI-enabled HR

AI seems like a very recent addition to the HR function, but in fact the
history of AI in HR can be traced back to the early 2000s. At that time, the
focus was on automating routine HR tasks such as resume screening and
candidate sourcing using rule-based systems. It wasn’t until the mid-2010s
that AI began to be used for more sophisticated HR tasks such as predicting
employee turnover, identifying high-potential employees and identifying the
best candidates for specific roles. In one example, IBM used AI to predict –
with 95 per cent accuracy – which employees were most at risk of leaving
their job.1 The predictive attrition program saved the company almost
$300 million.
From the mid-2010s onwards, AI was also being used to automate cer-
tain HR functions. For example, AI-powered chatbots were introduced to
improve the candidate and employee experience by providing immediate
responses to frequently asked questions.

The explosion in data

Without data, these advancements in AI would not be possible. Data is what
allows an AI system to identify patterns, spot trends and make predictions
about the future. Lucky, then, that people generate so much data! Almost
everything in a business context generates data, from an employee sending
an email to the sensors on production line machinery.
What’s more, employers can now collect types of data that simply weren’t
available before. This includes things like capturing employees on closed-
circuit television (CCTV), scanning social media data for employee senti-
ment, analysing the content of emails and even monitoring where employees
are using the data from sensors in corporate phones. I’m not saying it’s
necessary for employers to gather data on every little thing – indeed, data
for data’s sake is something we want to avoid – just that it is now possible
to gather more data, and more varied types of data. We’re no longer talking
only about neatly structured data in databases and spreadsheets, but also
unstructured data such as text and video content. There’s also a wealth of
HR-specific data, such as recruitment data, career progression data, training
data, absenteeism figures, productivity data, personal development reviews,
competency profiles and staff satisfaction data.
Bottom line, businesses have more data to mine for insights than ever
before. In the past, people-related data mostly went unused or, if it was used,
it was put into charts and tables for something like a corporate performance
pack. It’s the advancement of AI that has allowed HR teams to truly harness
this data and extract value from it.
 How data and AI have come to revolutionize HR 15

Amidst this explosion of data and AI, HR has become the function that
supports the strategic use of people-related data – gathering that data, ana-
lysing it and sharing the insights with decision makers across the organiza-
tion. I guess you could say the name ‘people analytics’ is slightly misleading,
since it is much more about improving strategic decision making and perfor-
mance in order to drive the business forward – but since that data is gener-
ated by the people working within the business, ‘people analytics’ it is! By
the mid-2010s professionals were being recruited into specific people ana-
lytics roles within HR functions. And in 2020, LinkedIn’s global talent
trends 2020 report listed people analytics as one of the most important
trends in human resources.2 People analytics had officially arrived.

The changing technology landscape –

and its impact on HR
It’s clear that data and AI have advanced enormously in the last decade. In
the last few years, technology has advanced at a pace that even I – an expert
in future tech trends – find eye-opening. So let’s spend a brief while explor-
ing the wider technology context and the major tech trends that will increas-
ingly impact the HR function.

Future internet technologies like the metaverse

and blockchain
We’re entering the third evolution of the internet, often referred to as Web3
or Web 3.0. Blockchain is one of the core features of this future internet,
since blockchain is the technology that underpins things like cryptocurrency
and non-fungible tokens (NFTs).
In its most basic sense, blockchain is a way of storing data. Like a database,
but built for the future internet and with much better security than your aver-
age computer system. Just one of the ways blockchain can improve HR pro-
cesses is in validating resumes, in a system where people’s credentials live ‘on the
blockchain’, and can be validated by previous employers, universities and even
government departments (validating the candidate’s identity, for instance).
So, blockchain is one future internet technology to watch. Another fea-
ture of the future internet – and potentially one that has wider-reaching
implications for HR – is that the online world will become much more im-
mersive. And this brings us to the metaverse…
16 Laying the groundwork for data-driven, AI-enabled HR

The emerging concept of the metaverse is geared around persistent,

shared virtual worlds – immersive digital spaces where we can play, social-
ize, shop, learn, exercise and even work. Mark Zuckerberg once described
the metaverse as being inside the internet, rather than just looking at it
through a screen, and I think that’s a great description. Largely this will be
enabled by immersive metaverse technologies like virtual reality (VR).
The Fortnite universe gives us a glimpse of how the metaverse will impact
life. In Fortnite, users aren’t just there to game; they buy outfits and acces-
sories for their digital selves, hang out with friends and even attend virtual
gigs (by superstars like Ariana Grande). So, the metaverse enables people to
spend more time in virtual environments, and those environments will be
more immersive, realistic and engaging than today’s online experiences.
Crucially, in this new evolution of the internet, the lines between the real
world and the virtual world become even more blurred than they are today.
And this will influence the world of work. Instead of logging on to a Zoom
meeting, for instance, remote employees will be able to put on their VR
headset and join a virtual meeting room, with realistic avatars of their col-
leagues. Employees will be able to work and train in the metaverse, in vir-
tual co-working spaces or digital replicas of real-world offices. And there
will be new ways for colleagues to collaborate on projects virtually in real
time. Consider, for example, that Meta and Microsoft are teaming up to put
Microsoft tools, including Teams and Windows, on the new Oculus VR
headset.3
The metaverse will even change the way we hire and onboard employees.
For example, Samsung hosted a virtual recruitment fair to attract candi-
dates,4 and PwC has created its own metaverse space, Virtual Park, to give
candidates an insight into what it’s like to work for the company.5
We’ll touch on the metaverse and blockchain throughout the book, but if
you want a really deep dive on the topic, you might be interested in my
book, The Future Internet: How the metaverse, Web 3.0, and blockchain
will transform business and society.

The importance of AI
The metaverse and blockchain are both enabled by AI, which again puts AI
(and, in turn, data) at the centre of the digital revolution. Note that I typi-
cally use ‘AI’ as a catch-all description, but what we’re often talking about
is machine learning – the specific subset of artificial intelligence focused on
creating machines (computers, software, etc.) that are capable of learning
from data in a way that’s similar to the human brain.
 How data and AI have come to revolutionize HR 17

AI capabilities are advancing fast, and AI is increasingly able to take on

the work that was traditionally thought of as human. For example, we have
generative language AI models that can write text that’s pretty much indis-
tinguishable from that of a human writer. The GPT tool by OpenAI can
create anything that has a language structure – which means it can answer
questions, write essays, summarize long texts, translate languages, take
memos and even create computer code. The technology is advancing to such
an extent that it can create computer code based on natural language
prompts. In other words, GPT can build apps and software for users who
don’t have a background in software design. Even just a few years ago this
wasn’t possible.
AI is evolving at an incredible rate, then. And, as such, it is inextricably
linked with other technology trends. Trends like the Internet of Things, the
ever-increasing network of smart devices that are connected to the internet –
from watches and phones to fridges and coffee machines. Such devices are
constantly gathering and transmitting data, further fuelling the growth in
data and AI. And trends like chatbots and voice interfaces (think Alexa and
Siri), which are able to communicate so well because of AI. And trends like
robots and cobots (collaborative robots), which are becoming more intelli-
gent, learning to perform tasks without human intervention or work seam-
lessly alongside humans, thanks to AI.
Naturally, such widespread technological change is impacting how peo-
ple do their jobs and how HR serves the organization. We’ll talk about that
throughout the book. For now, let’s also acknowledge that the very nature
of work itself – and our attitude to work – is changing.

The changing nature of work

As the HR function has become more data-driven, it has also had to grapple
with some widespread changes in the world of work.

The employee experience

These changes were drastically accelerated by the Covid-19 pandemic,
which fuelled the adoption of technology and remote working in many sec-
tors, and caused many people to re-evaluate what they want from work. On
top of that, more young people are entering the workforce. And these
younger generations expect, more so than older generations, to connect with
their employer in a meaningful way.
18 Laying the groundwork for data-driven, AI-enabled HR

So, while the human resources functions of the past might have focused
more on grievances and compliance, they are now very much focused on
creating an employee experience that drives business performance and
boosts the acquisition and retention of talent. Things like an awesome on-
boarding experience, flexible working arrangements, generous benefits (be-
yond salary), a strong organizational purpose, and so on.

Changing organizational structures

The way organizations are structured is changing too. The traditional, rigid
and hierarchical organizational structures of old are giving way to flatter,
more agile structures that allow the business to quickly reorganize teams
and respond to change. This move towards flatter organizational structures
is also, in part, a response to the proliferation of freelance workers.
More and more people are engaged in gig working, whether it’s on a full-
time basis or as a side hustle. The rise of gig working, coupled with a wider
move towards remote working, has massively opened up the labour market.
An organization’s next hire could be based on the other side of the world,
for instance. They could be a freelancer who has to integrate and work with
an existing team. This creates both opportunities and challenges for compa-
nies. In a global labour market where candidates are no longer limited to
local employers, organizations must prioritize the employee experience and
demonstrate a strong purpose if they want to attract and retain the best tal-
ent. As we’ll see throughout this book, data and AI can add huge value to
this process of creating an enriching employee experience – from recruit-
ment and onboarding to employee wellbeing.

The impact of automation

There is also rising automation to consider, particularly when it comes to
striking the right balance between human workers and intelligent robots.
We now have increasingly capable robots and artificial intelligence systems –
chatbots, for example – that can take on tasks previously done by humans.
This leaves the HR function with some key questions, such as how do we
find the balance between intelligent machines and human intelligence? What
roles should be given over to machines? And which roles are best suited to
humans? There’s no doubt that automation will affect every industry, and
much of the responsibility for preparing organizations – and people – for
this change will fall to HR. If you’re interested in reading more about these
 How data and AI have come to revolutionize HR 19

(and other) business trends, you might like to check out my book, Business
Trends in Practice: The 25+ trends that are redefining organizations.

New skills for a new era of work

The changing nature of work and rapid acceleration of technology will im-
pact the kinds of skills that employers need. The majority of children enter-
ing school today will do jobs that don’t yet exist. (Sound far-fetched?
Consider that we didn’t have social media managers, podcast producers,
influencers and blockchain developers a generation ago.) And for many peo-
ple already in the workforce, their jobs will evolve or even become obsolete.
In a world in which more and more jobs (and parts of jobs) are given over
to intelligent machines, the essential skills that organizations require will
evolve. Naturally, organizations will need more digital skills, but distinctly
human skills will also become more valuable – the kinds of things that ma-
chines can’t do as well as us. Skills like complex decision making, emotional
intelligence and empathy, creativity, teamwork and adaptability. We’ll talk
more about essential skills later in the book but – as you might have
guessed – I also have a book on the subject. It’s called Future Skills: The 20
skills and competencies everyone needs to succeed in a digital world.

So what lies ahead for data-driven HR?

Overall, the use of data in HR has evolved from simple record keeping to
sophisticated people analytics that inform key HR decisions. As technology
continues to advance, it is likely that data and AI will play an even more
important role in HR in the years to come.

Challenges and opportunities

The increasing use of data and AI brings challenges. I’ll talk more about this
in Chapter 5, but one of the biggest challenges will be ensuring that AI is
used ethically and fairly. There’s a risk that AI-powered systems could rein-
force biases or discriminate against certain groups (there are many examples
of this already happening, often because the underlying data itself is skewed
for or against certain groups). To address this, HR professionals must be
trained to use AI responsibly and ensure that their algorithms are transpar-
ent and unbiased. And that the underlying data isn’t biased.
20 Laying the groundwork for data-driven, AI-enabled HR

There will also need to be transparency around exactly what data or-
ganizations gather from their employees, and why. Consent is key. But going
above and beyond consent, we also want employees to fully buy into the use
of data and analytics because they understand how it will benefit their job
and the organization as a whole. Careful communication is needed.
Despite the challenges that lie ahead, the use of AI in HR is expected to
continue to grow in the coming years. This is a good thing. With the increas-
ing availability of data and advances in machine learning, AI has the poten-
tial to revolutionize HR by providing insights into employee behaviour,
predicting workforce trends and improving the overall employee ­experience.

Data, AI and major HR trends

Before we close out this chapter, let’s briefly explore some significant future
trends in HR. Because it’s clear that data and AI have an important role to
play in these trends…
One HR trend that we’ve already touched on is the hybrid working envi-
ronment that employees will increasingly enjoy. The role of technology in
general is pretty obvious here, since it will allow remote employees to col-
laborate with colleagues in new ways. But while flexible working brings
many benefits to employees, it can sometimes have a negative impact on
engagement, connection and satisfaction. Data and AI can help HR func-
tions monitor these factors and ensure that employees remain engaged even
as the physical working environment changes.
Employee wellness is a perennial trend in HR, but more attention is being
paid to mental health. Among other things, data and AI can help organiza-
tions monitor how employees feel about their mental health, assess the ef-
fectiveness of employee assistance programmes, and design an employee
experience that supports good mental health.
Another trend is addressing the talent exodus. In the wake of the pan-
demic, many people left their jobs, or searched for jobs that offered them a
better work–life balance (or, simply, more meaning). In certain sectors –
technology and IT being a great example – there are more job openings than
there are candidates, so holding on to talent is a primary objective. Data and
AI can be used to improve the hiring and onboarding process, and identify
the benefits that resonate most with people. And that’s just for starters. As
we’ve already seen in this chapter, there are predictive AI systems that can
identify the employees most at risk of quitting – which is incredibly valuable
information to a proactive HR team.
 How data and AI have come to revolutionize HR 21

AI analytics can also add real value to diversity, equity and inclusion
(DEI) programmes – not just when it comes to assessing DEI in the work-
place, but also finding a way forward. For example, analytics can help busi-
ness leaders understand employees’ thoughts on belonging, intersectionality
and bias in the organization.
And since upskilling the workforce will be a major task in this age of
rapid digital transformation, it makes sense that learning management is
another major trend in HR. AI-enhanced learning systems allow employees
to learn online at their own pace and enjoy a more personalized learning
experience.
These trends give us just a small taste of what AI and data can bring to
HR practices in future. But don’t worry, we’ll talk more about specific HR
practices in Part Two of the book.

Key takeaways

Let’s finish with a quick summary of what we’ve learned in this chapter:

●● In its relatively short history, HR has undergone a major evolution – from

the admin-focused approach of old, to a more strategic approach where
the work of HR aligns with organizational goals.
●● Particularly in the last 10 years or so, HR has adopted more of a people
analytics approach, using data and AI to inform decision making and
drive business performance.
●● The last few years have seen major digital transformation – not just in
terms of data and AI, but also in terms of future internet technologies like
blockchain and VR. In the future internet, as the metaverse evolves, we
will spend more time in virtual environments, and this will impact the
world of work.
●● At the same time, there are other major shifts taking place in the world of
work. From hybrid working and changing organizational structures to the
need for a stellar employee experience, HR functions are grappling with
widespread change.
●● Data and AI will play an increasingly important role in HR, helping HR
professionals navigate the challenges that come with rapid
transformation, and leverage the many opportunities that will arise from
new ways of working.
22 Laying the groundwork for data-driven, AI-enabled HR

No doubt about it: this is an exciting – and challenging – time to be an HR

professional. I hope this chapter has given you a sense of how we’ve arrived
at this point in time, where people analytics plays a central role in HR. Now
let’s explore these AI tools in a little more detail. Turn the page and discover
some of the revolutionary data, analytics and AI tools available to HR.

Notes
1 E Rosenbaum. IBM artificial intelligence can predict with 95% accuracy which
workers are about to quit their jobs, CNBC, 3 April 2019. www.cnbc.com/2019/
04/03/ibm-ai-can-predict-with-95-percent-accuracy-which-mployees-will-quit.
html (archived at https://perma.cc/DSK9-VVFC)
2 LinkedIn. Data-driven insights into the changing world of work, LinkedIn, May
2023. business.linkedin.com/talent-solutions/global-talent-trends (archived at
https://perma.cc/FBF4-372U)
3 A Kidwai. Meta and Microsoft form an unlikely partnership on workplace VR,
HR Brew; 19 October 2022. www.hr-brew.com/stories/2022/10/19/meta-and-
microsoft-form-an-unlikely-partnership-on-workplace-vr (archived at https://
perma.cc/AQ3H-ABCA)
4 L Handley. Looking for a job? You might get hired via the metaverse, experts
say, CNBC, 30 November 2021. www.cnbc.com/2021/11/30/looking-for-a-job-
you-might-get-hired-via-the-metaverse-experts-say.html (archived at https://
perma.cc/8TXT-2JCE)
5 PwC. PwC’s Virtual Park, PwC, 2023. www.pwc.co.uk/careers/early-careers/
ourevents/virtual-park.html (archived at https://perma.cc/J9UE-P525)
 23

The data, 03
analytics and AI
tools available
to HR
These days, most HR functions are rich in data. But that’s not the same as
being rich in insights. For that, you need to be able to turn data into valuable
insights that answer your biggest people-related questions, and help you
solve problems, drive performance and boost the employee experience. This
is where analytics comes into play.
Analytics is the process of collecting, processing and reporting data to
generate insights. In most cases, we use analytics tools and algorithms to do
this. By analysing data with algorithms and analytics tools, you can extract
critical insights that will help the HR function perform its role to the fullest.

The amazing capabilities of analytics

and AI tools (and how to use them in HR)
Traditionally, data would have been queried using tools like Excel and struc-
tured query language (SQL – used to interrogate databases). These methods
of analysis have served businesses fine for decades and still form the founda-
tion of many essential business analytics, like analysing revenue. When
you’re dealing with neat, structured data in spreadsheets and databases,
these methods work great.
However, the amount of data that businesses have access to these days
increasingly doesn’t conform to this traditional notion of data – i.e. that it’s
numerical and lives in a database or spreadsheet. Much of the data busi-
nesses have access to today is in the form of text, video, voice recordings and
24 Laying the groundwork for data-driven, AI-enabled HR

so on. These sorts of data are more difficult to work with, and as such new
analytical methods have arisen to cope with this. Bottom line, we can now
use AI-based analytics to do so much more than traditional analytics tools.

A quick overview of AI

AI is a term I use throughout this book as a useful catch-all description. But

the term ‘AI’ loosely groups together specific technologies and methods –
the most common being machine learning, deep learning and generative
adversarial networks.
AI is best thought of as an aspiration – we aspire to build intelligent
machines that are capable of learning for themselves, just as naturally
intelligent creatures such as humans and animals can. This ultimate
aspiration of generalized AI – AI that is capable of adapting to pretty much
any task that a human can do – is way beyond our current capabilities.
Instead, today’s applications of AI – certainly in a business context – fall
under the category of specialized AI, which basically means AI tools that
are designed to become very good at one task (or group of tasks),
constantly learning to improve as they go. And this brings us to machine
learning…

Machine learning – the current cutting edge in AI

Machine learning uses techniques that have been around since at least the
1960s to build algorithms that are capable of getting more and more
accurate, as they are given more information. In other words, they are
capable of learning from data. And the more data you give a machine
learning system, the more accurate it becomes. There’s been a huge
increase in machine learning over the last 10 or so years, largely because
there has also been a huge increase in data availability.
But how does an algorithm ‘learn?’ Well, just like humans, they do it
through training and then putting what they’ve learned to use, assessing
the results they achieve and modifying their behaviour accordingly until
they work out how to get better results.
Take image recognition, for example. If we have an image and want to
know whether it represents a cat, in a very simplified machine learning
model we could ask whether it has a tail, whether it has whiskers, whether
it has four legs, whether it is covered in fur… and so on. Finally, informed by
 The data, analytics and AI tools available to HR 25

the decisions it has made already, it attempts to tackle the question ‘Is this
a picture of a cat?’ If the algorithm decides that the answer to all of the
initial questions is yes, then it might report to us that there’s a high
probability that the image contains a cat. If that’s all it did, then in our very
simple use case it might be quite effective, but we still couldn’t say it was
‘learning’. If we were to show it the same picture over and over, we would
keep getting the same answer, regardless of whether it was right or wrong.
So, machine learning algorithms include a feedback loop, and a system
of ‘weighting’. Say it looks at a picture of a human, but they are on all fours,
and therefore mistakenly answers ‘yes’ to the question ‘Does it have four
legs?’ It goes on to mistakenly identify the image as a cat. Realizing, thanks
to the feedback loop, that it has got this one wrong, it can examine each
step of the process and work out where it was most likely to have made a
mistake. Through this process – repeated thousands or millions of times – it
is able to ‘learn’ which questions are the best indicators of the correct
answer – and adjust its answers (or predictions) in order to take this into
account. This explains why we can consider the algorithm to be ‘learning’.

Deep learning and neural networks

Think of deep learning as another, more advanced, form of machine

learning. As with other forms of machine learning, the principle involves
mimicking elements of the human cognitive process. Specifically, deep
learning involves the construction of artificial neural networks (ANNs) –
decision-making frameworks similar to the natural neural networks in the
human brain. Due to their size, which is often vast, the ANNs used in deep
learning are often called deep neural networks.
Deep learning is best used when the problems you are trying to solve are
very complex and involve a large amount of unstructured data – for
example, understanding and extracting meaning from human speech or
images.

Generative adversarial networks

Generative adversarial networks (GANs) are an even more recent

development in the world of machine learning, which essentially work by
pitting two neural networks against each other, forcing them to learn to
work with increasing efficiency in order to outsmart the other one.
Essentially, one network, known as the generative network, is given the
26 Laying the groundwork for data-driven, AI-enabled HR

task of creating a set of data that mimics the rules of a predetermined

training set. A piece of data is then taken from either the training set or the
newly created set, and presented to the other network, which is known as
the discriminator. That network then has to determine whether the piece of
data comes from the training set, or the newly created set. If it gets it right,
it wins the round. If it gets it wrong, then that round goes to the generative
network. Both networks ‘learn’ from the process, and go on to become even
more ‘creative’ in their attempts to thwart the other. It’s a machine-off,
basically.
What are GANs used for? They’re widely used in the creation of
‘deepfakes’ – fake (but highly realistic) photographs and videos. They’re
also used to create AI art, music and writing. In another example, a GAN
system created by tech company NVIDIA was able to recreate the Pac-
Man game from scratch. After being fed visual data of the game being
played, the GAN system managed to generate the entire code for a game of
Pac-Man that can be played by humans!1
Now that we have a basic grounding in AI, let’s delve into the specific
types of analytics.

The types of analytics available to HR

Before we get into the analytics tools themselves, let me start with a quick
word of warning: it’s very easy to get caught up in all of the exciting oppor-
tunities that analytics offers. This is truer now than ever before, as thou-
sands of vendors are taking advantage of the vast appetite from businesses
for anything labelled ‘AI’. You’ll no doubt see lots of organizations doing
very cool things with analytics, but what works for one business may not
work for yours. The challenge in using data and AI to great effect is deciding
which approach will work best for you, without losing sight of your goals.
Having said that, things are moving so quickly with AI that it’s safe to
assume that new and improved ways of doing just about anything will
emerge even as you’re working on your initial use cases. So, while you don’t
want to get distracted by the ‘new and shiny’ toys on offer, you do want to
keep an open mind about new ways of doing things.
With that out of the way, let’s get to the analytics themselves. Because
advanced analytics, including all of the machine learning techniques de-
scribed above, can be used to extract insights in some very exciting ways.
Here, we’ll take a look at the most applicable methods for HR.
 The data, analytics and AI tools available to HR 27

Image and video analytics

These analytic processes are used to extract insight and meaning from any
kind of image – still or moving – including photographs and video footage.
Machine learning has proven itself to be awesome at this process – provid-
ing it’s given enough training data and powerful enough computers to
crunch through it. Image and video analytics might also include analysing
metadata, such as Global Positioning System (GPS) tags on photographs or
timestamps on CCTV footage. These give extra insights that wouldn’t be
available simply from looking at the data by itself.
In the case of video analytics, this gives us the added capability of meas-
uring and analysing behaviour over time. Which is why Tesla’s Elon Musk
has said that he believes self-driving cars will eventually operate on video
data alone,2 rather than using other sensing technologies like LIDAR. This
means they will operate in much the same way as a human driver does,
mainly relying on what they see in order to navigate.
In a business context, image analytics has a large number of potential
uses, including facial recognition for security purposes, and recognizing
products in pictures shared on social media. Video analytics are proving use-
ful in a safety context – for example, to detect when employees aren’t wear-
ing the appropriate safety wear or following safety protocols – or simply to
understand how employees behave when they’re on site.

Text analytics
This is the process of extracting meaning from large quantities of text data.
Most businesses have huge amounts of text data – emails, company docu-
ments, customer records, websites, press coverage, blogs, social media posts,
etc. Thanks to text analytics, this data can now be mined for insights.
The most common text analytics methods include:

●● Categorization – which involves applying structure to text so that it can

be classified by factors such as subject, content, relevance, whether it is
fiction or non-fiction, and so on.
●● Text clustering – which involves grouping text into topics or categories
to make filtering easier, in the same way that search engines do.
●● Summarization – which involves pulling key or relevant points from the
document and automatically creating summaries, perhaps personalized
for certain people.
●● Sentiment assessment or analytics – which involves extracting opinion
or sentiment from text and categorizing it. (More on this coming up.)
28 Laying the groundwork for data-driven, AI-enabled HR

Text analytics is especially helpful for understanding more about your em-
ployees. For example, text can be mined for patterns such as an increase or
decrease in positive feedback from customers, and this may help to identify
customer service representatives who are performing well and those who
may need extra support to improve in their role. I know one organization
that uses text analytics tools to scan and analyse the content of emails sent
by their staff as well as their social media posts. This allows them to accu-
rately understand the levels of staff engagement, meaning they no longer
need to carry out traditional (and expensive) staff surveys, and they no
longer have to wait to assess staff engagement on an annual basis.

Sentiment analysis
Sentiment isn’t only found in text – it can be extracted from video and audio
data too, or any other form of data that can be used by someone to express
an opinion! Here, the aim is to extract subjective opinions and feelings, in
order to determine the attitude of an individual or group towards a particu-
lar topic, concept or idea. Sentiment analytics is regularly carried out across
social media to understand our reaction to marketing campaigns or new
products and services. It’s also used by governments and opposition politi-
cians to assess the popularity of policies.
As mentioned, sentiment analysis is very useful when you want to be able
to understand stakeholder opinion, be it from customers or employees.
Generally, sentiment analysis looks at feelings and opinions as expressed by
groups of people, rather than individuals, so it doesn’t have to rely on the
ability to collect and store personal data. That said, it can also be used to
identify individual voices – for example, employees who are amazing advo-
cates for your brand online.
Sentiment analysis seeks to understand the meaning in data beyond what
is directly expressed – so by analysing choice of words, context and metadata
it’s possible to understand, for example, what emotional state the creator
was in, even if they don’t specifically say ‘I’m feeling happy’ or ‘I’m feeling
frustrated.’ It can include monitoring and analysing body language in image
and video data, for example, or stress levels in voice data (great for analysing
customer service calls).
Sentiment analysis is also very useful when you want to understand the
attitude of an individual or group regarding a particular topic (such as pro-
posed changes to the company’s incentive scheme, for instance) or overall
context (wider company culture, for example), and whether that attitude is
positive, negative or neutral. In this way, sentiment analysis helps us get at
 The data, analytics and AI tools available to HR 29

the real truth behind communication. And, in this age of constant digital
connectivity and our increasing desire to share our thoughts and feelings
about all sorts of things – including companies – online, sentiment analysis
has become mainstream.

Voice and speech analytics

This is the process of extracting information from audio recordings of con-
versations or message logs. This can include analysing the topics and actual
words and phrases that are used, as well as the emotional content (senti-
ment) of the conversation.
Voice and speech analytics is commonly used in customer service settings.
For example, it can be used to identify recurring issues in customer com-
plaints or persistent technical problems that occur frequently. Obviously,
you can also use this type of analytics to assess the quality of customer ser-
vice that your representatives are providing, identifying star players whose
performance can then be highlighted to others as an example, or poor per-
formers who may benefit from additional training.
You could also use voice analytics to help identify recurring themes
around employee satisfaction, or, in the case of a call centre, employee per-
formance. Voice analytics can also help you identify when your employees
are getting frustrated or angry. By analysing the pitch and intonations of
conversations taking place in your call centre, you can gauge the emotional
state and performance of workers.

Analytics as an alternative to some traditional

HR tools
I’m not a big fan of those big annual employee surveys. They’re expensive.
Employees don’t find them particularly engaging. And the big danger is peo-
ple might just tell you what they think you want to hear.
The wide range of analytics tools on offer now can be used to give better
and more accurate insights for HR teams. You can, for example, collect data
from internal comms systems and social media and then analyse that data
using text analytics and sentiment analysis to understand employee senti-
ment. You can also gain a much deeper understanding of your corporate
culture.
Say, for example, you believe that your corporate culture is efficient but
fun. You may think that your business operates like a family, with a strong
30 Laying the groundwork for data-driven, AI-enabled HR

focus on excellent customer service, and those are values that have been
driven home to employees. Your orientation for new recruits draws their
attention to those values and the corporate culture that you believe exists.
But what happens after six months – are those employees embodying those
values or is something else calling the shots?
You could ask your new employee or you could conduct annual surveys,
but the quickest way to get to the truth is to assess what your employees are
saying and doing as part of their everyday work. That’s your culture, after
all. If you believe you’re driven by high-quality customer service and yet no
one answers the phone after 4.45pm, chances are that culture is not quite as
real as you’d like it to be.
New analytics approaches can even replace traditional methods of assess-
ing performance. Most companies assess employee performance annually,
but, in this world of data, once a year just isn’t enough. In order to be effec-
tive, performance should be assessed on a regular and less formal basis, and
modern data collection and analytics allows this.
Today, we have many innovative ways of collecting and analysing perfor-
mance. In one innovative example, wearable technology company Humanyze
(formerly Sociometric Solutions) has created electronic employee badges
that capture information from conversations as they go about their day, in-
cluding the length of the conversation, the tone of voice involved, how often
people interrupt, how well they show empathy and so on. Data from the
badges can be analysed in a number of ways, including text analytics, senti-
ment analysis and voice analytics, and then used to predict which teams are
likely to be more successful, which employees are more productive and cre-
ative, and which show signs of being great leaders.3 One of Humanyze’s
clients, a major bank, noticed that its top-performing employees at call cen-
tres were those who took breaks together. Based on this knowledge, they
instituted group break policies and performance improved 23 per cent.
(Read more about monitoring and managing performance in Chapter 9.)

Tapping into no-code and AI-as-a-service tools

If you’ve read this far and are interested in putting some of these ideas into
practice, but haven’t a clue how to build your own machine learning and
deep learning algorithms, you’re in luck. Because these days, businesses big
and small can access AI through a vast number of tools and applications
that are available as cloud services. These are often known as no-code solu-
tions or AI-as-a-service. In fact, AI may already be available in some of your
existing cloud-based HR apps. If your people-related data exists in the
 The data, analytics and AI tools available to HR 31

cloud, you can often just ‘plug’ it into existing tools and switch on AI
­capabilities.
What’s more, since the GPT tool that we mentioned in Chapter 2 can cre-
ate computer code based on natural human language instructions, several
companies are now creating solutions that harness this potential. So it’s
likely that, in the near future, it’ll be easy for people with zero experience of
computer programming to create their own bespoke software applications.
No doubt about it, we’re seeing the ‘democratization of AI’ happening
­before our eyes.
Many of the off-the-shelf analytics tools will also include a data visuali-
zation component that helps you communicate the insights you’ve u ­ ncovered.
This is important because data-driven HR is about turning people-relevant
data into insights and actions that add value to the business. In order to do
this successfully, you need to ensure it’s easy for the various decision makers,
whomever they may be, to extract insights from the data. The easier it is to
understand the data and pull out key insights, the easier it is for people both
within the HR team and beyond to make decisions and act on that data.
This is why data communication and data visualization have become such
big topics in recent years.
As mentioned, many commercial analytics platforms will come with their
own built-in visualization tools. But there are also inexpensive data visualiza-
tion tools like Tableau and Qlik available as a service. You can make data look
very attractive with impressive visuals, but I think that using a combination of
visuals and (short) narrative is much more powerful than using just visuals. For
instance, a graph may be a good way of showing employee churn trends over
time, but a simple narrative alongside it can pull out the key messages and put
that information into context – explaining what might be behind those trends
and why there was a spike in churn in late 2022, for example.

ChatGPT as an example of no-code AI in action

I’ve mentioned the GPT deep learning language model a couple of times al-
ready. Now it’s time to meet ChatGPT, an AI-based chatbot system that uses
GPT’s natural language processing (NLP) technology to generate conversa-
tions. Launched in late 2022, ChatGPT enables users to ask questions or tell
a story, and the bot will respond with relevant, natural-sounding answers
and topics. The interface is designed to simulate a human conversation. As
such, ChatGPT is capable of explaining, remembering what was said earlier
in the conversation, elaborating on ideas when asked, and even apologizing
when it gets things wrong.
32 Laying the groundwork for data-driven, AI-enabled HR

ChatGPT can be used for a variety of business applications, but for the
HR function some of the most enticing uses include:

Recruitment
ChatGPT can automate repetitive tasks in the recruitment process, like
screening resumes and scheduling interviews. ChatGPT can also assist in
candidate engagement by providing real-time support and answering fre-
quently asked questions about the company and the application process.
Plus, HR professionals can use ChatGPT to identify potential candidates
who may be a good fit for the organization based on their skills, experience
and education.

Employee onboarding
HR teams can set up ChatGPT to provide real-time support and guidance to
new hires – including answering common questions about company p ­ olicies,
procedures and benefits, and providing guidance on completing required
paperwork. ChatGPT can also help to automate administrative tasks, such
as scheduling orientation sessions or sending reminders to new hires about
required training.

Training
ChatGPT can provide employees with instant access to training materials
and answer questions about workshops and programmes. HR professionals
can also automate administrative tasks with ChatGPT, including scheduling
training sessions or providing reminders to employees about upcoming
events. ChatGPT can even create personalized training plans for employees
based on their specific needs and skill sets.

Performance management
ChatGPT can assist in the performance management process by providing
managers with guidance on conducting performance evaluations and an-
swering employee questions about performance metrics or feedback.
Meanwhile, HR professionals can get real-time insights on employee
­performance, and they can set up alerts that go to the human resources team
as well as to individual managers.

HR chatbots
ChatGPT can be used to develop conversational chatbots for HR departments
to improve the overall employee experience. Through chatbots ­powered by
 The data, analytics and AI tools available to HR 33

AI, employees can get instant support for common HR-related q

­ uestions
about things like benefits, vacation policies or payroll.

Employee engagement
ChatGPT can be a useful tool for improving employee engagement by pro-
viding workers with real-time support, personalized assistance and instant
access to information and resources. It can assist in answering questions
about company policies, culture, procedures and benefits, as well as provide
guidance on completing various forms and requests.

Compliance
Finally, HR professionals can also use ChatGPT to ensure that their HR
policies and practices are consistent, accurate, and compliant with local and
national regulations – so they avoid legal or reputational risks that may
­result from noncompliance.
ChatGPT can also assist in providing employees with up-to-date infor-
mation on compliance-related matters like employment laws, payroll and
tax regulations, and health and safety protocols. The model can also help
HR departments track and monitor employee compliance with HR policies,
such as attendance, leave requests and work hours.
Bottom line, ChatGPT can be a powerful tool for HR professionals, help-
ing to automate repetitive tasks, provide real-time support to employees and
enhance the overall employee experience. Automating these types of tasks
can free up HR professionals to focus on more strategic activities.

Getting started with ChatGPT

At the time of writing, ChatGPT is in beta and has already attracted millions
of users. If you want to have a go at using the tool, you can. Simply go to
chat.openai.com and click ‘Sign Up’ to set up your account. Once you’ve
created your account, ChatGPT will provide examples of what you can do
with the tool, and help you through the process of running your first queries.
For now, it’s free to use, but it will probably be monetized in future.

A quick word on combining analytics

Often, to get the most out of data-driven HR you won’t be able to rely on
one analytics tool alone – rather, you’ll need to combine different types of
analytics. For example, corporate culture analytics may tell you that your
culture is moving away from the values you’ve prioritized, but you may
34 Laying the groundwork for data-driven, AI-enabled HR

need text and sentiment analysis to tell you why that is. The idea behind
combining analytics is to base your decision making and HR operations not
just on what one set of analysis is telling you. Combining information from
more than one source and using different analytic approaches allows you to
verify insights from more than one angle.
All the approaches I’ve outlined in this chapter show only some of the
analytic possibilities available to HR teams today. Just a few years ago,
much of this wasn’t possible; we couldn’t do sentiment analysis on text,
for example. Analytics in particular has made such huge leaps that no one
knows for sure what’s going to be possible in 10 or even five years’ time.
Therefore, an important part of data-driven HR is staying open to new
opportunities that data and analytics may provide further down the road.

Tapping into other data-driven

and AI-enabled tech
Before we move on, I want to briefly pause and look at other data-driven
and AI-enabled technologies that will prove useful to HR teams – specifi-
cally, blockchain and the metaverse. I briefly mentioned both in Chapter 2,
but let’s explore them in a little more detail to see how these technologies
relate to HR.

The potential uses of blockchain in HR

Blockchain first came about as the technical backbone of Bitcoin, but has
since evolved into many other uses. Blockchains are essentially a (relatively
new) way of storing data online, built around two core features: encryption
and distributed computing. Encryption means that the data stored on a
blockchain can only be accessed by people who have permission to do so,
with the use of a special ‘key’. And distributed computing means that the
data is stored across many, many computers or servers, not on one central-
ized server. This means no one (other than the person who owns the data)
can access or alter the data without permission.
Another thing that makes blockchain unique is the way in which block-
chains are structured – basically, as blocks of data chained together, hence
the name. Each ‘block’ is linked to the previous block, thus forming a chain
of information. Each block also contains a timestamp to record when the
information in that block was created or edited. And, of course, each block
is encrypted. As for the information contained within the block, it could be
 The data, analytics and AI tools available to HR 35

anything: a record of a transaction, a record of ownership rights, personnel

information or any other record of information.
Because users can only access and edit the block they ‘own’ through a
private cryptography key, blockchain is very secure. Which is why block-
chain is commonly used to help secure and streamline transactions, espe-
cially cryptocurrency trading. Such transactions may also involve smart
contracts – contracts that execute automatically when agreed-on conditions
are met. In the future, smart contracts and blockchain could streamline any
processes that are heavily reliant on transactions and contracts.
But blockchain is also finding interesting uses beyond transactions. Some
governments, including South Korea, are exploring the use of blockchain to
manage government records, including digital credentials such as driving
licences.4
In HR, blockchain uses may include:

●● Verification of credentials. Blockchain can be used to securely store and

verify employee credentials, such as education and employment history.
This can help streamline the hiring process and reduce the risk of
fraudulent resumes.
●● Payroll and compensation. Blockchain can be used to automate payroll
management and ensure that employee compensation is distributed fairly
and transparently.
●● Benefits administration. Blockchain can be used to securely store and
manage employee benefits information, such as health insurance and
retirement plans. This can help reduce administrative costs and reduce
errors.
●● Performance management. Blockchain can be used to track and verify
employee performance metrics, such as sales numbers or customer
satisfaction scores.
●● Employee engagement. Blockchain can be used to create decentralized,
employee-driven platforms for communication and collaboration. This
can help to foster a sense of community and engagement among
employees, and enable them to contribute to company goals in a more
meaningful way.

HR and the metaverse

The concept of the metaverse has existed for a while. In the Matrix movies,
where humans are locked into a shared virtual world created by machines,
that’s essentially a (disturbing) depiction of a metaverse. The same with
36 Laying the groundwork for data-driven, AI-enabled HR

Ready Player One (a book and film in which people take refuge from our
dystopian future world in a massive online role-playing game and virtual
society, complete with its own currency). So it’s not a new concept invented
by Mark Zuckerberg, despite Facebook’s clever rebrand to Meta. The
metaverse is a concept that society has been building towards for a while –
ever since the emergence of the internet, social media, virtual reality and
early attempts at creating shared digital environments, like Second Life (the
online 3D environment where users engage in an alternative life and assume
an alter ego). The metaverse will, in essence, be the next generation of the
internet.
The concept of the metaverse is still evolving, but based on Mark
Zuckerberg’s vision of an internet that you’re inside of, rather than just
looking at, it makes sense that more of our everyday lives will take place
inside these shared metaverse worlds. And that includes work.
Which means HR teams must get ready for the metaverse. Some of the
ways in which HR could harness the metaverse include:

●● Virtual recruitment. The metaverse could provide a platform for virtual

recruitment, allowing employers to interact with potential candidates in
a virtual environment. This could save time and money by reducing the
need for physical meetings and travel.
●● Virtual training. The metaverse could also be used for virtual training,
providing employees with a more immersive and interactive learning
experience. This could be particularly useful for remote workers.
●● Virtual collaboration. The metaverse could provide a platform for virtual
collaboration, allowing teams to work together in a virtual environment.
This could be particularly useful for remote teams, who may struggle
with communication and collaboration in a purely digital environment.
●● Virtual meetings. The metaverse could also provide a platform for
virtual meetings, allowing teams to interact with each other in a more
natural and immersive way than straight-up video conferencing. This
could help to reduce the fatigue and disengagement that can come with
long video calls (farewell, Zoom fatigue).
●● Virtual workspaces. The metaverse could provide a virtual workspace
for employees, allowing them to work from anywhere in the world in a
fully immersive and customizable environment. This could help to reduce
the cost and environmental impact of physical offices, and provide greater
flexibility for employees.
 The data, analytics and AI tools available to HR 37

Both blockchain and the metaverse are enabled by AI, which in turn means
they’re reliant on data. So let’s turn to the magic ingredient that makes all
this possible.

The secret sauce: Data, data and more data

If you’re going to use AI and analytics, you need data. Let’s explore the data
sources available to HR teams, including some of the newer types of data
available.

A quick primer on the different types of data

Before we jump in, it’s important to remember that no one type of data is
inherently better that any other kind. Using data successfully is about find-
ing the data that works best for you and your use cases, and that’s likely to
be very different to what’s best for another use case, or a different business.
With so much data available these days, the trick is to hone in on the exact,
specific data that holds the most value for you.
The more data, and the greater variety you have, the fuller the picture
that you can draw – and the more accurate your analytics will be. In my
experience it’s often a combination of structured and unstructured, internal
and external data that provides the most valuable insights. (Don’t worry,
we’ll define these types of data next.) For example, you may well need to
mix some structured, internal data (such as ratings from employee micro-
surveys) with some structured external data (data from job portals,
­perhaps), as well as some unstructured internal data (transcripts from em-
ployee interviews, for instance) and unstructured external data (like social
media data). Often, it’s by combining data that we find the most valuable
insights.
Now let’s properly define those types of data.

Structured data
Structured data is any data or information that’s located in a fixed field
within a defined record or file, usually in databases or spreadsheets.
Essentially, it is data that is organized in a predetermined way, usually in
rows and columns. As such, structured data doesn’t really require smart AI
algorithms; rather, it’s based on simple ‘if this, then that’ logic.
38 Laying the groundwork for data-driven, AI-enabled HR

The average business has the potential to tap into a vast amount of struc-
tured data, including employee data, customer data, performance data, sales
data, number of website visits and any kind of machinery data points.
Structured data is, for now, the most commonly used type of data in busi-
ness – yet it represents just 20 per cent of all the data available in the world.
The remaining 80 per cent of data out there is unstructured (which we’ll get
to next). As more of the world becomes covered by cameras and more of our
conversations and activities take place online, we can expect that the pro-
portion of unstructured data will increase even further. There’s a lot of un-
discovered value waiting to be found, in other words.
As well as representing just a small portion of all data available, struc-
tured data is also less ‘rich’ than unstructured data. It offers us a more lim-
ited picture. Which is why it’s often wise to use other data sources alongside
structured data to enrich your insights. For example, structured data might
tell you that customer satisfaction dipped in the last quarter. But it won’t tell
you why. For that, you might need to tap into some unstructured data.
On the upside, structured data is usually cheap to use, easy to store and
easy to analyse with simple tools.

Unstructured and semi-structured data

Unstructured data is the term for any data that doesn’t fit neatly into tradi-
tional structured formats or databases – including email conversations,
website text, social media posts, video content, photos, audio recordings
and so on. It’s often text-heavy, but may also contain numerical data, or
other types of data such as voice recordings and images. Now, thanks to AI
analytics, it’s becoming increasingly possible to make use of this messy,
­complicated data.
Semi-structured data is a cross between unstructured and structured data.
This is data that may have some structure that can be used for analysis (like
metadata in a photo) but lacks the strict structure found in databases or
spreadsheets. For example, a tweet can be categorized by author, date, time
and length, but the content itself is generally unstructured. To analyse the
text in that tweet, you’d need a text analysis tool, ideally one that can carry
out sentiment analytics to understand the ‘mood’ behind the text.
It should be clear by now that the main downside of working with
­unstructured data is that it’s complex stuff, usually requiring specialist ana-
lytics tools. Unstructured data also tends to be much bigger than structured
data, which means you need bigger and better storage, too. None of this
should put you off using unstructured data. I firmly believe that it’s where a
 The data, analytics and AI tools available to HR 39

lot of the undiscovered value of data lies. But it does mean you need to be
very clear about what you’re aiming to achieve and what data you need in
order to do that.
Of course, the big advantage of unstructured and semi-structured data is
that there is so much of it, and it is highly descriptive and rich in content.

Internal data
No surprises here, internal data refers to all the information your business
currently has or has the potential to collect internally. It can be structured in
format (like an employee database or transactional records) or it could be
unstructured (like conversational data from customer service calls or feed-
back from employee interviews). It is your private or proprietary data that
is owned by your business – and this means that only your company ­controls
access to the data.
One downside to internal data is that you’re responsible for maintaining
and securing it. This costs money and, particularly in the case of personal
data, there are strict data protection and privacy regulations to comply with.
(Whereas, with external data, the data supplier takes on that liability for
you.) Also bear in mind that internal data on its own may not provide
enough information, and you may need to supplement it with external data,
such as social media data. Rather like blending structured and unstructured
data to get a really rich picture of what’s going on, often it is necessary to
combine internal data with external data to get the most useful insights.
On the upside, internal data is easy to access. You’re never at the whim of
a third party that can jack up prices or cut off access to the data at any time.
What’s more, there’s real value in your internal data because it’s already
tailored to your business or industry. So, while you may need to look at
some external data alongside your internal data to get the best results, you
should never overlook it altogether.

External data
External data is the infinite array of information that exists outside of your
organization. This can be publicly available (like certain government data)
or privately owned by a third party (like LinkedIn), and it can also be struc-
tured or unstructured in format. Key examples of external data include so-
cial media data, Google Trends data and government census data. There are
plenty of ready-made datasets available, but if you need something more
bespoke you can pay a third-party provider to gather the data for you.
40 Laying the groundwork for data-driven, AI-enabled HR

The obvious downside to external data is that you don’t own the data
and you will often have to pay for access. You’ll need to weigh up the risks
and the costs of accessing external data against the risks and costs of not
having that data. Would you have to go to the trouble of creating it yourself,
for example? What would happen if you didn’t use that data at all? Would
it stop you meeting your strategic goals?
There are some significant advantages to external data, though. The
Amazons and Metas of this world are lucky enough to have huge amounts
of internal data to work with. That’s great for them. But many businesses
can never dream of having that much data at their disposal. External data
fills that gap and helps to democratize data for all types of businesses.

Exploring newer types of data

Almost everything we do these days leaves a digital footprint. And that
means many new types of data are popping into existence. Let’s explore
some of these newer types of data and how they may prove useful to HR
teams.
Just to be clear, all of these newer types of data still fit into the category
of either structured, unstructured or semi-structured data, and can exist as
internal or external data. I have simply grouped them together here because
they represent some of the biggest leaps in data – which makes them useful
considerations for any HR function.

Activity data
Activity data, which provides a record of human activities or actions
(whether online or offline), can be incredibly valuable from an HR perspec-
tive. Think about all the things you do in the course of a normal day – it all
generates activity data. If you wear a fitness band with a sleep tracker, like I
do, even your sleep generates activity data. Then you travel to work, per-
haps paying for a ticket with a bank card or touching in with a travel card.
Your phone generates records of your location while you’re on the move. If
you make or receive a phone call on your way to work, or post a photo on
Instagram, that generates data. Then you get to work and send countless
emails, type hundreds or maybe thousands of words a day, and look at nu-
merous web pages.
The sheer volume of activity data available to HR teams can be over-
whelming, so it’s important to always refer back to your strategic objectives
and focus only on the data that helps you achieve your goals. But the real
 The data, analytics and AI tools available to HR 41

advantage of activity data is that it allows you to assess what your employ-
ees actually do, as opposed to what they’re supposed to do, what they say
they do or what you assume they do.
By tracking activity data, companies can accurately monitor individual
performance, and use this information to identify top performers and those
who may need help. And, of course, when you know who performs well and
what characteristics top performers share, you can focus on hiring more
people who match those characteristics.
One example of this comes from video game start-up Knack’s collabora-
tion with Royal Dutch Shell. Knack’s video games, which were designed by
a team of data scientists, psychologists and neuroscientists, aren’t just about
having fun – they’re about measuring human potential. All sorts of factors
are logged as a player participates in the games: every move they make, how
they solve problems, how long they pause, and so on. This builds a thorough
picture of the player’s level of persistence, creativity and even intelligence, as
well as their ability to prioritize tasks, and how quickly they learn from
mistakes.
Royal Dutch Shell’s GameChanger unit, which is charged with identifying
the best business ideas from inside and outside the company, was extremely
interested in the potential of these games to improve the process of identify-
ing the best ideas. So they devised an experiment: 1,400 Shell employees who
had previously proposed ideas to the GameChanger team were asked to play
a couple of Knack’s games. The GameChanger team then shared with Knack
information on how well three-quarters of the players had fared as idea gen-
erators (whether their ideas made it all the way, for instance). Knack used
this info to developed game-play profiles of the best idea generators in com-
parison to the weaker ones. Using information based on these top innova-
tors’ game profiles, the Knack team was then asked to guess from the
remaining quarter of the players who’d had the best ideas. They did this with
startling accuracy, clearly identifying those who had previously generated
winning ideas based only on the way they played the games.
Based on this, Knack and Shell were able to identify the key characteris-
tics of top idea generators, such as social intelligence and task-switching
ability. This allows the GameChanger unit to devote more time to those
employees whose ideas are likely to have more merit.

Conversation data
‘Conversation’ in this context doesn’t just mean two employees having a
chat around the coffee machine. It covers any conversation people may have
in any format, whether it’s a call with a customer, an instant message sent
42 Laying the groundwork for data-driven, AI-enabled HR

via phone or computer, company emails, social media posts and more. It’s all
conversation data.
This type of data is incredibly valuable to HR teams because it can give
in-depth insights into how happy and engaged your employees really are, as
well as how positive your employer brand is. Thanks to advances in analyt-
ics, conversations can now be mined for the content itself (what is said) as
well as context (how it’s said). This means companies can tell how happy,
irritated or stressed an employee (or customer) is, or even if they’re telling
the truth, just by analysing the tone of their voice.
As an example, conversation data can be useful in understanding what
makes an employee want to leave (or, for that matter, stay with) the com-
pany. By analysing text from open-ended questions on surveys and in exit
interviews, as well as social media posts, emails and team assessments, HR
teams can now accurately predict what makes an employee more likely to
leave or stay with the company.
Hiring is another area that can benefit from conversation data. It’s not
uncommon for employers to scour social media profiles for glimpses into
what potential hires are really like. This could potentially be done on a
larger scale to identify the types of content and the sentiment behind things
that successful employees post on social media, and use that knowledge to
assess potential candidates in future.
For many companies, email is an especially rich source of conversation
data, giving insights into employees’ productivity, treatment of colleagues,
and so on. Text analytics software is getting better and cheaper all the
time, making it possible for companies to search through employees’ email
traffic, hunting for words, phrases or patterns of communication that are
linked to certain success (or failure or attrition) metrics.
It goes without saying that there are implications to gathering and ana-
lysing conversation data. When it comes to phone calls, for instance, gener-
ally speaking, you can’t record customers or employees just because you feel
like it; what you’re recording must be relevant to the business. You may also
need to inform the parties that they are being recorded. Use of email data
can also be restricted depending on where you are in the world. But we’ll
talk more about such pitfalls and challenges in Chapter 4.

Photo and video data

The amount of photo and video data has exploded in recent years, largely
thanks to the proliferation of smartphones and the increasing use of CCTV
(especially where I am in the UK). Photo and video data can be large, which
can make it more expensive to store and work with. However, your ­company
 The data, analytics and AI tools available to HR 43

may already be collecting this data as a matter of routine (perhaps through

security footage), so it may not be very difficult or expensive to find new
ways to use that data more intelligently. If your company isn’t collecting
photo or video data already and you’re interested in doing so, make sure
you have a clear business case for working with this data (purely because it
can be expensive).
But how could you use photo and video data? One example is using
video data and video analytics to detect automatically safety violations or
where employees are not wearing required personal protective equipment.
Indeed, during the Covid-19 pandemic, many PPE detection tools were de-
veloped, some just using data from ordinary CCTV cameras.

Sensor data
Sensors are being built into an increasing number of products, from factory
machinery to office chairs. And these sensors generate a wealth of data that
can help HR departments improve their functions, including employee per-
formance, employee safety and more.
Because sensor data tends to lack context (it’s just telling you what the
sensor recorded at any given time, not what might have caused that), keep
in mind that it may need to be combined with another dataset to get the best
results, depending on what you’re trying to achieve. However, sensor data is
self-generating, meaning it’s very easy to capture, once the data capture
tools have been put in place. Some devices, such as smart phones, contain
ready-to-use sensors that can be used to the company’s advantage (for ex-
ample, a delivery company using their drivers’ phones to track driver
­behaviour, for instance).
Clearly, wearable technology has a huge role to play in this type of data,
and the workplace wearables market is booming. Honeywell’s Connected
Worker solution is just one example of this. Using a series of connected
wearables sensors, the solution measures an employee’s heart rate, breath-
ing, motion and posture to assess whether they are under physical stress or
in potential danger (it can detect toxic gas, for instance). This kind of tech-
nology will become more and more common in the future, especially for
workers in physically demanding jobs, or those who work in dangerous or
isolated locations.

Gathering the data you need

Once you know what data you need, your next step is to identify how you’ll
access or collect it. I’ll talk more about building the right data and t­ echnology
44 Laying the groundwork for data-driven, AI-enabled HR

infrastructure in Chapter 14, but for now let’s look at some of the general
considerations around collecting data.

Sourcing data internally

A good starting point is to see whether the data you need already exists in-
ternally – or whether you have the capability to generate it yourselves, for
example, by collecting data from your systems, customers or employees. As
we’ve already seen in this chapter, you can now gather data from almost any
activity undertaken by the company’s employees, including test scores,
­interview answers, performance reviews and more.
And wherever the HR team and your company’s employees are currently
having conversations, there is an opportunity to collect conversation data.
For example, if you operate a telephone sales department or customer ser-
vice department, you can record those conversations and analyse the con-
tent and sentiment for useful insights into how staff are performing. Surveys,
emails, customer feedback comments, social media platforms, and so on, all
provide useful sources of conversation data.
Video and photo data can be obtained by simply starting to collect it
using digital cameras. For example, retailers can use their network of CCTV
cameras to analyse how the presence of staff members in certain sections of
the shop floor impacts how likely a customer is to buy something. And with
modern sensors being smaller and cheaper than ever (small enough to fit
into an employee’s badge, for instance), they can be incorporated into al-
most anything, from manufacturing equipment to office equipment.
Wherever possible, you want to put in place systems to collect or generate
the data you need automatically. Whether you want to collect activity data
related to employee productivity, or sensor data in a hazardous working
environment, or whatever, the data collection should ideally take care of it-
self. The whole point of data-driven HR is to free up HR time and resource
to focus on adding greater value to the company. If HR professionals are
engaged in lengthy data collection exercises, this completely defeats that
purpose. Of course, with any new data project, time is needed to set up, fine-
tune, maintain and assess those processes, but once that is done, you should
be looking to collect data with the minimum effort possible. That leaves the
HR team to focus on turning that data into insights… and then acting on
those insights. You’ll also need to consider when you will collect the data.
Does it need to be gathered in real-time, for example, or is it not particularly
time-sensitive?
 The data, analytics and AI tools available to HR 45

It’s clear that internal data is a vital part of any data-driven HR strategy.
But you may also need to combine that data with some external data to get
a fuller picture that truly answers your strategic questions.

Sourcing external data

There is a wealth of external data that is already out there. Indeed, many
companies exist purely to supply other companies with data. LinkedIn is
perhaps one of the biggest sources of HR-related data, but there are also lots
of smaller, more industry-focused data providers. So, even if you’re looking
for quite specialized data, there’s a good chance someone out there has it.
Social media platforms are obviously key sources of useful data, and they
provide a wealth of information on current, past and potential employees.
You can, for example, use sentiment analysis to find out what past employ-
ees are saying about your company culture online or how happy current
employees are with their working environment. Sentiment analysis can tell
us a lot about users’ feelings, opinions and experiences, without having to
trawl through individual posts one at a time.
Other sources of HR-relevant external data may include census data,
which provides a very useful source of population data, geographic data and
education data. This could be useful, for example, if you were looking to set
up a new office in a new location and wanted to assess the potential work-
force in that area using local demographics. In addition, weather data is
often used by companies to help plan staffing levels according to the number
of visitors expected on a sunny weekend, for example.

When the data you want doesn’t exist: The value

of synthetic data in HR
One increasingly important type of data is synthetic data. The idea here is
that, when real data isn’t available – perhaps because it would be too expen-
sive or intrusive to collect – you can create artificial data that mimics real
data in every way. This can then be used to train AI algorithms and can be
analysed in exactly the same way as real data can.
Synthetic data has a few advantages over real data. One is that it’s com-
paratively cheap to ‘collect’. (It isn’t collected, of course, it’s generated by
algorithms, but the end result is the same.) And, somewhat counter-­
intuitively, synthetic can sometimes represent the real world more accurately
than real data, because it’s not affected by biases that can influence real-
world data (more on this coming up in Chapter 4).
46 Laying the groundwork for data-driven, AI-enabled HR

But how is synthetic data generated? A popular technique involves gen-

erative adversarial networks that use the same algorithms as those used to
create ‘deepfake’ images of celebrities. In the case of synthetic facial data, for
example, the GAN can come up with thousands or millions of images of
entirely made-up people. These images are just as useful for training facial
recognition algorithms as pictures of real people, but can be generated far
more quickly than real facial images can be captured, without any of the
privacy or consent or bias implications.
Imagine how this approach could improve your HR datasets. If your
dataset isn’t very diverse, for instance, you can generate synthetic data to
overcome that bias. Synthetic data can also be used to increase the sample
size of HR data, leading to more accurate insights and better decision mak-
ing. What’s more, synthetic data is useful in HR contexts because it protects
employee privacy, and allows for the testing and refinement of algorithms
and models without exposing real employee data.
In other words, synthetic data enables HR professionals to work with
accurate, unbiased data, leading to more equitable HR policies and prac-
tices. As such, it’s well worth considering synthetic data for your use cases.
Here are just a few ways synthetic data may prove useful to HR:

●● Synthetic resumes can be generated to test and improve resume screening

algorithms. This allows you to ensure that the hiring process is free from
bias while maintaining employee privacy.
●● Synthetic employee data can be generated to test and develop HR tools
such as employee performance evaluation systems, employee satisfaction
surveys and employee retention models. This way, you can improve the
accuracy and effectiveness of these tools without compromising employee
privacy.
●● Synthetic job postings can be created to test and improve job matching
algorithms. This ensures that job openings are matched with the most
qualified candidates without discriminating against any particular group
of applicants.
●● Synthetic training data can be generated to improve machine learning
algorithms used in HR applications, such as chatbots, language
processing and sentiment analysis. Again, this helps you develop more
effective HR tools.
●● Synthetic diversity data can be generated to test and develop diversity
and inclusion policies and programmes. With such data you can analyse
the effectiveness of DEI programmes without compromising employee
privacy.
 The data, analytics and AI tools available to HR 47

●● Synthetic compensation data can be generated to test and develop pay

equity models and tools. As such, you can ensure that employees are paid
fairly and without any bias – again, while maintaining employee privacy.

Key takeaways

We’ve covered a lot of ground in this chapter, so let’s recap the main points:

●● There are many different types of AI-enabled analytics tools available to

HR teams, including image and video analytics, text analytics, sentiment
analysis and voice analytics.
●● Many tools are available as off-the-shelf solutions, requiring little or no
AI knowledge. The ChatGPT text generator is a great example of such a
‘no-code’ solution.
●● Don’t overlook the importance of other future technologies that are
underpinned by data and AI – particularly the metaverse and blockchain.
These will create many opportunities for HR to deliver a better service
and employee experience.
●● When it comes to data, HR teams can choose from structured data,
unstructured and semi-structured data, internal data and external data.
Within these categories, there are also new forms of data that can be
mined for insights – specifically, activity data, conversation data, photo
and video data and sensor data. It’s also possible to generate synthetic
data, allowing HR teams to work with unbiased, accurate data.
●● Remember, no one type of data is better than another. It all comes down
to knowing what it is you want to achieve and finding the data that best
helps you do that. Often, the real value lies in combining different types
of data and different analytics to get the richest insights.

As I’ve alluded to throughout this chapter, there are some significant

challenges to working with data – potential bias being just one of them.
In the next chapter, we delve into the challenges and pitfalls of data-
driven HR.
48 Laying the groundwork for data-driven, AI-enabled HR

Notes
1 J Vincent. Nvidia’s AI recreates Pac-Man from scratch just by watching it being
played, The Verge, 22 May 2020. www.theverge.com/2020/5/22/21266251/
nvidia-ai-gamegan-recreate-pac-man-virutal-environment (archived at https://
perma.cc/9K7J-6R2P)
2 S Alvarez. Elon Musk explains Tesla’s pure vision approach to Autopilot and
Full Self-Driving, Teslarati, 10 April 2021. www.teslarati.com/tesla-autopilot-
fsd-camera-approach-explained-elon-musk (archived at https://perma.cc/
NH5X-HAVE)
3 D Peck. They’re watching you at work, The Atlantic, December 2013. www.
theatlantic.com/magazine/archive/2013/12/theyre-watching-you-at-work/
354681 (archived at https://perma.cc/KVT4-3PUX)
4 F Hersey. South Korea’s digital identity blockchain prepares to add new
credentials, go international, Biometric Update, 27 December 2022. www.
biometricupdate.com/202212/south-koreas-digital-identity-blockchain-prepares-
to-add-new-credentials-go-international (archived at https://perma.cc/93DZ-
2CPS)
 49

Challenges and 04
pitfalls of AI and
data-driven HR
Data and AI can bring great rewards. But they also bring challenges, particu-
larly when it comes to the use of highly valuable personal data. Which is
why every organization needs to practise good governance when it comes to
data and AI.
Practising good governance means being aware of the legal requirements
concerning every aspect of your data- and AI-related activities, and ensuring
that you’re operating in an open, ethical, and transparent manner. It also
extends to having policies in place to determine exactly who has access to
data, and who is responsible for maintaining the quality and accuracy of
that data.
Assuming your organization has a data and AI governance policy already
in place (and it really should), you need to ensure your HR activities operate
within the scope of that policy. But you may also need to put in place poli-
cies that are specific to your HR remit. This may include defining exactly
who owns the various people-related data within the organization and who
is accountable for various aspects of the data (such as updating data and
making sure data is accurate). It’s also a good idea to appoint a data and AI
champion from within your team to coordinate with others in the company
on issues of governance, quality and privacy.
In this chapter, we’ll explore this notion of data and AI governance, in-
cluding the pitfalls and challenges you may encounter along the way.

Understanding the value of data

Good data governance is about treating data as a valuable business asset
that must be protected. The HR function in particular deals with data that
is very sensitive and extremely valuable. Why? Because it is personal data.
50 Laying the groundwork for data-driven, AI-enabled HR

Personal data is any information that’s related to an identifiable person.

If we think about the average HR team, some of the data routinely collected
on employees (and potential employees) includes:

●● Personal information, like name, contact details, date of birth, gender,

nationality and social security number.
●● Financial information, including bank account details, tax information
and other financial records.
●● Employment history, such as previous employers, job titles, employment
dates and reasons for leaving.
●● Health information, which may include disability status, medical records
and workers’ compensation claims.
●● Criminal records, including arrest records, convictions and other criminal
history information.
●● Compensation and benefits data, such as salaries, bonuses and healthcare
costs.
●● Education and qualifications, like degrees, certifications, institutions
attended and dates of completion.
●● Job application data, for instance, resumes, cover letters, references and
interview notes.
●● Recruitment and selection data, including assessment scores, background
checks and hiring decisions.
●● And performance metrics, such as sales numbers, customer satisfaction
scores and productivity rates.

Clearly this is all extremely sensitive data, and as such, it carries a heavy
governance burden. Depending on where you are in the world, there are
likely to be strict regulations that govern the use of personal data (more on
this coming up), obviously with a key emphasis on protecting that data, but
also gaining consent to use the data.
You will need to ensure that everyone who comes into contact with your
people-related data is aware of the value of this data, and the associated
governance issues. They must be aware of the privacy and permissions issues
surrounding that data. And they must be aware of the rules and procedures
for how that data can be used. Basically, if your people do not give data the
respect it deserves – if they don’t treat data as a precious business asset – it
can very quickly become a liability.
With that liability in mind, let’s dwell on the biggest risks that may apply
to your people-related data.
 Challenges and pitfalls of AI and data-driven HR 51

The biggest data-related challenges

Please do keep in mind that these are all huge topics in their own right – and
regulations and the threat landscape are constantly changing. Specialist
legal and technical advice is therefore recommended.

Legal and regulatory compliance

HR departments must comply with various data protection laws, such as
the General Data Protection Regulation (GDPR) in Europe, the Health
Insurance Portablility and Accountability Act (HIPPA) in the US, and the
California Consumer Privacy Act (CCPA). Failure to protect employee data
can result in penalties, fines and legal action – not to mention the reputa-
tional risk that comes with a data breach.
Data privacy laws vary greatly around the world. The EU arguably has
some of the most stringent rules, and in the US it can vary from state to
state. It’s therefore vital that any HR team gathering sensitive personal data
makes sure they’re operating within the laws of their country.
In Europe, the GDPR is designed to enhance data protection and the right
to privacy for EU citizens, giving them greater control over their personal
data and how it is used. It is also implemented into law in the UK.
The GDPR states that companies can only use personal data for the ex-
press purpose for which it was given. And that includes the personal data of
employees. Consent is therefore a critical pillar of GDPR. Employees must
explicitly opt in to allow a company to use their personal data, and they
must be made fully aware of how that data will be used. This can be clarified
through a simple data privacy statement distributed to employees. Crucially,
you can then only use the data for the purpose for which it was handed over.
The GDPR also sets out strict mandates around reporting the theft or loss
of personal data. So, in the event of any breach that affects employee data,
you will need to inform the supervising authority (in the UK that’s the
Information Commissioner’s Office) within a maximum of 72 hours. You’ll
also have to inform those individuals whose data is affected.
You may also be affected by rules that govern the transfer of data from
one country to another. Say, if you’re a European company but you have a
US office, or your data analytics provider is based in the US, you’ll be af-
fected by GDPR data transfer rules.
In the US, regulation around the use of personal data may be a little less
stringent, but there are still many things that can trip a company up. As Felix
52 Laying the groundwork for data-driven, AI-enabled HR

Wu, Professor of Law at the Benjamin N Cardozo School of Law, once told
me, ‘Unlike Europe, the US does not have comprehensive privacy regulation,
but this may actually make things more difficult for companies, which must
comply with a patchwork of varying state and federal laws.’ California, for
instance, has the CCPA, which is similar to GDPR. Various (but not all)
states have similar laws, drawing upon parts of GDPR or even the CCPA.
But the specifics vary across state lines, making it difficult for organizations
that operate nationally. That said, the US does have nationwide data privacy
and data protection laws for certain industries. A good example is HIPPA,
which governs how personal health information can be used. Basically,
wherever you are in the world, specialist legal advice is a must.
Even working within these regulations, employers can still gather (with
consent) huge amounts of useful data. In the relatively strict European
Union, for instance, employers can monitor staff emails, other electronic
messages and the websites visited during work hours, providing they have
good reason for doing so and have consent. This doesn’t give you carte
blanche to read all employee communications just because you feel like it,
but it is allowed when you have legitimate reason for monitoring employees’
communications on company resources (again, with consent and after in-
forming employees of how you intend to use that data). Telephone calls may
also be monitored for business purposes (with consent, of course).
What this means for HR teams is that any monitoring of employee com-
munications must be clearly explained in a privacy policy, employee hand-
book or contract and you should get employee consent for that monitoring.
You need to make it very clear what data you’re gathering in terms of emails,
instant messages, website usage, etc. and why you’re doing that. If there isn’t
a clear business reason for gathering the data – for example, to improve
performance – you shouldn’t be doing it.
I would also steer well clear of messages that are obviously of a very per-
sonal nature and are not related to the business. Essentially, you should seek
to strike a balance between the privacy of your employees and the needs of
the business, and be transparent about what you’re doing at all times. Which
brings me to…

Ethical issues and the need for transparency

As well as sitting on the right side of the law, you also need to ensure that
your data usage sits within the ethical boundaries set by the company. Say
your organization espouses a culture of openness and honesty, you can’t
 Challenges and pitfalls of AI and data-driven HR 53

then have your data-driven HR activities fly in the face of that culture.
Clumsily implemented or poorly communicated data projects can do a lot
of damage when it comes to employee trust and morale. So it’s important
not to gloss over this aspect of data-driven HR.
On the whole, though, we’re all getting a lot more used to the wealth of
data being collected and generated about us. When we sign up for a free
email service, we acknowledge the email provider’s right to read those
emails. When we use an app, we agree to the provider’s right to use our loca-
tion data, among other things. We’re all getting used to our everyday activi-
ties being tracked.
This doesn’t mean generating or gathering people-related data can be a
free-for-all. As I’ve emphasized, you must have a clear business case for col-
lecting data on your employees and this must be properly communicated.
Transparency is one of the key pieces of advice I give to every company I
work with. Another is emphasizing the benefits of data. You ideally want to
achieve widespread buy-in for the use of data, from the top-level executives
to front-line employees. And it’s easy for people to get on board with data
when they understand how it will benefit the company and them as employ-
ees. Just as hundreds of millions of people are seemingly happy for Google
to scan their emails in return for a free email service, your employees are
more likely to be happy with you using their data if they understand that
information will be used to improve their working environment, for ­instance.
In other words, transparency is vital, but so is adding value for employ-
ees. People are far happier for their data to be used when they feel they’re
getting something valuable in return, whether it’s better working conditions,
more effective management, a safer environment, or whatever.
The key to success in data-driven HR is therefore to be open about how
you want to use the data you collect, to operate ethically and offer genuine
value to your employees in return. When you provide value and can demon-
strate a clear business case, most people will be happy, especially if the data
is anonymized (more on this coming up).
So far we’ve talked about the ethical and legal risks around data. But
there are also some major technology risks around data.

The devastating impact of data breaches

The downside of generating more and more data is that it introduces new
vulnerabilities to the organization, by creating more data that someone
could potentially steal. These days, it seems like barely a week goes by
54 Laying the groundwork for data-driven, AI-enabled HR

­ ithout reports of yet another large-scale loss or theft of personal data.

w
Data breaches can lead to huge losses for businesses, in terms of legal costs
and financial compensation, as well as the damage done to the company’s
reputation.
And if you think that no one would be interested in stealing your em-
ployee-related data (as opposed to, say, customer credit card details), think
again. As we’ve already seen in this chapter, the wealth of people-related
data available to the average HR team contains personally identifiable infor-
mation that’s valuable to criminals. Names, addresses, social security
­numbers, banking details… if this data got into the wrong hands, it could
clearly lead to identity theft, causing significant harm to affected employees.
It could also seriously tarnish your employer brand.
Another threat to be aware of is ransomware attacks – where a virus in-
fects the system and makes valuable data unusable unless a ransom is paid.
Phishing attacks are often the way in to the company system. And now, with
powerful AIs like ChatGPT (see Chapter 3), the danger is attackers can
­create authentic-sounding, highly personalized phishing messages very ­easily.
There is also a rising threat from the Internet of Things, and its ever-­
expanding network of connected devices. The theory is simple – the more
devices that are connected to the internet means more possible attack vec-
tors for intruders who want data. The how and the why is a bit more
­complicated – what benefit would an attacker gain from taking control of a
sensor on a factory machine, for instance? Well, aside from causing mischief
(which is certainly the main motive for a good deal of hacking activity) the
likelihood is that they want to use it to take advantage of network vulnera-
bilities which would allow them to get at the real jackpot – other devices on
the network such as PCs or phones which are far more likely to hold s­ ensitive
and valuable information. Any HR team using IoT-related devices to gather
data needs to take their security very seriously, and apply the same level of
security (passwords, regular system updates, etc.) as you would for any
other device.

Defending your data

When you’re dealing with personal data, you’re responsible for its protec-
tion. You’re responsible for preventing data loss and breaches.
You can’t properly protect data if you aren’t entirely sure what data you
have. Therefore, an important first step is being aware of exactly what
people-related data you have, including where that data resides, exactly
 Challenges and pitfalls of AI and data-driven HR 55

what data is involved (critically, does it include personally identifiable in-

formation?), who that data is divulged to, how that data is processed or
analysed and how it’s then used within the organization. Don’t forget to
consider any data that may be used or processed by third parties (a payroll
company, for instance) or anything that’s stored with off-site data p­ roviders
or in the cloud.
Another important step is to follow a strategy of data minimization. In
other words, don’t collect every scrap of data that you possibly can ‘just in
case’. Rather, you should seek to gather only the very essential data, i.e. data
that can help meaningfully improve the company and add value. Indeed,
these days, data privacy regulations often restrict data collection to a mini-
mum. The GDPR, for instance, insists that any personal data collected must
be ‘adequate, relevant and limited to the minimum necessary for the ­purposes
for which the data are processed’.
Wherever you are in the world, restricting the amount of data you collect
to the minimum necessary is a wise approach, but you can also take steps to
anonymize personal data by removing any personally identifiably markers.
This way, it can be used for analytic purposes without being linked to an
identifiable person.
Say, for example, you’re analysing the performance of sales colleagues to
identify the key traits of successful salespeople in order to inform future re-
cruitment decisions. In this case, what you’re really aiming to do is hire the
best talent for your sales team and remove some of the guesswork from the
recruitment process. And if that’s the goal, do you really need the data you
gather to identify individual sales colleagues?
You must also take measures to ensure the data is secured, especially
when it isn’t anonymized. There are certain safeguards any business can put
in place to secure data and prevent data breaches. Such measures can
­include:

1 Encrypting your data, thereby rendering it far less useful to anyone who
might want to steal it. Homomorphic encryption is an excellent option.
Here, data is encrypted in such a way that it can be analysed while
remaining in its encrypted form – even the analytical algorithms don’t
‘see’ the unencrypted data.
2 Restricting access to personal data to only those who need access.
3 Having systems in place to detect and stop breaches while they are
happening.
4 Training your staff so they never give away secure information and are
aware of threats such as phishing emails.
56 Laying the groundwork for data-driven, AI-enabled HR

Do keep in mind that data security is a specialist field and it’s always a good
idea to consult with a data security expert, either inside or outside your
­organization.

AI pitfalls
So far, we’ve focused on issues surrounding data. But the use of AI also cre-
ates unique challenges – some related to the data itself, and others related
more to ethical questions around AI. Let’s explore these AI-related pitfalls in
more detail.

The thorny issue of AI ethics

AI truly has the potential to change business and wider society in ways that
we can’t yet imagine. The fundamental change that we’re seeing right now
is that, instead of only carrying out repetitive tasks – such as applying for-
mulas to figures in a spreadsheet (which computers already do exceptionally
well) – intelligent machines are increasingly being used to make decisions.
Inevitably, this will include decisions that affect the people who work for
your organization. Amazon, for instance, has used AI algorithms to auto-
matically track and fire hundreds of fulfilment centre employees.1 This is the
tip of the iceberg as we grapple with huge ethical questions around decision
making in machines. One commonly used example in AI ethics is how
should a self-driving car act when it has to decide whether it should crash
into a pedestrian or a brick wall (potentially injuring its own driver)?
The truth is, AI can be used for enormous good, but it can also be used
for nefarious purposes. Machine vision can detect cancerous growth in med-
ical scan images but it can also be used by totalitarian regimes to carry out
surveillance on their citizens. Natural language processing like ChatGPT
makes it easier than ever for humans to interact with machines, but it can
also be used to orchestrate scams and phishing attacks by impersonating
people.
Yes, these are extreme examples, but if you’re deploying AI systems
within your organization you’ll likely have to make some ethical choices. It’s
therefore important to think about how your use of AI fits with the com-
pany’s overall culture and ethics. What’s more, because AI systems rely on
data, you have to contend with the same issues around consent and privacy.
You have to gain consent for any work you do with personal data and
­ensure your AI use stays on the right side of the law.
 Challenges and pitfalls of AI and data-driven HR 57

But there’s often a difference between what’s legal and what’s ethical.
Amazon’s highly automated firing of employees is not likely to be illegal. But
ethical? You can make a strong argument that giving machines the power to
make decisions that massively impact employees’ lives is ethically iffy at
best. And it’s far from clear that the employees involved ever gave consent
for AI to have this much power over their lives.
Even if you aren’t giving intelligent machines the power to make such
life-impacting decisions, but merely to run processes more efficiently, care
must be taken. In 2018, IT contractor Ibrahim Diallo arrived at his Los
Angeles office to find he couldn’t get in because his security pass had been
revoked. When he managed to gain access, he couldn’t log into his computer
or any of the systems he needed to do his job, and shortly afterwards secu-
rity guards arrived to escort him from the building. His pay was also stopped.
Neither his manager or any other senior member of staff knew what had
happened, but after being forced to work from home for three weeks they
eventually found that an HR error had mistakenly flagged him up as having
been dismissed. At that point automated systems kicked into action, which
had no way of being manually overridden! The stress he experienced, and
his leaders’ inability to do anything about it, caused him to leave the com-
pany and find work elsewhere.2
And what about when employees aren’t fired by AI, but because of AI? As
more responsibilities are given over to machines, this changes the work that
humans do, and renders some jobs obsolete. The World Economic Forum
predicts that 85 million human roles will have been automated into redun-
dancy by 2025. However, over the same period, the boom in AI and other
advanced, automated technology will lead to the creation of 97 million new
roles.3 So AI and automation will create more jobs than are displaced.
But that doesn’t mean automation isn’t an ethical concern, as far as re-
placing human jobs is concerned. You must still consider the impact of AI,
automation and data initiatives on your human workforce. Is there a risk
that anyone will become redundant? If so, can they be reskilled and rede-
ployed elsewhere in the organization? (Read more about reskilling and pre-
paring the organization for changing skill needs in Chapter 6.)
Crucially, is there a danger that people would become redundant anyway,
regardless of whether you use AI? For example, would failing to deploy AI
create a situation in which your company is no longer competitive or viable?
There are many situations where not using AI is an unsound strategy. If
problems exist in your company that could be tackled with AI, do you have
an ethical obligation to use it?
58 Laying the groundwork for data-driven, AI-enabled HR

I can’t answer that question for you. But I do recommend that every or-
ganization looking to roll out AI should set up some kind of ‘ethics council’.
The size and scale of your council and the amount of resources dedicated to
it will depend on the size of the organization. But the important thing is to
have someone who has the responsibility of considering all the issues raised
here and how they impact your initiatives.
If you want to understand more about AI ethics, I highly recommend
reading the Organization for Economic Co-operation and Development
(OECD) AI Principles, which are designed to promote the ethical use of AI.4

Transparency around AI
Just as with data, transparency is vital when you’re working with AI. It’s
really important to explain how and why AI is being used. After all, if intel-
ligent machines are going to be making decisions that potentially impact the
people in your organization, then you should at least be able to explain the
decision-making processes used by the machines.
In AI circles, this is known as the ‘black box’ problem, whereby algo-
rithms can be so complex that it’s very difficult for humans to precisely un-
derstand how they do what they do. In other words, ‘We know it works, but
we don’t understand how it works!’ This is compounded by the fact that
some AI is deliberately made opaque by those who sell it, in order to prevent
it being easily copied.
You want to strive for ‘AI explainability’, so people at every level of the
organization can understand how AI is used and how it might impact them.
If you’re using off-the-shelf software, for example, choose a provider that
explains how their AI works. It’s also a good idea to have a grievance pro-
cess in place, whereby employees can raise concerns if they feel they’ve been
treated unfairly because of AI-related decisions.
Remember when we talked about getting buy-in for collecting data, so
that employees understand how the use of data benefits them, their job, and
the organization as a whole? The exact same thing goes for AI. So, as well as
being transparent about how and why you’re using AI, do be sure to sell the
benefits as well.

The environmental cost

I can’t talk about AI and data concerns without mentioning the environmen-
tal impact. All that computing power needed to train and run AI systems
 Challenges and pitfalls of AI and data-driven HR 59

burns up a lot of electricity. As an example, according to estimates, training

OpenAI’s GPT-3 language model took about as much electricity as 120 US
homes would consume in a year.5
Of course, AI creates the potential to drive efficiencies that can lead to
reductions in environmental impact – think of smart home thermostats, for
example, which help homeowners use energy more efficiently and in turn
cut emissions. It’s not as clear-cut as saying ‘AI is terrible for the environ-
ment’, but the environmental impact certainly warrants consideration.

A quick word on AI and ownership

IP rights and ownership is another issue that requires consideration. This is
simple enough – you just need to be sure you have the rights to use any al-
gorithms you’re employing, either under licence from your AI provider, or
because you’ve developed and own them outright yourselves.

Bias and the importance of ‘clean’ data

Now we bring it full circle back to data. Because AI algorithms are only as
good as the data they are trained on. A key part of governance is therefore
making sure that the data you are using is as clean as possible. What do I
mean by ‘clean’? I mean two things: data that is of high quality, and data
that’s free from bias. Let’s look at both issues.

Working with high-quality data

In essence, data quality refers to a standard of measures that can be used to
assess whether the data is fit for purpose. These measures are all equally
important for data quality:

Measurements for data quality

●● Consistency: This first measure means that the data in a dataset is all
recorded and collated in the same way. For example, if you have multiple
fields in a record, then every record should have all of the fields
complete. Fields should be used the same way across every record, and
if we ‘know’ something about one piece of information in the dataset, we
should know it about every piece of information, so we are always able
to use the data together.
60 Laying the groundwork for data-driven, AI-enabled HR

●● Accuracy: This means that the data is error-free. In the case of

automatically collected data, the tools or sensors that have been used to
collect the data must be audited so that you know they’re working
properly. Of course, errors can occur with human-inputted data too, so
we have to ensure that those with data-entry responsibilities are trained
and aware of the need for accuracy.
●● Uniqueness: This means that there are no duplicate entries. If the same
piece of data is recorded more than once, it’s very likely that your
database will start to become inaccurate.
●● Validity: Validity is a way of measuring whether every record or piece of
data in a database is fit for the purpose it’s intended for. For example, are
dates all stored in the correct format, and are all of the figures stored in
the same way?
●● Timeliness: Timeliness measures whether your data is likely to be
relevant, with regard to the time at which it was collected. Some
processes need only be measured very occasionally in order to monitor
and understand them. Others need to be measured much more
frequently. And for those operations that require real-time datasets,
measurements must be taken and recorded with a delay that is as close
to zero as possible.
●● Completeness: Finally, completeness is a measurement of how much of
the total availability of data on a subject is captured in your dataset. The
greater the completeness of your dataset, the more grounded in reality
your insights will be.

If you want to ensure you’re working with high-quality data, it’s really im-
portant you audit your data using these metrics.

The problem of data bias

The second element of ‘clean data’ is bias, by which we mean data that is
not truly representative of the data subject. Usually this relates to the way
in which the data was collected. For example, if you are trying to measure
customer satisfaction using feedback forms, and you only send the forms to
customers who have left positive reviews, your data is going to be i­ nherently
biased.
 Challenges and pitfalls of AI and data-driven HR 61

With the huge and complex datasets used in AI initiatives, the potential
for bias to creep in is ever-present. This is a big challenge for many data ini-
tiatives, because biased data means your insights will not be informed by
objective reality. In fact, preventing bias (or at least reducing the damage it
can cause) is seen by AI thought leaders as one of the biggest challenges our
society will have to overcome, if the potential of AI is to be realized.
Biased data can be the result of poor data quality, as outlined above, but
sometimes, even if your data measures up well against all of the quality
metrics, bias can still creep in. This is because bias isn’t necessarily ‘wrong’,
in terms of whether the data is accurate, unique, valid or timely. It might just
mean you aren’t casting your net widely enough in order to get diverse
measurements, viewpoints or opinions. As a result, the models and simula-
tions you build won’t represent the real world.
There are very serious implications to data bias. In the US, when facial
recognition systems used by police forces to identify suspects in crowds were
audited, it was found that young, female, black citizens were far more likely
than any other age group to be misidentified.6 When applied to people in
this demographic, the accuracy rate of the algorithm was found to be 34 per
cent lower than with other groups. If left unchecked, this could clearly lead
to higher rates of wrongful stops, arrests or searches of young black women.
Recruitment is one area where data bias can cause real problems for HR
teams. Noel Sharkey, a Professor of AI and Robotics at the University of
Sheffield, has said that datasets used by recruitment algorithms that he has
studied are so riddled with bias that they simply shouldn’t be used until they
can be regulated and audited with the same degree of thoroughness as is
used for data in pharmaceutical trials.7 In 2018, Amazon stopped using a
machine learning algorithm that was designed to assess job applicants, after
discovering that it was ‘sexist’. Because far fewer women than men had ap-
plied to work for the company over the previous 10 years, the dataset that
the algorithm used was found to discriminate against women, passing them
over for opportunities for no reason other than the fact it did not have
enough data on female applicants for these roles.8
Clearly, you will need to take a careful look at the data and AI algorithms
you use to check for possible bias. Ask yourself questions like, ‘Does this
data exclude certain groups or skew in favour of certain groups?’ and ‘How
does this algorithm make decisions?’ (There’s that explainability issue
again.) And remember, as we saw in Chapter 3, synthetic data can help to
overcome inherent bias in data by creating new data that’s more ­representative
than your existing data.
62 Laying the groundwork for data-driven, AI-enabled HR

To make things even more complicated, sometimes it might be appropri-

ate to introduce bias into a system deliberately, in order to compensate for
social factors that lean towards unfairness or intolerance. In recent years,
Microsoft and IBM have both released AI-powered chatbots that later had
to be adapted (deactivated, in Microsoft’s case) to stop them acting in a rac-
ist and abusive manner. This was because they were learning how to con-
verse based on social media interactions which, of course, can often be racist
or abusive themselves. So they had no choice but to introduce an element of
deliberate bias into the system – in this case, telling the bot that it shouldn’t
be learning from racist or abusive data. Of course, this inevitably means that
the data the bots learn from is less representative of real life. However, as it’s
clearly unacceptable for a bot representing a company like IBM to be using
racist language and quoting Hitler, there wasn’t a lot of choice! Balancing up
the harm that can be done by introducing biased data against the harm done
by excluding it is an essential part of the process of governance, too.
We must also remember that AI, particularly when it comes to something
like recruitment, can help to overcome human biases. A well-trained algo-
rithm working with unbiased data will be less prone to prejudice and bias
than many human decision makers. So despite the risk of bias, AI has much
to offer when it comes to making better, fairer decisions.

Key takeaways

Let’s briefly recap some of the main lessons from this chapter:

●● HR teams deal with highly personal, sensitive data – data that is valuable
to the organization, but also attractive to hackers. Good data governance
is essential.
●● Data governance involves working in line with data protection
regulations, having clear data policies in place that govern the use of
data, and being open and transparent with employees about what data is
collected (with consent) and how it is being used. Data regulations vary
around the world (and from state to state in the US) so do seek specialist
advice for your jurisdiction.
●● It’s vital you protect data from threats such as ransomware attacks. To
protect your data, you first need to be crystal clear about what data you
have, and who has access to it (including third parties). You should also
encrypt data, restrict access to only those who need it, have systems in
 Challenges and pitfalls of AI and data-driven HR 63

place to detect and stop breaches and train staff to be alert to threats.
You should also follow good practices around data minimization and
anonymization wherever possible.
●● AI has its own ethical and technical challenges. For one thing, you
should be able to explain to employees how AI systems make decisions
(thereby avoiding the ‘black box’ problem). You must also be aware of
the potential for bias to creep in, and take steps to counter bias in your
data and AI systems (for example, using synthetic data). Plus, your data
needs to be ‘clean’ (accurate, complete, etc.).

Another challenge with data and AI is working out how best to use it in
your organization! With so many potential uses, prioritizing your use cases
is no easy feat. In the next chapter, we’ll explore how to find the most
­strategic uses for data and AI in your HR function.

Notes
1 V Tangermann. Amazon used an AI to automatically fire low-productivity
workers, Futurism, 26 April 2019. futurism.com/amazon-ai-fire-workers
(archived at https://perma.cc/N8SK-TZNT)
2 J Wakefield. The man who was fired by a machine, BBC News, 21 June 2018.
www.bbc.co.uk/news/technology-44561838 (archived at https://perma.cc/
R4FU-J6E4)
3 M Kande and M Sonmez. Don’t fear AI. It will lead to long-term job growth,
World Economic Forum, 26 October 2020. www.weforum.org/agenda/2020/10/
dont-fear-ai-it-will-lead-to-long-term-job-growth (archived at https://perma.cc/
LQ4P-CNMH)
4 OECD. OECD AI principles overview, OECD, May 2019. oecd.ai/en/ai-­
principles (archived at https://perma.cc/97VP-NQH2)
5 J Saul and D Bass. Artificial intelligence is booming – so is its carbon footprint,
Bloomberg, 9 March 2023. www.bloomberg.com/news/articles/2023-03-09/
how-much-energy-do-ai-and-chatgpt-use-no-one-knows-for-sure (archived at
https://perma.cc/EC33-4UBB)
6 A Najibi. Racial discrimination in face recognition technology, Harvard
University, 24 October 2020. sitn.hms.harvard.edu/flash/2020/racial-­
discrimination-in-face-recognition-technology (archived at https://perma.
cc/5EAQ-5PV2)
64 Laying the groundwork for data-driven, AI-enabled HR

7 N Sharkey. AI expert calls for end to UK use of ‘racially biased’ algorithms, The
Guardian, 12 December 2019. www.theguardian.com/technology/2019/dec/
12/ai-end-uk-use-racially-biased-algorithms-noel-sharkey (archived at https://
perma.cc/4WYH-TZ62)
8 J Dastin. Amazon scraps secret AI recruiting tool that showed bias against
women, Reuters, 11 October 2018. www.reuters.com/article/us-amazon-com-
jobs-automation-insight-idUSKCN1MK08G (archived at https://perma.cc/
MRZ6-PPPQ)
 65

Finding the right 05

ways to use data
and AI in HR
In order to get the most out of data and AI, it’s vital you act strategically.
You need a strategy that maps out how HR can best use data and AI to help
drive the business forward. In practice, this means looking at the work that
HR functions do, and how data and AI can improve those processes. With
this in mind, you can identify your priority objectives (within the context
of the organization’s wider goals, of course), pinpoint the data you need,
and so on.
This chapter sets out the process of creating a strategy for intelligent HR.
But why is strategy so important when it comes to data and AI?

Why it all starts with strategy

As we saw in Chapter 2, the explosion in data is affecting almost every area
of our lives, including work. We now live in a world in which the amount of
data being generated every day – even every second – is, frankly, astonishing.
And when it comes to what we should do with all this data, I’ve found that
many companies, or functions within organizations, fall into one of two
camps: some are so eager to ride the data train, they dive in and start collect-
ing all kinds of data simply because they can, with no thought as to how that
data benefits the business; others prefer to bury their heads in the sand,
often because they’re so overwhelmed they don’t know where to start.
The same can be said of AI. So far in this book, we’ve seen just a taste
of the cool things that organizations can do with AI (there are plenty more
examples coming up in Part Two), but it should be clear already that the
possibilities are evolving fast. And with AI-driven advancements like
blockchain and the metaverse coming into focus, things are only going to
accelerate.
66 Laying the groundwork for data-driven, AI-enabled HR

I don’t say that to put you off or intimidate you. Only to emphasize that,
with the technology landscape changing so rapidly, it’s easy to get lost in the
weeds. Which is why it’s so important to stay focused on your most pressing
goals – and how you can best use data and AI to achieve those goals. This is
where strategy comes in.
Having a clear strategy will also help you avoid overdosing on data.
Given the risks and pitfalls around data – especially personal data – it’s
never a good idea to start collecting huge amounts of people-related data
that you don’t really need. I always say that the power of data isn’t in the
impressive amount of data you can collect or the super-cool AI analytics
that are available. The power of data lies in how you use it. It’s about how
you use the insights that you glean from the data to improve decisions, bet-
ter understand your employees, optimize operations and add value to the
company. Therefore, you need to be very clear about what it is you want to
achieve, and what kind of data and AI analytics will help you achieve that.
Creating a robust data strategy will help you develop and maintain a laser-
like focus on your goals and map out your next steps.
But where do you start?

Begin with your objectives

Finding the best uses for data and AI means identifying the areas where data
and analytics can provide the most value for the organization.

Understanding potential uses of data and AI in HR

Within the context of HR’s role in the organization, there are three main
ways in which data and AI can add value to the business. They are:

●● generating better HR insights to aid decision making across the

organization
●● delivering a better service for employees (and candidates)
●● driving efficiencies in HR services

Let’s briefly explore each one.

 Finding the right ways to use data and AI in HR 67

Generating better HR insights to aid decision making across

the organization
Here, we’re talking about getting information to the people within the or-
ganization who need it most, providing people-related reporting, helping the
business understand what’s going on now, informing decision makers about
people-related challenges that may be on the horizon (using AI’s predictive
capabilities to create early warning systems), and helping the business plan
for the future (addressing skills gaps, for example).
As an example, networking hardware company Juniper Networks uses
LinkedIn data and analytics to analyse where top-performing employees
come from – but also where they go when they leave the company. This
helps the company better understand career paths and, in turn, make better
decisions on how to attract and retain talent.1
We’ll delve more into the topic of generating insights in Chapter 6, but
suffice to say that data and AI are absolutely fundamental to this process of
generating better, more forward-thinking insights.

Delivering a better service for employees (and candidates)

The HR function is all about serving the people who work in the organiza-
tion – providing a great recruitment service, awesome training and develop-
ment opportunities, a thoughtful wellbeing offering, a safe working
environment and all those other ways in which HR serves workers. HR
teams can use data and AI to make these services better, create a richer em-
ployee experience and add more value for employees at all stages of the
employee lifecycle. Walmart, for example, uses AI to help employees find the
right medical provider for their needs.2 We’ll explore plenty more examples
from various HR services in Part Two.

Driving efficiencies in HR services

AI in particular has much to offer HR teams who want to improve the effi-
ciency of their services. So we’re talking about things like automating cer-
tain HR processes (using chatbots, for example), using AI in onboarding,
using metaverse environments to make training more immersive, and so on.
Again, we’ll explore these various HR services in Part Two. But as a quick
example, Johnson & Johnson has used an AI automated writing tool called
Textio to scan for unconscious bias in its job postings. Using this tool,
Johnson & Johnson recruiters uncovered that the language in many of their
68 Laying the groundwork for data-driven, AI-enabled HR

job postings skewed masculine – but thanks to some simple AI-powered

edits, the company saw a 9 per cent increase in female applicants.3
When thinking about how best to use data and AI, I suggest you keep
these three branches of HR activity firmly in mind. They should serve as a
useful starting point when defining your strategic objectives.

Now it’s time to define your strategic HR objectives

How can you best use data and AI in your organization? Well, it depends on
your specific HR challenges and objectives.
What is it you most want to achieve as an HR function? Maybe you want
to improve recruitment and selection, reduce turnover, boost employee per-
formance, enhance your training and development offering... Every organi-
zation is different. Some HR teams may, for example, face greater operational
challenges, while other companies are suffering significant morale problems
and need to develop a better understanding of their people. You can’t begin
to look at data and AI options until you’ve identified your most pressing
goals and challenges.
Of course, the best kind of HR data strategy is directly linked to the or-
ganization’s wider objectives and, in effect, should cascade down from those
corporate objectives. To put it another way, you’re not just looking at HR-
specific goals and challenges, but also how you can add value to the organi-
zation and help deliver the company’s vision.
Therefore, your organization’s overarching strategic plan should inform
your thinking at this stage. In an ideal world, your organization’s strategic
plan would be a concise, simple document that anyone in the organization
can read and understand – something like a ‘plan on a page’ that clearly sets
out where the organization needs to go. However, this isn’t always the case
and I recognize that some organizational strategic plans are overly long and
complex, making it difficult to determine what actions need to be taken.
Whatever your company’s strategy looks like, it should set out intended
outcomes for the company, including financial and non-financial objectives,
and (hopefully) the core activities and enablers that will lead to those out-
comes being achieved. If you struggle to understand this from the company’s
strategic plan, have a discussion with your leadership team before going any
further. It’s vital you understand where the business as a whole wants to go.
Say, for example, your company has a corporate objective to become a
top-three provider of specific consultancy services within the next three
years. That will translate into various HR-related actions such as assessing
and optimizing your employer brand in order to attract the best talent. In
 Finding the right ways to use data and AI in HR 69

this case, the biggest opportunities for data-driven HR are likely to lie in
generating better insights and delivering an awesome recruitment and on-
boarding service.
Basically, this stage is all about identifying where the biggest data- and
AI-related opportunities lie for your HR function, based on your HR objec-
tives and wider corporate objectives.

How many objectives is too many?

I cannot recommend strongly enough that you keep this objectives phase
simple. Don’t be tempted to create a list of 100 HR objectives that cover
everything you could possibly want to achieve. Instead, focus only on core
objectives. After all, you can’t create a robust data strategy if you aren’t
crystal clear on what exactly you need to achieve and, in turn, what areas or
activities you need to focus on to achieve those aims. A list of 100 nice-to-
have objectives will lead to a very muddled (and probably very expensive)
strategy that delivers little real value.
I recommend identifying around three top priority objectives as a starting
point. These should represent your most pressing HR challenges, or goals
that will add the most value to the organization, based on its overarching
strategic objectives.
Some companies and functions like to map out their objectives as a sim-
ple ‘plan on a page’ – a one-pager that visualizes where the organization (or
function within the organization) is heading. This certainly isn’t essential,
but it does make it easier to communicate your goals, and serves as a useful
document to refer back to in future. But a simple list of objectives will do.
With your objectives firmly in mind, you’re ready to create your HR data
and AI strategy. Which brings us to…

Creating your strategy

To be clear, you’re not looking to create a lengthy strategy document that no
one ever reads because it’s so complex. A good HR data strategy can be eas-
ily broken down into a few simple sections that determine what it is you
want to do with data and AI. The following six questions will help you un-
derstand and really clarify what you want to do, and, as such, they form the
basis of any good data strategy.
But before we get into the questions themselves, keep in mind that you
may need some expert help to pull your strategy together and put it into
70 Laying the groundwork for data-driven, AI-enabled HR

practice – particularly from your in-house IT professionals and data scien-

tists, who will advise you on which data and AI solutions are technically
feasible. If your organization is small and doesn’t have the data knowledge
and IT expertise in-house, there are many data consultants who will help
you determine the best course of action for your needs.
As well as the IT team, it’s also important to involve other stakeholders
from across the organization, particularly business leaders, to ensure that
your intended uses are aligned with the organization’s objectives. Remember,
the right use cases for data and AI in HR will depend on the organization’s
needs, so you want to take a collaborative approach to creating your data
strategy. Plus, if your company has an ethics council (see Chapter 4), you’ll
want to work with them to ensure your chosen uses are ethical.
Now let’s get into the six questions or steps that you can use as the foun-
dation of your data and AI strategy…

1. What questions do we need to answer to achieve

our goals?
Surprisingly, a good data strategy doesn’t start with data. It starts with your
unanswered business questions. Many of the organizations and functions I
work with tend to ask for as much data as possible – not because they plan
to do very clever analytics, but because they don’t know what data they re-
ally need. By pinpointing your unanswered questions, you can hone in on
the data you need most.
So, having set out what it is you want to achieve (your objectives, from
earlier in the chapter), you now need to pin down the big questions you
must answer if you’re going to deliver those goals. Some of the questions
might have been identified already as you worked out your objectives, while
others will need careful thought at this stage.
Defining these questions helps you identify exactly what you need to
know. And by making sure your questions are linked to your company’s
priorities, you can ensure they’re the most strategically important questions,
rather than asking every little ‘nice to know but not essential’ question.

2. What data do I need to answer those questions

or solve those problems?
Look at each question you identified in step 1 and then think about the data
you need in order to answer those questions. Much of that data may come
 Finding the right ways to use data and AI in HR 71

from within the company itself, but you may also need to make use of exter-
nal data providers.
Establish what data you already have access to (from HR systems, em-
ployee surveys, performance metrics, etc.) and what you don’t yet have ac-
cess to. For the data you don’t have access to, do you need to partner with
an external provider or can you set up new data collection methods to
gather the data internally? Circle back to Chapter 3 for more on the data
itself.

3. How will we analyse that data?

Having pinned down your information needs and the data you require, next
you need to look at your analytics requirements, i.e. how you will analyse
that data and turn it into valuable insights that help you answer your ques-
tions and achieve your goals.
As we saw in Chapter 3, much of the promise of data lies in using machine
learning to analyse unstructured data, like email conversations, social media
posts, video content, voice recordings, and so on. Combining this messy and
complex data with other more traditional data, like KPIs or sales data, is
where a lot of the value lies. And remember, these days there are many off-
the-shelf AI analytics solutions that will help you extract value from data.
Again, you can circle back to Chapter 3 for a refresher on the different types
of analytics, but we’ll also explore analytics more in Chapter 6.

4. How will we report and present insights

from the data?
Data is only really valuable if you can turn it into insights and actionable
knowledge. To put it another way, data is absolutely useless if it isn’t pre-
sented to the right people in the right way at the right time, so that the right
actions can be taken. This is how businesses gain competitive advantage.
Perhaps the most important thing to remember at this stage is to keep
your target audience in mind. Therefore, you need to define the audience for
your data (i.e. who requires insights from the data?) and work out how best
to get that information to them. The HR team itself may be the largest audi-
ence, but no doubt you will also need to present insights to others elsewhere
in the organization, including the leadership team and managers across the
business. Indeed, as we saw earlier in the chapter, communicating insights
from data and aiding decision making across the business is one of the
72 Laying the groundwork for data-driven, AI-enabled HR

c­ ritical ways for HR teams to add value through data and AI. We’ll delve
into this in more detail in Chapter 6.
What’s the best way to disseminate insights to the people that need them?
Options for reporting and presenting insights vary from fancy dashboards
with real-time data through to simple reports with key insights presented as
visuals. The right option for you depends on what you’re measuring, who
needs to know about it, and how you usually communicate across the com-
pany.
However you decide to disseminate the information, keep in mind that
people are less likely to take action if they have to work hard to understand
what the data is telling them. It’s therefore vital that insights are presented
in a clear, concise and interesting way – this ensures the critical messages are
understood and there’s no room for misinterpretation.
You might be wondering why you need to think about disseminating in-
formation at such an early stage. The reason is your method(s) for present-
ing data may influence your data infrastructure requirements. Which leads
us to…

5. What are the infrastructure implications?

Having defined what data is needed, how it will be turned into value, and
how it will be communicated, the logical next step is working out the infra-
structure implications of these decisions. Essentially this comes down to
what software and hardware you’ll need in order to capture, store, analyse
and communicate insights from the data you have identified.
For example, if you’re looking at gathering significantly more perfor-
mance data, is your current data storage technology up to the task of storing
all that new data, or do you need to supplement it with other solutions?
What current analytic and reporting capabilities do you have and what else
do you need to access?
We’ll talk more about technology and infrastructure in Chapter 14 but,
again, you will need to involve your in-house experts or an external data
consultant.

6. What action needs to be taken?

Having answered the five questions above, you’re now ready to define an
action plan that turns your HR data strategy into reality. Like any action
plan, this will include key milestones, actions and owners of those actions.
 Finding the right ways to use data and AI in HR 73

As part of this step, you will also need to identify training and development
needs to help you put this plan into action, and pinpoint where you might
need external help.

Making the business case for data- and

AI-driven HR
There’s no doubt that getting the leadership team and key decision makers
involved will help you create a more robust data strategy. Not only that,
getting leadership’s buy-in at this crucial early stage means they’re more
likely to put your people-related data to good use in their own decision
making.
Thus, an important part of creating a robust data strategy is making a
strong business case for a data-driven HR approach, to help get people
(both inside and outside the team) on board with the idea of data-driven
HR. The more people are aware of and excited by the possibilities of data
and AI, the more likely they are to buy into the idea.
This extends across all levels of the company and all functions, not just
the company leadership. After all, data-driven HR is about people – and
their data. When the people in an organization understand what data-driven
HR is all about, and how it benefits the company as a whole and them as
employees, they’re more likely to be on board with, for example, capturing
new kinds of employee data. When the business case for data-driven HR
isn’t communicated properly at every level, it can breed mistrust and have
serious negative consequences on the organization’s culture.
Making a business case for data-driven HR is a bit like an entrepreneur
making a business case (or business plan) for their new venture. So, natu-
rally, you’ll want to do the same sorts of things as an entrepreneur would in
their business plan. This includes giving a good outline of the data strategy
and its goals (i.e. what you’re hoping to achieve with data), as well as the
tangible benefits to the business and its employees. It’s also vital you’re open
and realistic about the timeframe, likely disruption to the business and
costs, especially in discussions with the leadership team. You need to make
the best case for data-driven HR, which means it’s important not to gloss
over these issues.
‘Selling’ data-driven HR is a crucial consideration on the way to intelli-
gent HR. It instils confidence in data, inspires feelings of trust and transpar-
ency and emphasizes the HR team’s value to the company as it works to
74 Laying the groundwork for data-driven, AI-enabled HR

achieve its goals. Plus, when you want your HR data to be used by other
functions across the company, ensuring everyone understands the value of
your people-related data means they’re much more likely to incorporate
that data into their decision making further down the line. By making a
business case now, you’re sowing the seeds for data-driven decision making
and adding value through data in the future.
How you communicate your plan for data-driven HR depends on a num-
ber of factors, like how big your company is and the usual process for kick-
ing off new initiatives. One good way to go about it is by distilling your data
strategy into key points that can be communicated in a short presentation.
Keep it simple and brief (there’s no need to go into masses of detail on ana-
lytic possibilities, for instance) and remember that your enthusiasm for this
new age of data-driven intelligent HR will be infectious. Use examples to
demonstrate how other companies are leading the way in data-driven HR
(you’ll find plenty of these in Part Two). And remember to focus on the ben-
efits that data-driven HR will bring, both to the organization as a whole and
the people who work there.

Returning to your strategy in the future

No strategy is ever set in stone. Things change, markets shift, organizational
priorities evolve, and so on. It’s therefore very likely that you will need to
revisit your data strategy on a regular basis (annually, at least) to check it’s
still in line with the company’s overall priorities. Even if nothing has changed,
revisiting the strategy helps you to stay lean and remain focused on your
desired outcomes.
Also keep in mind that, as you get further down the road of data-driven
HR, significant new opportunities or questions may present themselves. For
example, when answering one of your strategic questions, the data may
throw up other, more pressing questions that also need to be answered, and
this may lead to a slight tweak in your strategy. The technology around data
and analytics is evolving fast and what’s possible in one or two years’ time
may be completely different from what’s possible now. While you want to
follow through on the actions in your strategy, remember that the point of
data-driven HR is to add greater value to the organization and do things in
a more intelligent, streamlined way. It therefore makes sense to stay alert to
new ways of doing that.
 Finding the right ways to use data and AI in HR 75

Key takeaways

Let’s finish up with a quick recap on strategy:

●● It’s vital you have a strategy that maps out how HR will use data and AI
to help drive the business forward. Having such a strategy in place
ensures you stay focused on your goals and target your resources most
effectively.
●● There are three main ways data and AI can add value to the work that
HR teams do: generating better insights, providing a better service to
employees and driving efficiencies in HR.
●● Any good strategy starts with your objectives. What are your biggest
people-related challenges or most pressing HR goals? When identifying
your objectives, ensure they link to the organization’s overarching
strategic goals. And do try to keep your list of objectives to just a few key
goals.
●● With your objectives in mind, you can now begin to create your data
strategy, using the six key questions outlined in this chapter. It’s really
important you have wider buy-in for your data strategy, so be sure to
involve key stakeholders – including leadership and the IT function. If
you do not have in-house data/IT expertise, a data consultant will be
able to help you develop your strategy.
●● Finally, don’t forget to revisit your strategy on a regular basis to check it’s
still in line with the company’s overall priorities. Revisiting the strategy
also helps you remain focused on your desired outcomes.

As we’ve seen in this chapter, generating better insights is one of the most
impactful ways HR teams can use data and AI. So let’s delve into this topic
in more detail and see how the HR function can deploy data and AI tools to
improve decision making.

Notes
1 M McNeill. How 5 successful companies are using HR analytics, ICS Learn, 17
May 2020. www.icslearn.co.uk/blog/human-resources/how-5-successful-
companies-are-using-hr-analytics (archived at https://perma.cc/6LNJ-799Y)
76 Laying the groundwork for data-driven, AI-enabled HR

2 Business Insurance. Walmart using AI tools to help staff find right medical
provider, Business Insurance, 1 February 2022. www.businessinsurance.com/
article/20220201/STORY/912347597/Walmart-using-AI-tools-to-help-staff-find-
right-medical-provider (archived at https://perma.cc/GG5P-ZB3M)
3 A M Klahre. 3 ways Johnson & Johnson is taking talent acquisition to the next
level, Johnson & Johnson, 29 August 2017. www.jnj.com/innovation/3-ways-
johnson-and-johnson-is-taking-talent-acquisition-to-the-next-level (archived at
https://perma.cc/NRW2-4PQ6)
 77

PART TWO
Data-driven and
AI-enabled HR
in ­practice
Now that we’ve laid the groundwork for intelligent HR, let’s look at practi-
cal uses of data and AI. Each chapter in Part Two will focus on a specific
area of HR, setting out ways in which data and AI can optimize HR pro-
cesses, drive performance, and contribute to the organization’s success.
Throughout, we’ll explore real-world examples of how organizations have
deployed data and AI to great success.
78

THIS PAGE IS INTENTIONALLY LEFT BLANK

 79

Better HR 06
insights and
decision making
As we saw in Chapter 5, delivering people-related insights and aiding deci-
sion making are two of HR’s core functions. With data and AI-enabled tools,
HR teams can generate better insights than ever before and improve people-
related decision making across the organization. By analysing large datasets
and applying advanced AI algorithms, HR teams can identify trends and
patterns in key HR metrics such as turnover, employee engagement and
performance – allowing them to develop more informed and data-driven
strategies for recruitment, retention, performance and development. The
work of HR becomes more strategic and forward-thinking – to the extent
that you can even create early warning systems capable of identifying poten-
tial workforce issues before they become critical. Such early warning ­systems
can enable you to take proactive action to mitigate risks and optimize work-
force performance.
By leveraging data and AI in this way, HR teams can become more agile,
responsive and strategic, while at the same time creating a more human-
centric approach to workforce management. And as an added bonus, these
tools can also make compliance reporting much less cumbersome.
This all comes under the topic of HR analytics. Needless to say I could
write a whole book on HR analytics alone, so consider this chapter an
­overview of how HR analytics can give your organization better insights.

The role of HR analytics

HR analytics is, as the name suggests, the process of gathering and analysing
HR data to provide useful insights, improve decision making and boost
workplace performance. You might also be familiar with the terms people
analytics, talent analytics and workforce analytics – it’s all the same thing.
80 Data-driven and AI-enabled HR in practice

What’s involved?
HR analytics breaks down into a few different steps. In the first instance the
data is collected (and this can be internal or external data). Then the data is
analysed using statistical models and machine learning techniques to iden-
tify patterns and trends (this may involve comparing the data against other
data, such as benchmarks or historical information). And then the insights
gained from that process are used to inform decision making.
It’s really important that this whole process builds on your HR data strat-
egy (Chapter 5) – by which I mean the data that you collect and analyse
must help the HR function achieve its core goals and contribute to the or-
ganization’s strategic objectives. Of course, HR analytics also provides a
way for the HR team to measure and demonstrate how it contributes to the
organization’s performance. Win–win.

Why do we need HR analytics?

Because it’s how we turn data into insights. And as we’ve already seen in this
book, data in and of itself isn’t enough to enable key decision makers to take
action. You need HR analytics to pull out the meaning within the data. You
need HR analytics to identify trends and patterns within your data – which
can, in turn, help you predict upcoming issues. And you need HR analytics
to answer your most pressing questions. Questions such as ‘Which of our
employees are most likely to leave in the next 12 months?’, ‘What are the
best indicators of success in candidates?’ and ‘How does our development
programme impact employee satisfaction and performance?’ HR analytics
takes the guesswork out of answering these questions so that you can decide
on the best way forward.
In short, HR analytics enables more accurate decision making, both
within the HR function and for managers and leaders across the business.
That said, HR analytics isn’t without its pitfalls. You need good-quality data
(data that is free from bias, etc. – see Chapter 4). You may have multiple
existing management and reporting systems to deal with, which can make it
difficult to pull data together (more on technology and infrastructure in
Chapter 14). You need to invest in the AI tools to analyse that data, whether
that means creating in-house analytics systems or purchasing off-the-shelf
HR analytics tools. And you need to ensure that these technologies are used
ethically, transparently and with the proper oversight (again, see Chapter 4).
These are not insignificant issues but, as we’ll see in this chapter, HR analyt-
ics can deliver huge rewards.
 Better HR insights and decision making 81

HR analytics in action
Take German energy company E.ON as an example. When employee absen-
teeism rose above benchmark, E.ON’s HR team deployed HR analytics to
figure out which factors were driving unscheduled time off. One theory was
that employees selling holiday days back to the company increased absentee-
ism, but it turned out that wasn’t the case. Instead, analytics uncovered that
the duration and timing of vacation had the biggest impact on absenteeism –
as in, people who took multiple days off across the year and had at least one
longer vacation took less unscheduled time off.1 As a result, employees were
encouraged to take frequent short breaks and at least one longer break across
the year – and managers were encouraged to be more accommodating when
it came to approving time off. If the HR team had simply acted on gut in-
stinct and stopped the practice of employees selling back holiday – which
was not proven to have any statistical impact on ­absenteeism – it likely would
have been an unpopular and ineffective ­decision.

The evolution of HR analytics

In Chapter 2 we saw how the work of HR has evolved, and how data has
come to play a more vital role in HR. Now let’s explore how HR analytics
has evolved significantly over the past few decades. Driven by new technolo-
gies, data availability and increasing recognition of the strategic value of HR
in organizations, HR analytics has evolved from basic data reporting and
descriptive analysis to more sophisticated analytics, incorporating AI and
machine learning technologies.

From the pre-analytics era to descriptive

and diagnostic analytics
Before the advent of HR analytics, HR decision making was largely based
on a mixture of gut feeling and experience. In terms of data, HR profession-
als relied on traditional sources like annual performance reviews, exit inter-
views and employee surveys to inform their decisions.
As HR analytics began to evolve, systems largely focused on descriptive
analytics, which involved gathering and presenting data on employee demo-
graphics, retention rates, turnover and other fundamental HR metrics. This
stage was marked by the emergence of HR information systems and
82 Data-driven and AI-enabled HR in practice

­ anagement systems that helped organizations store and manage employee

m
data more effectively.
Then, as HR analytics matured, diagnostic analytics emerged, enabling
HR professionals to identify patterns and relationships in their data. This
phase involved the use of data visualization tools and basic statistical analy-
sis techniques to explore the causes of specific HR issues, such as high turn-
over or low employee engagement.

The move towards predictive and prescriptive analytics

With advancements in technology and the increasing availability of data,
HR analytics has increasingly shifted towards predictive analytics. This in-
volves applying advanced statistical models and machine learning ­techniques
to historical data in order to forecast future outcomes (such as employee
turnover, performance, recruitment success, and so on). This leap forward in
analytics has allowed organizations to make more informed decisions by
anticipating potential challenges and identifying opportunities for improve-
ment. In other words, the emphasis has shifted somewhat from solving
­current and past problems to predicting future problems and solving them
before they occur.
The most recent stage in the evolution of HR analytics is prescriptive
analytics, which goes beyond simply predicting outcomes to recommending
specific actions. By incorporating optimization algorithms and decision
analysis, prescriptive analytics enables HR professionals to identify the best
course of action for a given situation, such as talent development, workforce
planning or employee engagement initiatives.
In one example, Microsoft used HR analytics to develop statistical pro-
files of employees who are most at risk of leaving their jobs. Based on these
insights, the HR team could enact a variety of interventions, such as discuss-
ing learning opportunities or assigning a mentor. With this approach,
Microsoft’s HR team was able to cut attrition rates in high turnover parts of
the business by more than half.2
Clearly, the integration of AI into HR analytics has significantly expanded
its capabilities. Advanced algorithms can analyse vast amounts of data, un-
cover hidden patterns and generate insights that were previously unattain-
able. As a result, HR professionals and executives can make more informed
decisions, identify trends, and develop more tailored interventions for indi-
vidual employees or teams.
 Better HR insights and decision making 83

So, what can be measured and analysed

with HR analytics?
Pretty much anything to do with the workforce. Essentially, you’re looking
for metrics that help you understand what’s working well at the moment,
where you might have room for improvement, and what trends and issues
may be coming up on the horizon.
The following aren’t exhaustive lists by any means, but they indicate
some of the metrics that are commonly used in HR analytics. Which metrics
are right for you? Largely this will be informed by your HR data strategy
(Chapter 5). But you will also need to talk to decision makers across the
organization to determine which metrics they need most in order to make
better decisions. We’ll talk more about getting information to decision
­makers later in the chapter.

Example recruitment metrics

●● cost per hire
●● time to hire
●● acceptance rate
●● new hire turnover (typically, employees who leave within the first year)
●● time to productivity

Using metrics like these you can, for example, understand how long it takes
to hire for specific roles, which in turn allows managers to better plan their
hiring. And with advanced tools, it’s even possible to scan your organiza-
tion’s entire employment history to determine the key attributes for a par-
ticular position. That’s the idea behind Watson Recruitment, an AI tool by
IBM. Watson can also be used to scan applicants to find the most appropri-
ate candidates based on desired attributes.3
How are companies using recruitment metrics in practice? One example
comes from media information company Nielsen. Using HR analytics, the
team determined that the first year of employment played a critical role in
how likely an employee was to stick around. Based on this knowledge,
Nielsen implemented a system to ensure that critical contact points for first-
year employees were met. For example, if the first check-in with their man-
ager didn’t happen within a set timeframe – an important condition for new
hire retention – it triggered a notification.
84 Data-driven and AI-enabled HR in practice

Elsewhere, Johnson & Johnson used HR analytics to analyse the impact

of industry experience on new hire retention. Previously, recruiters had as-
sumed that those with industry experience would be more likely to stay with
the company and be quicker to contribute to company performance, as op-
posed to people who were fresh out of college. But the data actually revealed
that employees hired straight out of college remained with the organization
significantly longer than more experienced candidates – and there was no
real difference in how quickly they contributed to performance. As a result,
the company increased hires of recent graduates by 20 per cent.

Example performance metrics

●● employee performance
●● performance and potential
●● goal tracking
●● capability analytics (identifying core competencies in the workforce)
●● capacity analytics (operational efficiency of employees)
●● leadership performance

With metrics like these, you can better grasp employee, team and organiza-
tional performance – identifying, for example, which employees are star
performers and which attributes top performers have in common. This
knowledge can inform future hiring decisions, as well.
In one example, a mining company in Zimbabwe was concerned about
losing money as a result of understaffed or overstaffed departments. To
identify optimum staffing levels, they measured the number of employees in
a business unit and compared that to the business unit’s activity over 17
quarters. They found a strong relationship between business activity and
number of employees, and this helped them identify which departments
were understaffed and which were overstaffed. Employees from overstaffed
units were either retrenched or relocated to understaffed departments, and
within three months the company was saving money.4

Example employee retention and satisfaction metrics

●● employee engagement score (for example, using the employee net
promoter score)
●● retention rate
 Better HR insights and decision making 85

●● turnover rate/employee churn

●● absenteeism

Using metrics like these, you can identify trends in retention, predict which
employees might be at risk of leaving, and develop strategies to better en-
gage with those employees.
You can also explore the relationship between engagement and perfor-
mance, which is what shoe retailer Clarks did. By analysing 450 data points,
the company was able to confirm that high engagement does indeed lead to
higher business performance, with each 1 per cent of improvement in engage-
ment delivering a 0.4 per cent improvement in business performance. The
team also analysed the characteristics of the 100 top-performing stores and
one of the things they found was that a longer tenure of the store manager
was a significant predictor of store performance. So switching store manag-
ers on a regular basis was a no-no for team engagement and performance.

Example training and development metrics

●● cost of training per employee
●● training effectiveness
●● training completion rate
●● time to completion
●● time since last promotion

Using training and development metrics, you can better understand how
employees engage with your training offering, identify areas for improve-
ment and find out which employees may be overdue for development. In one
example, International Scholarship and Tuition Services used HR analytics
to better spot potential development opportunities, which meant they were
able to ensure the right employees got on track for leadership roles.

Using HR analytics to inform decision

making: Reporting and dashboards
Much of the time, HR analytics will be relevant to, well, the HR team! But
that’s not exclusively the case. Because decision makers across the business
will also need access to the insights you uncover. In this final part of the
86 Data-driven and AI-enabled HR in practice

chapter, let’s dwell on the process of turning insights into decisions and
­actions – both within HR and across the business.

Reporting to others in the business

For me, intelligent, data-driven HR is about HR becoming the strategic
partner for the organization – by which I mean providing leaders and man-
agers with key insights about people, developing early warning systems, and
generally preparing the organization for the future. Because the nature of
work is changing fast, and what it takes to gain competitive advantage is
fast evolving. The HR function will play a vital role in making sure the or-
ganization is in the best position for success (a great example being building
future skills and identifying skills gaps – see Chapters 10 and 13).
But to achieve this vision of HR as the company’s strategic partner, you
need to be able to communicate insights to those who need them most. The
good news is that most HR analytics tools make reporting easier than ever,
but here are a few overarching tips on reporting insights:

Reporting insights

●● Identify the audience for each metric: Some metrics will be exclusively for
the HR team, some may be relevant to middle managers, while others
are needed by the executive team. By identifying the right audience, you
can avoid reporting metrics to people who really don’t need them and
ensure reports are always relevant. With this in mind…
●● Ask decision makers what information they need: What are their reporting
needs? What information are they currently not getting? What
information are they getting that they currently don’t use? Which leads
me to…
●● Revisit your HR analytics frequently to ensure they’re actually useful: For
example, if you commonly report on data that it later turns out isn’t
helpful for decision makers (in HR and elsewhere), then you may want to
adjust your reporting accordingly. Remember, data should lead to
insights which lead to action. If data doesn’t lead to action, do you really
need to keep measuring that metric?
 Better HR insights and decision making 87

Of course, sometimes you may need to deploy HR analytics for more ad hoc
projects, as opposed to regular reporting. Say your organization wants to
open an office in a new location but isn’t sure of the right location. In that
case, the HR team can contribute significant value to the decision-making
process.
That’s exactly what happened at technology multinational Cisco. When
the company wanted to open a new regional office in California, they turned
to HR analytics to pinpoint the best location and even the right building.
The goal was to identify the best spot according to a number of factors, but
especially the ability to attract talent in that location. The People Planning
Analytics and Tools team sourced data such as office usage rates and costs,
and availability of talent (specifically from neighbouring universities).
Interestingly, the data ended up pointing Cisco leaders to a different location
than one they originally had in mind, because the data indicated it would
have been more difficult to attract talent in that area. Now, Cisco’s process
for selecting and opening new offices routinely involves people analytics.

Using HR dashboards
HR dashboards are a great way to analyse and communicate data, and aid
decision making. Think of the dashboard as a tool that pulls all the HR
metrics together in one place, summarizes information, identifies insights
and turns those insights into easy-to-digest reports and visualizations.
Basically, the HR dashboard is a decision-making tool both for HR and the
wider organization.
There are many different types of HR dashboards available – many of
them cloud-based and incorporating AI techniques, with some that special-
ize in specific areas (such as recruitment or performance). They’re typically
very user-friendly, highly customizable, and regularly updated with new fea-
tures. Well-known examples include BambooHR, Personio, QuestionPro
and Bob.

Key takeaways

To recap the main points on HR analytics:

●● With HR analytics, HR teams can generate better insights than ever

before, which enables the organization to make more accurate people-
related decisions.
88 Data-driven and AI-enabled HR in practice

●● As HR analytics has evolved from descriptive analytics to predictive and

prescriptive analytics, this has allowed the work of HR to become more
strategic and forward-thinking – to the extent that you can even create
early warning systems capable of identifying potential workforce issues
before they become critical.
●● Pretty much anything people-related can be measured with HR
analytics, including recruitment and retention, performance, training and
development, employee engagement and more.
●● The right metrics for you will largely depend on your HR data strategy.
But, essentially, you’re looking for metrics that help you understand
what’s working well, where you might have room for improvement and
what trends and issues may be coming up on the horizon.
●● HR analytics platforms make reporting to decision makers (within HR and
across the organization) easier than ever. Remember, data-driven HR is
about HR becoming the strategic partner for the organization – by
providing leaders and managers with key insights about people,
developing early warning systems and generally preparing the
organization for the future. HR dashboards and other reporting tools will
help you get the information to those who need it most.

It’s clear that data and AI can drastically improve the decision-making pro-
cess in HR activities, including recruitment. But how else can data and AI
enhance the recruitment and selection process? Turn to the next chapter to
find out.

Notes
1 M Ankum. People analytics: 5 real case studies, Effectory, 4 March 2022. www.
effectory.com/knowledge/people-analytics-5-real-case-studies (archived at
https://perma.cc/67KP-T4XK)
2 M McNeill. How 5 successful companies are using HR analytics, ICS Learn.
17 May 2020. www.icslearn.co.uk/blog/human-resources/how-5-successful-
companies-are-using-hr-analytics (archived at https://perma.cc/XJY6-8PAN)
 Better HR insights and decision making 89

3 HR 360. Here’s how IBM is using artificial intelligence to help boost their
human resource capabilities, HR 360, 2021. hr360.wbresearch.com/blog/
ibm-watson-ai-for-hr-recruit-reskill (archived at https://perma.cc/6V7N-DZGF)
4 T Fica. The 29 most important HR metrics you need to track examples of how
companies use HR analytics, BambooHR, 2 February 2023. www.bamboohr.
com/blog/key-hr-metrics#examples-of-how-companies-use-hr-analytics (archived
at https://perma.cc/3WFZ-78YE)
90

Recruitment 07
and candidate
selection
AI and data are increasingly being used in recruitment and selection to op-
timize the hiring process – and I firmly believe that those HR teams who
embrace data and AI are the ones who will recruit most successfully in the
coming years. Why exactly? Because AI algorithms can identify patterns in
large datasets and extract valuable insights that help HR professionals to
better match job candidates with the requirements of the role, improve the
candidate experience and reduce bias in the hiring process. Plus, as we’ll see
in this chapter, automation is helping to streamline intelligent recruitment
by automating aspects of candidate assessment and selection.
In this chapter we’ll explore some of the main ways in which data and AI
can help to enhance your recruitment activities. Broadly speaking, they are:
boosting your employer brand; identifying the best recruitment channels;
identifying and assessing the most suitable candidates; and streamlining
­recruitment with AI-enhanced automation tools.

Boosting your employer brand

As any marketing colleague will tell you, branding is absolutely vital when
it comes to attracting and retaining customers (in this case, employees).
What’s more, your employer brand should ideally align with your overall
company, service or product brand. If the two are sending different mes-
sages – ethical manufacturer with questionable employee ethics, for exam-
ple – you could struggle to attract the best talent. One paper by Randstad
RiseSmart stated that nearly 70 per cent of unemployed job seekers would
not take a job with an employer who had an iffy reputation.1 On the flip
side, the same paper stated that 84 per cent of employees would consider
 Recruitment and candidate selection 91

jacking in their current job to move to an employer with an excellent reputa-

tion, even if the salary increase was less than 10 per cent.

Using data to monitor your brand

Your employer brand tells employees and potential employees who you are
as a company, what you stand for, what it’s like to be part of the company
and what makes you different from other employers. When developing your
employer brand, you’ll want to consider what kind of talent you want to
attract; or, to put it another way, what kind of people best fit with your
company’s culture and goals.
Having identified this, you need to be sure this brand image chimes with
reality. And this is where data and AI algorithms come into play. You should
look to test your employment brand at regular intervals. Sentiment analysis
of interview and survey responses, as well social media posts, can really help
establish how successful your employer brand is. And if your company goes
through major changes, such as a restructuring, you should absolutely look
to assess the impact of this change on your brand. Measuring sentiment
before, during and after the changes will give vital insights that will help you
manage the transition and maintain a positive employer brand.
No doubt you’ve heard of the employer net promoter score. Indeed, these
days, ‘How likely are you to recommend the company?’ is a common ques-
tion on employee surveys. However, rather than take the temperature of this
once a year, many companies are using anonymous ‘pulse’ surveys, asking
just this one simple question on a weekly, monthly or quarterly basis.
Creating and maintaining a positive employer brand is not just about
keeping your current employees happy; it’s also about how attractive your
company appears to outsiders. What factors do you think attract potential
employees to your company or make an employee more likely to recom-
mend you as an employer? Salary? Flexible working? No and no, according
to data. Josh Bersin’s company analysed Glassdoor data from more than
6,000 companies and 2.2 million employees to get some interesting findings
on employer brand.2 It turned out the biggest factor in whether employees
would recommend their company as a place to work was ‘culture and val-
ues’. In fact, an employee’s rating on culture and values is almost five times
more predictive of a company being recommended than salary and benefits.
Among those under the age of 35, ‘career opportunities’ was the top driver
of employer brand. This demonstrates the need for employers to consider
culture and employee development as much as, or even more than, the usual
salary and benefits.
92 Data-driven and AI-enabled HR in practice

I’m not saying decent salaries and benefits don’t impact employer brand,
but it’s clear that people want to feel truly at home in a company. They want
to enjoy the culture, they want to be proud of the company they work for,
and they want their input to be valued through good development opportu-
nities. These are all elements to focus on when it comes to boosting your
employer brand and promoting that brand outside the organization.

Immersing potential employees in your employer

brand
Another key element of branding is of course raising the profile of your em-
ployer brand among potential employees. This is especially worthwhile if
you’re in a highly competitive industry, like the tech world, where you’re
struggling to attract talent. Some innovative organizations are using data-
related technology like virtual reality to capitalize on this idea, delivering
360-degree videos that show people what it’s really like to work at the com-
pany. This helps employers show off their culture, give an authentic feel for
everyday life within the business and attract the best candidates.
One of my favourite examples comes from the unlikely world of US col-
lege football. The University of Minnesota’s Golden Gophers team created
a slick VR experience to help sign up in-demand players. Described as a
‘day in the life of a Golden Gopher’, the VR experience serves to immerse
candidates and tell a compelling story about what it’s really like to be part
of the team – with footage of practices, games, workouts, those all-impor-
tant UM campus experiences, city life and even the Minnesota weather. In
the competitive world of college football it can be hard to stand out and
convey the benefits of signing with a smaller team. VR is a great way to do
this authentically.
Taking the idea of immersive VR experiences to the next level, some or-
ganizations are now using the metaverse to show off their employer brand
and attract candidates. French retail group Carrefour, for example, held a
virtual recruitment session in the metaverse.3 For a retail company with
huge digital ambitions, the event appeared to be designed to attract those
with digital skills and to promote Carrefour as an innovative employer – in
other words, to help the company stand out in a field where competition for
digital skills is intense. For many organizations, showing up in the metaverse
will be a key way to attract talent and demonstrate their appeal as an em-
ployer. From a marketing perspective – which is, after all, what employer
branding is all about – the metaverse makes a lot of sense. But we’ll talk
more about the metaverse later in the chapter.
 Recruitment and candidate selection 93

Identifying the most effective recruitment

channels for you
Most companies use a mix of recruitment channels, including headhunters,
job forums, social media campaigns and LinkedIn searches. Different chan-
nels work better for different industries, or even different positions within
the same company. Given the diversity of recruitment channels, it’s impor-
tant to know which deliver the greatest return on investment. In my work
with clients, I’d estimate that maybe 50 per cent of recruitment spend is
wasted. If a channel isn’t driving actual recruitment, you should stop recruit-
ing through that channel immediately and focus on those channels that de-
liver the most value for your spend.

Understanding more about your recruitment channels

The beauty of data is that it allows you to test your recruitment channels
and measure their success rate. These days, it’s possible to measure every-
thing minutely. So, rather than focus on obvious indicators like how many
resumes you get in response from different channels (which only tells you
volume, not quality), you could look instead at more valuable indicators like
how many offers were made to candidates from particular channels. You
could even take this further and assess your most successful employees in
particular roles and pinpoint which channels they came from. Many of to-
day’s HR analytics platforms (see Chapter 6) can help you assess this and
more, so that you can target your recruitment appropriately.
A good example of this comes from Marriott Hotels. The hospitality
giant has a popular recruitment page on Facebook with, at the time of writ-
ing, more than 1.3 million likes and high volumes of people engaging with
the page on a daily and weekly basis. The page obviously lists available jobs,
but it also beautifully demonstrates what it’s like to work for the chain
through photos and videos of life behind the scenes in the hotels. It makes
sense that a hospitality business like Marriott wants to attract the classic
‘people person’, which explains why social media is such a useful recruit-
ment channel for Marriott. It’s an excellent lesson for any company looking
to maximize their employment channels – go with the channel most used by
the type of people you want to attract. To build on their Facebook success,
Marriott even created a Facebook game called MyMarriottHotel, where
potential employees could learn how to manage hotels.
94 Data-driven and AI-enabled HR in practice

Experimenting with not-so-obvious recruitment

channels
Sometimes, though, you may need to think a little further outside the
­recruitment box and look for entirely new channels to recruit talent – par-
ticularly in areas where there’s a lot of competition to hire the best talent.
Data scientists are one such group that are in great demand at all kinds of
businesses, yet there just aren’t enough people with the required skills. To
overcome this problem, Walmart decided to get creative. Rather than adver-
tise through traditional channels, they turned to crowdsourced analytics
competition platform Kaggle to find the talent they needed. At Kaggle, an
army of ‘armchair data scientists’ apply their skills to analytical problems
submitted by companies, with the designer of the best solution being
­rewarded – in this case with a job at Walmart.
Candidates were provided with a set of historical sales data from a sam-
ple of stores, along with associated sales events, such as clearance sales and
price rollbacks. They were asked to come up with models showing how
these events would affect sales across a number of departments. As a result
of the competition, several people were hired into the analytics team.4
Interestingly, the crowdsourced approach led to some appointments of peo-
ple who, ordinarily, wouldn’t have been considered for an interview based
on their resumes alone. One candidate, for example, had a very strong
­background in physics but no formal analytics background.
While you may not be looking to recruit an army of data scientists like
Walmart, this example shows why it’s important to consider some of the
more unusual recruitment channels, find out where your talent hangs out
and use that knowledge to focus on recruitment channels that deliver the
most bang for your buck.

Identifying and assessing the best

candidates for your business
I’ve often heard that recruiters and hiring managers make up their minds
whether or not to hire a prospective candidate within five minutes of sitting
down with them. It’s hard to say whether this is true or not, but many HR
professionals or hiring managers would probably admit that they had made
appointments based on a gut feeling – simply whether or not they felt the
person was the right fit for the role.
 Recruitment and candidate selection 95

As it is in many other areas of business, data and analytics is helping to

take the guesswork out of recruitment. Rather than relying on the famous
gut feeling, those teams taking a data- and AI-enabled approach are finding
it leads to more suitable hires who stay happy and on the job for longer. In
my opinion, this is where the future of recruitment lies.

Predicting a candidate’s suitability

When recruiting a new candidate, you’re not just looking at a list of skills
and experience on paper. Personality and fit are just as important.
Traditionally, these factors have been considered difficult to judge, but not
with the help of data and predictive analytics…
We already know that companies like Facebook and Google can predict
an awful lot about our intelligence, behaviour and personality attributes
based on our profiles and online activities. A study by researchers at
Cambridge University and Microsoft Research Labs showed how the pat-
terns of Facebook likes can be used to automatically predict a range of
highly sensitive personal attributes.5 For example, likes for curly fries,
­science, Mozart, thunderstorms or The Daily Show predicted high intelli-
gence, while likes for Harley Davidson, Lady Antebellum, and ‘I love being
a mom’ predicted low intelligence. The study was even able to predict
­sexuality and religion with extraordinary accuracy.
Creepy? A little. But the same sort of predictive capabilities can be put to
good use when it comes to identifying the right employees. It sounds obvi-
ous, but the first step is to identify exactly what you’re looking for. JetBlue
Airlines gives us a great example of this. Originally, the company had
­focused on ‘niceness’ as the most important attribute for its flight attend-
ants. Then, after carrying out some customer data analysis with the Wharton
Business School, the company was interested to find that, in the eyes of their
customers, being helpful is actually more important than being nice – and
can even make up for people being not so nice. The company was then able
to use this information to narrow down candidates more effectively.6
What does your ideal employee look like, in terms of skills, desirable at-
tributes, personality traits, qualifications, experience and fit? Armed with
this ‘shopping list’, if you like, it’s relatively easy to use AI tools to sift
through potential candidates and identify certain data points in applica-
tions, resumes or profiles and find the candidates with the best fit.
96 Data-driven and AI-enabled HR in practice

While the final hiring decision should always come down to a human, AI
tools can save a lot of time by narrowing the field down from maybe h ­ undreds
of candidates to the most suitable 10 or 20. Clearly, then, this is one area
where AI-driven automation is playing an increasingly important role. We’ll
talk more about automation a little later in the chapter, but suffice to say that
AI tools are helping employers filter out much of the ‘noise’ when it comes to
narrowing down candidates and focusing on what they really want. Crucially,
such automation can also remove the biases that humans inevitably bring to
the recruitment process. But, again, we’ll get to that later.

CASE STUDY Identifying candidates for top-level positions

Data and analytics can be used to identify the best candidates for any position,
even right up to the ‘C-suite’ of executives who guide the direction of the
business. These leaders carry a large amount of responsibility, and in return
often take home a sizeable chunk of a business’s earnings. When mistakes are
made appointing people at this level, it can spell disaster. So, of course, it makes
sense that filling these vacancies should be done with as little guesswork as
possible.
Corporate headhunters Korn Ferry have taken steps to ensure C-level
recruitment is firmly rooted in data and analytics. Thanks to the wealth of data
collected in the company’s long history, they were able to draw up detailed
profiles of the competencies, traits and experiences needed to succeed at the
top level. The data revealed some strong patterns about the important traits and
qualities required for C-level positions, including being a lifelong learner, higher
levels of emotional intelligence and empathy, communication skills and a
tolerance for risk.7
In partnership with data scientists at the University of Southern California,
Korn Ferry built an analytics-based people placement platform called Four
Dimensional Executive Assessment – or KF4D for short – which can assess traits
and competencies, experience and even cultural fit. When it comes to the latter,
clients looking to fill positions have the ability to choose whether they’re happy
with their organizational culture, in which case the system will find someone
who is likely to fit in. Or, if they’re looking to change their culture, candidates will
be suggested who are likely to be agents of change.
 Recruitment and candidate selection 97

Appointing a new CEO is undoubtedly one of the biggest challenges a busi-

ness will ever face. Most companies wouldn’t make decisions about which
products or services to offer without solid data analysis, so it makes sense to
bring the same analytical approach to hiring for top-level positions.

Sourcing freelancers and remote workers

In this connected world, it’s now increasingly common for companies to
draw upon remote talent without making traditional hires. As more and
more businesses head for the cloud, it’s becoming much easier for companies
to tap into gig economy talent. The obvious advantage for workers is they
can work whenever and wherever they like. And for employers, they can
benefit from talent without the expense of hiring people full time.
And even when companies do want to hire full-time talent, more and
more are letting employees work remotely – particularly since the pandemic.
But what does this mean for recruitment? Well, the same sorts of analytic
techniques we’ve talked about in this chapter can be incredibly useful when
identifying and assessing candidates who may never set foot in your offices.
Also, when someone works remotely, they may not have access to the same
level of one-to-one mentorship that an in-house employee may have. This
may mean you need to focus your hiring on highly experienced people who
already have all the core attributes needed to succeed in the role. Data and
analytics can help you pinpoint such candidates quickly and easily – and
ensure you onboard remote workers successfully (more about onboarding
in Chapter 8).

Identifying and promoting suitable candidates inside

the company
As a final word on identifying candidates, many employers see it as more
cost-effective to promote from within than to recruit from outside the com-
pany. Another obvious advantage of internal candidates is they’re already
well versed in the company’s systems, processes and culture. So it makes
sense to apply the same sort of talent analytics to identifying suitable candi-
dates from within the company. We already know that the opportunity to
progress and grow within a company is hugely attractive to employees, so
promoting from within is another great way to boost your employer brand.
98 Data-driven and AI-enabled HR in practice

As we saw in Chapter 6, people analytics can help HR teams identify the top
performers in roles across the company, as well as those who are already in
the best position for them.

AI-enhanced automation in recruitment

As we’ve already seen in this chapter, AI and data are increasingly being
used by HR functions to automate and, in turn, streamline parts of the re-
cruitment and selection process. So let’s dwell on this notion a little deeper,
and see what can be automated – and where companies may fall foul of
automation.

What can be automated?

When it comes to recruitment, how do you, an HR professional, add the
most value for the organization and people you serve? I’d bet your answers
include making decisions based on your expertise and perspective, identify-
ing and plugging skills gaps, finding the very best talent, convincing sought-
after candidates to come work for your employer, improving your
organization’s employer brand and generally providing an awesome service
to internal and external candidates. And yet how much of your (or your
team members’) time is really spent on those mission-critical tasks? How
much of your team’s time is spent on mechanical, easily automated tasks like
scheduling interviews, answering basic questions or sifting through the
many ­irrelevant applications that come your way?
This is where automation can provide massive advantages for the average
HR team – taking care of the easily repeatable tasks and freeing up recruit-
ers’ time for more value-adding activities. In particular, the combination of
robotic process automation – software that automates business processes
that are rules-based, structured and repetitive – with machine learning is
where a lot of value lies. The robotic process automation takes care of re-
petitive tasks, while the AI element can enhance and speed up decision
­making.
There are many recruitment automation platforms out there that com-
bine robotic process automation and AI elements, often referred to as
 Recruitment and candidate selection 99

a­ pplicant tracking systems. Here are some of the recruitment and selection
tasks that they promise to automate:

●● You can automate the creation and posting of job adverts, according to
templates that you set. That way, you don’t have to reformat or rewrite
the same posting for different sites and manually upload them to dozens
or even hundreds of boards – the system will automatically populate
multiple boards, based on your job description and typical ad template.
Some tools are even capable of identifying the target audience for specific
job posts and automatically buying ad space (within a defined budget)
that’s optimal for that audience.
●● You can automatically pre-screen candidates according to must-have
credentials, from professional certifications to the legal right to live and
work in your part of the world. This ensures recruiters don’t waste time
on non-eligible candidates. But you can also take pre-screening further, by
adding questions to the application process that can be scored and ranked
according to desirability. This allows your recruiters to hone in on the best
(on paper) candidates first, potentially speeding up your time to hire.
●● You can also automate some candidate assessments, such as cognitive
testing or personality questionnaires. The system will then assess and
rank the candidates. In one example, IBM used AI to identify the best
candidates to join a sales team based on their social skills – with the
system even predicting the impact those people would have on the sales
team’s performance. As new team members are added to the team, the AI
tool continually re-assesses the skills needed with the aim of forming the
optimum sales team.8
●● You can even have the AI system automatically shortlist and contact the
best candidates, thereby quickly securing them an interview – and
automatically reject the least suitable candidates according to your pre-
set criteria, thereby giving those candidates a fast resolution. However,
automatically selecting and rejecting candidates without human oversight
should be used with caution – see later in the chapter.
●● You can also automate the sourcing of resumes and candidates. Basically,
you create a list of requirements, and then the system will scour platforms
looking for the most suitable profiles. This can extend to social media
platforms as well – meaning the system can identify suitable candidates
and send them tailored messages.
100 Data-driven and AI-enabled HR in practice

●● You can use AI-powered chatbots to automate communication throughout

the recruitment process – for example, answering candidates’ common
questions or sending them a progress update. This saves recruiters sending
the same messages over and over again, and ensures consistency. And for
the candidate, it helps to ensure they get quick responses and remain in
the loop.
●● You may also be able to automate due diligence checks, such as reference
requests and background checks.
●● You can even automate interview scheduling, which can be especially
helpful when you’re recruiting high volumes of candidates. Recruiters
and hiring managers define their availability and then the candidate can
choose the available slot that best suits them, so there’s no back-and-
forth between various parties.
●● You can automate first-round interviews with one-way video interviewing,
where the recruiter sets the questions in advance and then the candidates
record their responses. Hilton has used digital interviewing platform
HireVue to conduct one-way video interviews (for first-stage interviews)
with multiple candidates at once. This helped the company quickly assess
which candidates were the best fit, and which should therefore progress
to the next interview stage – and in doing so, Hilton was able to reduce
its time to hire from 43 days down to an impressive five days.9
●● You can automate internal recruitment workflows – for example,
prompting hiring managers to review a shortlist of candidates, or give
their feedback for candidates. This can save masses of time for busy
recruiters.
●● You can also use predictive analytics to forecast workforce needs and
identify potential skill gaps in your organization. This allows you to
proactively develop talent pipelines and recruitment strategies.
●● And don’t forget, you can automate recruitment reporting. (Circle back
to Chapter 6 on HR analytics.)

Clearly, automating tasks like these can help to free up recruiter time, im-
prove recruiter productivity (by up to 50 per cent, according to recruitment
automation specialists Entelo),10 speed up time to hire (Entelo says by up to
15 per cent), reduce organizational costs (because positions are vacant for
less time) and provide a slicker experience for candidates.
 Recruitment and candidate selection 101

Automation in action

CASE STUDY Johnson & Johnson

We’ve already seen a few quick examples of automation in action, but let’s
explore how Johnson & Johnson has automated one part of the recruitment
process with great success – in this case, using Google’s Cloud Talent Solution
to better match people visiting J&J’s career site with suitable job listings. This
had been a problem for J&J in the past. While the company receives around one
million applications for 25,000 jobs each year, the proportion of applicants that
were highly qualified for those positions was low. Recruiters noticed that even
when visitors to the site had a strong match for certain positions, those
opportunities weren’t displayed in search results. Candidates couldn’t easily find
the right job opportunities, and the system wasn’t working. This is where Cloud
Talent Solution comes in. By understanding the nuances of job seeker search
queries and job postings, the system uses machine learning to help candidates
find the jobs most relevant to them. This meant J&J recruiters didn’t have to
manually optimize job postings for search results, and job seekers could find jobs
on the J&J site more easily. As a result, J&J has seen a 41 per cent increase in
high-quality applicants per search, and a 45 per cent increase in click-through
rate on its job site.11

For those recruiters concerned that they may find themselves out of a job
thanks to automation software, I wouldn’t lose too much sleep. We still
have marketers and sales people, despite marketing and sales automation
software existing for years. But it’s fair to expect that the work of recruiters
will shift and evolve – moving away from repetitive tasks and spending
more time on the tasks that drive the business forward. Which sounds great.
However, it’s vitally important to ensure that these automation technologies
are used ethically and transparently, and that human oversight is main-
tained. If we abdicate all responsibility for hiring decisions to machines, this
can result in unintended consequences and unfair outcomes. As some or-
ganizations have discovered to their detriment…

The downsides of automation

Perhaps the best-known example of automated recruitment gone wrong
comes from Amazon. As we saw in Chapter 4, the company had to stop
102 Data-driven and AI-enabled HR in practice

using a machine learning algorithm for screening candidates because it was

biased against female candidates. So what happened? In essence, the data
the algorithm used to make decisions was flawed because Amazon’s recruit-
ment data from the previous 10 years included more male than female can-
didates (historically, more men had applied for roles). As such, the system
effectively taught itself that male candidates were preferable to female.12
The system was so skewed, it penalized resumes that included the word
‘women’.
Amazon made a couple of big mistakes with its system. For one thing, the
data used was purely internal – and even in a company as big as Amazon,
there clearly wasn’t enough data to ensure diversity and fairness. Synthetic
data would have helped to solve this issue. And secondly, as with any aspect
of business, it’s not always a great idea to base future decisions on what
you’ve done in the past. Just because certain candidates have traditionally
done well in the hiring process doesn’t mean that’s the best outcome for the
business going forward. Remember, Amazon was using recruitment data
from the previous 10 years and a lot can change in 10 years.
Interestingly, at the time of writing, leaked Amazon documents suggest
that the company has again been experimenting with software to screen its
job applicants – and is preparing, reports suggest, to replace many human
recruiters with software. The software can apparently predict which appli-
cants will be successful across certain jobs and fast-track them to an inter-
view – without the involvement of a human recruiter.13 The leaked documents
suggest that Amazon believes the new model isn’t subject to biases.
Bias in the training data was previously a huge issue for Amazon. But
there are other ways in which bias can creep into automated systems. Having
a training dataset that’s too small is one way. Another is programming the
system to rank candidates according to subjective parameters – for example,
let’s say you determine that having a postgraduate degree is more favourable
than just a graduate degree, and you therefore tell the system to prioritize
those candidates, that’s a subjective decision on your part. By doing what it
is told, the system may overlook brilliant candidates who don’t have a post-
graduate degree. Ranking candidates is a great way to automate recruit-
ment, but you should be careful about the data points you select and ensure
you’re ranking based on a broad range of criteria.
In another example, an algorithm designed to promote science, technol-
ogy, engineering, and mathematics (STEM) jobs online prioritized advertis-
ing to men purely because women are more expensive to advertise to.
Because the system was designed to optimize ad delivery while keeping ad-
vertising costs low, it delivered ads to more men than women!14
 Recruitment and candidate selection 103

Getting automation right

I don’t give these examples to put you off recruitment automation. Far
from it. I believe the benefits of recruitment automation outweigh the
downsides – particularly, and ironically, when it comes to reducing the risk
of bias and discrimination in recruitment decisions. Human bias is one of
the biggest challenges in recruitment. Because, as much as we’re supposed
to judge people objectively, and most of us set out to do just that, the fact
is humans are prone to cognitive bias (however subtle or subconscious it
may be). This can be as irrational as disliking a certain phrase or buzzword
used by a candidate.
Automation has the potential to take our human bias out of the equation
and assess candidates in a truly objective way. But that only works when the
data and algorithms are unbiased. Awareness and oversight are your best
weapons here – meaning it’s really important to be aware of the possible
downsides, carefully assess your data for potential bias, and think hard
about the assessment criteria used by algorithms. You could, for example,
have your algorithms assess candidates purely on a skills basis, and ignore
other factors such as age, gender, name and education. You should also im-
plement checks and balances throughout the process, and find a balance
between human expertise and machine decision making. Basically, you don’t
want to be relying solely on what the machine says.
I would also strongly urge you to ensure any automation is as transparent
as possible. You don’t, for example, want a chatbot to pretend to be a human
HR professional. Always be clear with candidates when they’re dealing with
a machine, how their data is being used, and so on.

Recruitment in the metaverse era

We’ve touched on the metaverse briefly, but since it’s likely to play a larger
role in recruitment in future, let’s close out the chapter with a quick look at
metaverse-based recruitment.
For now, metaverse recruitment is likely to take the shape of virtual re-
cruitment sessions and career fairs, largely aimed at attracting tech-savvy
talent, as per the Carrefour event from earlier in the chapter. Other compa-
nies to have held metaverse career fairs include Capgemini and Samsung –
with the latter enabling job seekers to virtually consult with recruiters (or
rather, recruiters’ avatars) during the event.15
104 Data-driven and AI-enabled HR in practice

Looking a little further ahead, we may see organizations create their own
virtual recruitment centres where candidates attend virtual interviews and
assessments. Indeed, the metaverse could simply be the next evolution of
gamification in recruitment and selection. Instead of candidates playing, for
example, a social media game to assess their suitability for a role, they could
immerse themselves in a virtual assessment experience that replicates what
they’d face in the real-world role. I could see this being of great interest to,
say, hospitality companies and retailers, but it could no doubt prove useful
in a vast range of industries.
Adding an extra, more immersive dimension to recruitment and selection
is one clear advantage of the metaverse. But another advantage is, quite
simply, enhancing your employer brand. For example, if you’re trying to at-
tract innovative, creative, tech-savvy people, they’re likely to appreciate you
taking a similarly innovative, creative, tech-focused approach to your re-
cruitment. It makes sense that recruiting in the metaverse will also prove
especially useful for companies who want to hire freelancers and remote
employees on a large scale. Because, in the metaverse, you can conduct huge
campaigns without geographical borders.
It’s still early days for recruiting in the metaverse, but HR professionals
should definitely watch this space…

Key takeaways

Let’s finish with a quick summary of the main takeaways on intelligent

recruitment and selection:

●● AI algorithms can identify patterns in large datasets and extract valuable

insights that help HR professionals to better match candidates with the
requirements of the role, improve the candidate experience and reduce
bias in the hiring process.
●● The main ways in which data and AI can boost your recruitment
activities include improving your employer brand; identifying the best
recruitment channels; identifying and assessing the most suitable
candidates; and streamlining recruitment with AI-enhanced
automation tools.
●● Automation in particular can provide massive advantages for HR teams
by taking care of the easily repeatable tasks and freeing up recruiters’
 Recruitment and candidate selection 105

time for more value-adding activities. Some of the tasks that can be
automated include pre-screening candidates, assessing candidates (for
example, personality testing), scheduling interviews, sourcing resumes,
posting job ads, and more.
●● Automation also has the potential to take bias out of the equation and
assess candidates in a truly objective way. However, it’s important to
note that bias can creep into algorithms – for example, if the training
dataset is limited. It’s therefore extremely important to be aware of the
possible downsides, carefully assess your data for potential bias and
think hard about assessment criteria.
●● Finally, HR professionals should be on the lookout for opportunities to
recruit in the metaverse – immersive virtual recruitment fairs, for
example, or VR experiences that show what it’s really like to work for
your company. In the future, more talent will be recruited in this way.

Of course, attracting and hiring the best talent is only part of the challenge;
you must also onboard that talent successfully. In the next chapter, we’ll
explore how data and AI is enhancing the onboarding process.

Notes
1 Randstad RiseSmart. The connection between HR analytics and employer
brand, Randstad RiseSmart, nd. info.randstadrisesmart.com/wp-rg-insight-
whitepaper (archived at https://perma.cc/W5MF-Z2L4)
2 J Bersin. Data proves that culture, values and career are biggest drivers of
employment brand, Josh Bersin, 25 July 2016. joshbersin.com/2016/07/
data-proves-that-culture-values-and-career-are-biggest-drivers-of-employment-
brand (archived at https://perma.cc/N4KT-B2YH)
3 P-N Schwab. Recruitment in the metaverse: Examples and perspective ­(analysis),
Into The Minds, 25 May 2022. www.intotheminds.com/blog/en/recruitment-
metaverse (archived at https://perma.cc/WHG9-594D)
4 B Marr. Walmart: The big data skills crisis and recruiting analytics talent,
Forbes, 6 July 2015. www.forbes.com/sites/bernardmarr/2015/07/06/walmart-
the-big-data-skills-crisis-and-recruiting-analytics-talent (archived at https://
perma.cc/5SNW-8PFV)
5 R Baldwin. Study: Facebook likes can be used to determine intelligence,
sexuality, Wired, March 2013. www.wired.com/2013/03/facebook-like-research
(archived at https://perma.cc/WFK3-23G9)
106 Data-driven and AI-enabled HR in practice

6 B Aslan. To all recruiters – use machine learning to hire better candidates,

Medium, 18 June 2016. medium.com/@deadlocked_d/to-all-recruiters-use-
machine-learning-to-hire-better-candidates-c5aad22f3319 (archived at
https://perma.cc/3YJM-YXHF)
7 B Marr. Can big data find your next CEO? Forbes, 27 June 2015. www.forbes.
com/sites/bernardmarr/2015/07/27/can-big-data-find-your-next-ceo (archived
at https://perma.cc/SQK2-DEBC)
8 M Mahto and D Miller. AI for work relationships may be a great untapped
opportunity, Deloitte, 24 October 2022. www2.deloitte.com/uk/en/insights/
industry/public-sector/how-social-ai-is-transforming-the-workplace.html
(archived at https://perma.cc/R4N6-GQN4)
9 H L Kurter. How Hilton reduced their time to hire from 43 days down to
5, Forbes, 19 September 2019. www.forbes.com/sites/heidilynnekurter/
2019/09/19/how-hilton-reduced-their-time-to-hire-from-43-days-down-to-5
(archived at https://perma.cc/R9P4-BZ8Y)
10 Entelo. What is recruiting automation? Entelo, nd. www.entelo.com/recruiting-
automation (archived at https://perma.cc/BEQ5-7CWG)
11 Google Cloud. Johnson & Johnson: Reimagining recruiting with Jibe and
Google, Google Cloud, nd. cloud.google.com/customers/johnson-and-johnson
(archived at https://perma.cc/PPM4-78JK)
12 BBC. Amazon scrapped ‘sexist AI’ tool, BBC News, 10 October 2018. www.
bbc.com/news/technology-45809919 (archived at https://perma.cc/D3JN-
32AR)
13 J Del Ray. A leaked Amazon memo may help explain why the tech giant is
pushing out so many recruiters, Vox, 23 November 2022. www.vox.com/
recode/2022/11/23/23475697/amazon-layoffs-buyouts-recruiters-ai-hiring-
software (archived at https://perma.cc/LB22-J4HL)
14 M Lavanchy. Amazon’s sexist hiring algorithm could still be better than a
human, IMD, November 2018, www.imd.org/research-knowledge/articles/
amazons-sexist-hiring-algorithm-could-still-be-better-than-a-human (archived
at https://perma.cc/2G95-RBZZ)
15 Multiplier. Hiring in an alternate digital universe with metaverse, Multiplier,
nd. www.usemultiplier.com/blog/hiring-in-the-metaverse (archived at
https://perma.cc/T8RA-9HJ5)
 107

Improving 08
employee
onboarding
Onboarding has always been a vital part of the HR function. Indeed, this
process of initiating new employees into the company can make or break the
employee experience.
Data and AI can play a significant role in onboarding new hires, helping
to create a more personalized and engaging experience for new hires – while
at the same time saving time and effort through automation. By using data
and AI, organizations can enhance the onboarding experience for new hires,
increase engagement and retention, and create a more positive impression of
the organization. It’s no wonder that, according to a 2022 report, 68 per
cent of organizations in the US are already using AI in their onboarding
processes.1
But what does this look like in practice? Let’s find out…

Onboarding in the AI era

There are many onboarding software tools out there that promise all sorts
of benefits. But, for me, there are two main ways in which these tools can
enhance the onboarding process: saving time through automation, and pro-
viding a more personalized experience for new hires. It almost seems like a
contradiction, doesn’t it? That data and AI could on the one hand automate
many onboarding processes, and on the other, help to provide a more human
experience for employees. But that’s exactly the case. Let’s explore automa-
tion in the first instance, since this is where many efficiencies lie.

Automating aspects of onboarding

While every employee is unique and valued as an individual, there’s no deny-
ing that onboarding new hires involves a lot of mundane, repetitive tasks
108 Data-driven and AI-enabled HR in practice

that are the same for each and every one. AI can help to reduce the load
these processes place on HR professionals, thereby freeing up their time for
employee interactions that deliver the most value.
Use cases vary from employer to employer, depending on the size of the
organization and the scope of the onboarding programme. But, in general,
we’re talking about automating simple, easily repeatable tasks. (As we saw
in Chapter 7, using software to carry out such tasks is known as robotic
process automation, and this can be combined with AI for maximum effect.)
Here are some of the tasks that can be automated with onboarding
­software:
●● You can automatically gather required data from new employees.
●● For example, sending requests for pay slips from their previous
employment, or proof of identity. (In fact, many of the compliance-
related onboarding tasks can be easily automated.) Automated systems
can not only handle such communications, but also analyse the
resulting data, depending on your needs – for instance, segmenting an
employee’s data according to demographics. And, for employees, the
good news is they only need to provide the required data once, instead
of completing multiple forms.
●● You can automatically send messages to employees at various stages of
their onboarding journey, such as signing the offer letter, sending the
employment contract and in the run-up to their first day.
●● You can use chatbots to automate certain conversations.
●● Before, during and after a person joins your organization, you’ll no
doubt have multiple interactions with them that go beyond sending
automated welcome messages. In particular, many new hires will have
questions they need answering. While not all of these questions can be
managed with chatbots, your frequently asked questions can certainly
be automated.
●● You can use software to automatically create and assign system profiles
for new hires – such as creating user accounts and email addresses, and
assigning access to internal systems.
●● You can automate onboarding checklists.
●● There’s typically a long list of items for new hires to do in their first
days, weeks and months with the company. AI can automate this
checklist process and track the employee’s progress against tasks,
providing reminders and prompts when tasks remain unchecked.
 Improving employee onboarding 109

●● You can automate training workflows.

●● Training will obviously be different for different roles within the
company, but onboarding software can help to manage the process
(for example, by scheduling training or providing reminders). This
ensures each employee gains the essential knowledge they need to
perform their role – and gets it at the right time.
●● Chatbots can be used to informally gather feedback from employees
about the onboarding process.
●● Remember that AI can automatically detect the sentiment in written
responses, giving you quick insights on how employees feel about the
onboarding process, and helping to pinpoint where people may need
extra support.
●● You can also automatically analyse the progress and performance of new
hires. (We’ll talk more about performance management in Chapter 9.)
●● You can even use AI to help new hires boost their network and get to
know helpful colleagues.
●● AI tools can make suggestions for each employee on which personnel
or team leaders they might want to get to know, based on their role
and training needs.
●● All of these tasks apply to onboarding remote hires, as well.
●● AI tools can also help you provide tailored support to remote
employees. For example, AI can be used to automatically translate
onboarding documents and communications into another language.
(What’s more, thanks to the metaverse, you can deliver a highly
immersive virtual onboarding experience. More on the metaverse
coming up later.)
●● And don’t forget that you can also automate reporting on your onboarding
trends and patterns, thanks to HR analytics tools (Chapter 6).

It’s worth noting that many of the above examples can also be applied when
employees are promoted or transitioning to a new role within the company,
when people rejoin the organization after a career break and even when
people leave the business altogether.
If you are able to automate onboarding processes, you stand to reap sev-
eral benefits. The most obvious is decreasing the HR workload and time
spent on mundane tasks – which, in turn, allows HR professionals to focus
their time on more strategic tasks. Automation also speeds up the ­onboarding
110 Data-driven and AI-enabled HR in practice

process, thereby allowing new hires to reach productivity more quickly.

And, most importantly, the onboarding experience is consistent and positive
for all new employees. With templates and automated workflows, nothing
gets overlooked and everyone enjoys the same initial experience. So, while
it’s true that setting up automated systems requires an investment of time
and money, the long-term pay-off can be significant.

Creating a more engaging, personal onboarding

experience
Because so much of onboarding involves routine, task-based processes, it
can very easily feel impersonal to the employee joining the organization. An
exercise in ticking boxes on a list and little more. While it may seem coun-
terintuitive, AI can be used to inject a little more personalization into the
onboarding journey, making it feel more human and welcoming.
In fact, one of the things I like most about data is it allows for so much
personalization. When you’re automatically tracking and monitoring the
employee onboarding process, you gather a wealth of information on how
that employee is doing and how they feel about the onboarding process.
Used well, this information can help you create a more personalized, engag-
ing onboarding process that’s tailored to each individual – and, in turn,
promotes a positive impression of the organization.
As an example, AI can leverage data from onboarding surveys and auto-
mated chats (not to mention social media and other data sources) to gain
insights into new hires’ preferences, strengths, and areas they wish to de-
velop. This information can be used to tailor onboarding programmes to
meet the unique needs of each individual, providing them with relevant and
meaningful content that plugs gaps in their learning – or better aligns with
their goals and interests.
To put it another way, AI tools can be used to connect new hires with the
right people, right information and right resources, all at the right time.
But to achieve this you’ll need data, and plenty of it. You’ll need to be
collecting data throughout the hiring process, throughout the onboarding
process and from your existing workforce. And you’ll need to be taking the
temperature of employees on a regular basis, so that you can gauge senti-
ment and track progress. Not only should you use all this data to deliver a
more personalized onboarding experience for new hires, you should also use
it to refine your onboarding experience in general. If the same feedback is
coming from multiple employees, for example, that gives you an opportu-
nity to refine the experience for everyone in future.
 Improving employee onboarding 111

What about the downsides of AI in onboarding?

As with any use of AI, you need to be aware of the potential pitfalls. The
good news about onboarding – compared to, say, recruitment – is it’s largely
objective (by which I mean the same tasks are carried out at the same time
for all new hires). So there’s less opportunity in general for bias to creep into
automated onboarding processes. Good news.
Perhaps the biggest challenge, then, is finding the right balance between
the human element and the tasks that are best delegated to machines. Data
and AI should not replace the role of human HR professionals in initiating
new people into the business – rather, it should be used as a way to allow
HR professionals to focus their time where it has the most impact. Sending
requests for documentation or confirmation letters to multiple new hires is
something that can easily be handed over to machines. But it’s vital that
employees receive a genuinely human welcome to the organization. From
interviews and reviews to mentorship opportunities, there are plenty of
touchpoints where the human element is absolutely vital.
Bottom line, you should seek to automate those mundane, easily repeat-
able tasks, while recognizing that humans will always be needed in the on-
boarding process.

Onboarding new hires in the metaverse

If we can recruit people in the metaverse (see Chapter 7), why not onboard
them in the metaverse as well? While this won’t apply to all employees or all
tasks in the onboarding journey, it’s certainly worth considering the
metaverse as an onboarding tool – especially for remote hires who will never
set foot in your office.
What do we mean by onboarding in the metaverse? Well, as the metaverse
is evolving so rapidly, there are no hard and fast rules. But, in general, we’re
talking about providing immersive, virtual onboarding experiences that
mimic ‘being there’ in person. Think virtual meetings or virtual training ses-
sions that provide a more engaging experience than your average Zoom call.
Onboarding in this way – especially for remote hires – can make the process
feel more fun and help to foster a deeper connection with the organization.
You can even take remote onboarding to a whole new level and create a
digital twin of your real-world office – a virtual space that looks exactly like
the office that new hires encounter in the real world. This could be used to
provide virtual tours (useful for any new hire who will be joining the office
112 Data-driven and AI-enabled HR in practice

in person, so they can orient themselves before their first day), or a realistic-
looking space where new hires can meet team members or colleagues.
It’s early days for metaverse onboarding, but one company leading the
way is Accenture. The IT services and consulting giant has created a virtual
campus known as One Accenture Park, located in what Accenture calls ‘the
Nth Floor’ (basically, the company’s metaverse presence). Resembling a cor-
porate theme park, One Accenture Park features a virtual boardroom and
conference room, with a virtual zip line that whizzes the employee (or rather,
the employee’s personalized digital avatar) off to different destinations in
the virtual campus.
In the past, the onboarding process at Accenture involved new employees
meeting senior leaders and other colleagues in face-to-face meetings. But
that wasn’t possible during the pandemic. So One Accenture Park was de-
signed to offer a more immersive place for new hires to learn and network
with colleagues – all they have to do is don their VR headset.
Accenture says this isn’t designed to replace real-world interactions be-
tween employees.2 Rather, the metaverse campus provides a focused space
for new hires to meet with leaders and colleagues – without the distraction
of emails and notifications, as is often the case in the physical world.
It’s not just about networking, though. In One Accenture Park, new em-
ployees can learn in a more fun, interactive way, thanks to a wealth of on-
boarding materials and exercises. For example, they can ascend the
‘leadership mountain’ attraction in the park while learning the fundaments
of effective management. There’s a ‘phishing pier’ where employees can
learn cybersecurity awareness. There are also realistic virtual replicas of
Accenture’s physical offices so new employees can learn their way around.
(These can also be used by any employees for virtual meetings.)
Unconvinced of the metaverse’s potential to deliver information to new
hires? Consider the 2019 study from a group at the University of Nottingham,
UK, which researched the effectiveness of VR training versus a traditional
PowerPoint training session. The researchers found that those who partici-
pated in VR training had better levels of engagement and long-term reten-
tion of information than those who endured the dreaded PowerPoint.3
Accenture has noticed similar results; the company says that providing in-
teractive training in the metaverse improved the retention of information by
33 per cent.4 Importantly, employees seem to love it – with Accenture’s vir-
tual onboarding experience receiving a feedback rating of 4.7 out of 5. With
feedback like that, it would be daft to ignore the potential of the metaverse
to provide a better onboarding experience.
 Improving employee onboarding 113

Key takeaways

To summarize the main points from this chapter:

●● Data and AI can play a significant role in onboarding new hires, helping
to create a more personalized and engaging experience for new hires –
while at the same time saving time and effort through automation.
●● When it comes to automation, some of the tasks that can be carried out
by software bots include sending standard communications, answering
frequently asked questions (FAQs), collecting feedback during the
onboarding process, and more. This frees up HR professionals’ time,
allowing them to focus on employee interactions that have the most
impact.
●● AI can also be used to inject more personalization into the onboarding
journey, making it feel more human and welcoming. AI can leverage data
from onboarding surveys and automated chats, for example, to gain
insights into new hires’ preferences, strengths and areas for
improvement. This can be used to tailor onboarding programmes to meet
the unique needs of each individual.
●● It’s really important to find the right balance between those tasks that
should be allocated to machines, and those that require the human
touch. So, while you should absolutely seek to automate mundane, easily
repeatable tasks, remember that human HR professionals will always be
needed in the onboarding process.
●● Finally, for many employers, the metaverse could provide exciting new
ways to deliver orientation and training sessions – providing an
immersive onboarding experience, while improving knowledge retention.

It’s clear that data and AI can help to smooth the onboarding journey. But
once an employee is established in their role, how can data and AI be used
to boost their performance? Turn to Chapter 9 to find out.

Notes
1 Leena AI. The state of employee onboarding in the US, Leena AI, 2022. leena.ai/
state-of-employee-onboarding-united-states (archived at https://perma.cc/BK6C-
WL7Z)
114 Data-driven and AI-enabled HR in practice

2 O Pickup. Why more companies are sending new hires straight to the metaverse
for improve onboarding, Work Life, 17 June 2022. www.worklife.news/
technology/why-more-companies-are-sending-new-hires-straight-to-the-
metaverse-for-improved-onboarding (archived at https://perma.cc/QMH3-
QDFB)
3 University of Nottingham. Study finds virtual reality training could improve
employee safety, EurekAlert, 16 September 2019. www.eurekalert.org/news-
releases/530825 (archived at https://perma.cc/9N65-Q428)
4 HRM. How Accenture is onboarding its employees in the metaverse, HRM,
7 December 2022. www.hrmonline.com.au/onboarding/accenture-onboards-
employees-in-the-metaverse (archived at https://perma.cc/7B67-4FKL)
 115

Performance 09
monitoring and
management
Generally speaking, measuring and reviewing the performance of employees
is done poorly by many companies. Traditional methods such as annual
performance reviews are often disliked by both the employees being re-
viewed and the managers conducting the reviews, and can be a huge waste
of time.
Thankfully, new approaches driven by data and AI are allowing HR
teams to optimize employee performance monitoring and management, and
gain valuable insights into employee productivity, performance, engagement
and areas for improvement. By leveraging data from various sources – such
as HR systems, performance reviews and employee feedback – AI algorithms
can identify performance trends, patterns and correlations, allowing HR
professionals to gain a deeper understanding of what drives employee per-
formance. This information can then be used to develop more targeted and
effective performance management strategies, such as personalized coach-
ing, goal-setting and recognition programmes. AI-powered performance
management systems can also provide real-time feedback and performance
metrics, allowing employees to monitor their progress and receive actiona-
ble insights for improvement. Furthermore, data and AI can be used to
identify high-performing employees and potential successors, allowing
­
­organizations to develop talent pipelines and succession plans.
As we’ll see in this chapter, organizations can now use data and AI to cre-
ate a more objective, transparent and effective performance management
process, improve employee engagement and retention, and drive better
­business outcomes.

A quick word of warning before we start

Measuring people performance with data and analytics can bring many
benefits, but it must be applied carefully. Because there’s a fine line
116 Data-driven and AI-enabled HR in practice

between performance improvement and employee surveillance, and

companies that have overstepped this mark have faced huge
backlashes. Most people don’t want their boss to monitor their every
move, and in fact this can be hugely demoralizing for staff, particularly
the most self-motivated members of the workforce. To avoid your
company coming across as some sort of Orwellian tyrant, you’ll need to
tread a fine line, gathering the data that you really need to give people
genuinely useful feedback, without freaking out your workforce or
damaging your employer brand.
Therefore, it is vitally important you ensure that data and AI are used
ethically, transparently and with human oversight to a) avoid unintended
consequences and b) maintain a human-centric approach to performance
management. This is a difficult balance to achieve and maintain, no doubt
about it, and my concern is that many employers won’t get this delicate
balance right. We’ll talk more about this topic later in the chapter and
explore how you can find the appropriate balance for your organization. But
I wanted to flag this topic now so you can keep it at the front of your mind
as you read the chapter.

Lessons from the world of sports

Why start with sports, you ask? Well, sport is often at the cutting edge of
data and analytics, and it provides a useful glimpse of how data can be used
to drive very real performance improvements. Coaches across a whole range
of sports, from cycling to football, are using data to assess and improve in-
dividual performance.

Measuring physical performance and sleep

A number of NFL American football teams use an athlete tracking system
called OptimEye, developed by Catapult Sports. A lightweight wearable de-
vice (worn in a small top that looks a bit like a sports bra) tracks metrics
such as players’ speed, motion and heart rate, and calculates player exertion.
Having this data helps coaches and support staff identify which players are
 Performance monitoring and management 117

working hardest in practice and who could work harder, as well as prevent
illness or injuries from players pushing themselves too hard. It also means
workouts and practice drills can be tailored to each individual on the team.
Plus, if a player does get injured, the historical data will help ensure the
player doesn’t reinjure themselves during recovery.
Such devices are also used in the UK by many soccer teams to monitor
and design individual training routines, and spot early warning signs of
injury. Looking to the future, devices are being developed to monitor fac-
tors like adrenaline and cortisol (the stress hormone) levels, as well as
­perspiration levels.
But sport performance isn’t just about physical exertion; good quality
sleep is another critical factor in getting athletes to perform at their best. A
Stanford University study found that basketball players who slept for an
extra 90 minutes improved both the accuracy of their shots and how fast
they could run.1 In football, many clubs give players wristbands to assess
sleep quality, in order to assess any potential problems and find solutions
that will help boost the player’s performance.

Moving to real-time analysis

When it comes to analysing player performance in matches, most analysis
was traditionally done post-match using video analytics. Not anymore. In
2015, the International Football (soccer) Association Board agreed to
change the rules governing the use of wearable devices, opening up the
potential for league and competition organizers to allow players to wear
such devices during matches themselves. This provides coaching teams
with a wealth of new possibilities to track actual player performance dur-
ing a match and potentially make changes at half-time based on what the
data is telling them. The hope is that the use of tracking devices would
also help reduce the number of deaths from cardiac arrest.
While in-depth monitoring like this clearly goes way beyond what the
average company is capable of, the use of cheap and readily available fit-
ness trackers could change this. It’s not inconceivable, for example, for
employers to use sleep data to understand who may be too tired for a cer-
tain job, especially if it’s a dangerous task. Even in a typical office setting, a
critical sales pitch or meeting could be allocated to the employee who is the
most rested.
118 Data-driven and AI-enabled HR in practice

Intelligently measuring employee

performance
There’s clear evidence that measuring performance delivers very real opera-
tional and financial improvements, as the UPS example later in this chapter
shows. However, ‘performance’ is the critical word in that sentence –
­companies need to make it clear they’re monitoring performance and not
individual behaviour.
This is important because, thanks to the sheer amount of data we can
now gather on employees, it’s possible for HR teams to understand more
about their employees than ever before – including how they think, what
they’re feeling, who they interact with and how productive they are. To im-
prove people performance in a meaningful way without alienating your
workforce, it’s important to drill down to the right metrics that will drive
performance while maintaining employee engagement. This is likely to in-
volve looking at metrics such as which factors motivate employees, what
stops them performing at their best, where people are dissatisfied in the or-
ganization and so on, rather than metrics like how many hours employees
spend at their desk or how long they spend in the bathroom.
In this way, data-driven, intelligent performance management is about
finding a more grown-up way of measuring performance, where intelligent
people understand exactly what they need to do to help the company suc-
ceed, and data is used to see how this is going in reality.

The Internet of Things – and how happy, connected

employees are more productive
The rise of Internet of Things-enabled devices, particularly wearables, plays
a huge role in HR’s ability to effectively measure performance. This can
mean measuring physical movements, such as how staff are coping in chal-
lenging physical conditions (there’s more on employee safety and wellbeing
in Chapter 11), or how people are interacting with each other, such as the
IoT-enabled smart employee badges that we saw in Chapter 3.
Using technology to drive efficiency isn’t a new thing. In the 1990s, tele-
coms company Bell Canada gave phone technicians devices to wear on their
wrists that let them enter repair data without having to go back to the com-
puter in their vehicle. And this reportedly saved each technician almost an
hour a day.2 But this technology has leapt forward with modern wearables,
such as employee badges and wrist bands.
 Performance monitoring and management 119

What particularly interests me is how wearables can help measure and

improve both productivity and wellbeing. After all, happy employees are
more productive, and connectivity plays an important role in this. In one
example, Bank of America used wearable technology to measure employee
activities, and found that call centre workers who took breaks together were
happier than those who didn’t. So the company instituted a group break
policy, and saw a double-digit bump in productivity as a result.3 This a great
example of using data and analytics to drive employee performance in a
positive, thoughtful way.
Wearables are also helping employees work smarter. (Note I said ‘smarter’
not ‘faster.’) One example comes from digital consulting company Accenture,
which helped Airbus boost productivity in one specific part of the business
by a whopping 500 per cent.4 They deployed wearable devices call ‘head-up
displays’ to help component assembly workers access assembly information
quickly, whenever they needed it. As a result, workers were able to assemble
more components more quickly and with dramatically fewer errors.

Other data-driven performance measurement systems

Aside from wearable devices, there are a number of different technologies to
help you measure performance. Once again, I want to stress that the idea
behind this should be to help individuals and the company as a whole per-
form better, not to punish individuals who aren’t performing well. If some-
one isn’t performing a task as well as expected, there may be a very good
reason, from fatigue or stress to systems not working properly or impeding
what the employee is trying to do. Data and analysis should help get at the
why of performance metrics, as well as the what.
Tracking computer usage has become relatively commonplace. Indeed,
it’s now possible to measure virtually everything an employee does on their
computer. Software from Veriato, for example, logs web browsing, docu-
ment use, email use, chat applications and keystrokes, and takes regular
screen grabs that are stored for a certain period of time. It also has the po-
tential to alert managers when certain thresholds are met. Personally, I think
this is edging very close to the line between what’s acceptable for boosting
performance and what’s infringing on individual privacy – more on that
coming up later. However, speaking to The Wall Street Journal, one Veriato
client said the system delivered real benefits. Celeste O’Keefe, CEO of Dancel
Multimedia, uses the system to measure a team of 16, made up of animators,
artists, administrators and salespeople. O’Keefe felt the system allows her
120 Data-driven and AI-enabled HR in practice

team to be more streamlined and focused, and she finds it useful for guiding
her people in the right direction. O’Keefe uses the system to skim graphs and
screengrabs to spot problems with employee productivity. Often these are
the result of someone not being familiar with certain software or systems –
thus identifying opportunities for training and guidance. However, O’Keefe
also acknowledges that her using the system has led to at least one firing.5
The market is also awash with productivity tools and apps that monitor
performance without necessarily tracking the employee’s every move. One
example comes from Basecamp, productivity and project management soft-
ware that allows staff to add their upcoming tasks for the day, week or
month and tick them off as and when they’re completed. This allows man-
agers to easily see what people are working on and how much they’re able
to get done. Similarly, the Asana app allows managers to assign tasks and
track their progress in real time. For sales teams, Salesforce details how
many sales calls and emails were made in a day and how much revenue has
been generated from that activity. Tools like these can drastically help to cut
down the amount of time managers and staff need to spend emailing each
other with updates on projects or holding team meetings.
Going a step further, there are also AI tools that can help you predict
performance. AI capabilities mean it’s now possible to identify characteris-
tics and activities that are linked to high and low performance, and predict
relationships between factors like employee characteristics, training invest-
ment, employee engagement and performance. For example, predictive an-
alytics firm iNostix (now a Deloitte company) provides predictive systems
that it claims can lead to faster time-to-contribution, predict organizational
effectiveness, accurately assess employee engagement and predict absentee-
ism or the risk of workplace accidents.

Intelligently reviewing employee

performance (and not on an annual basis)
The way that many companies manage employee performance is through
traditional annual reviews that evaluate employees against certain key per-
formance indicators (KPIs). Yet, in today’s fast-paced, technology-driven
workplaces, annual performance appraisals just aren’t working any more.
Business moves so much faster these days. It’s no surprise that one study
indicated that only 6 per cent of companies thought their performance man-
agement processes were working.6 To me, the traditional performance
 Performance monitoring and management 121

r­eview model is a perfect example of how not to review performance be-

cause, by its very nature, the process involves looking backwards far more
than looking forwards. Plus, employees dislike annual reviews because they
usually have to fill out lengthy questionnaires, and managers dislike them
because they’re incredibly time consuming. In fact, an organization’s pro-
ductivity can dip as much as 40 per cent during the annual review period.7
Now, companies are starting to move away from annual reviews, gener-
ating more regular discussions and looking to the future rather than the
past. Data-driven performance reviewing should be about creating an ongo-
ing dialogue between employees and management, all based on and facili-
tated by data and evidence. This may include using AI-driven systems and
conducting much more regular (but shorter) reviews, as we’ll see through-
out this section.

A word on linking incentives to performance

Before we get into that, although designing incentive schemes is beyond the
scope of this book, I think it’s important to note that data-driven employee
performance isn’t about simply hardwiring KPIs and performance reviews
to the incentive system. So many companies design narrow metrics that
drive all the wrong employee behaviours; when people know they’re being
evaluated on certain metrics only, those are the activities they focus on,
sometimes to the detriment of other value-building activities. Say, for exam-
ple, I ask my children to tidy their room and promise a cinema trip in return,
but they know I only evaluate how tidy the floor is and never look under the
bed or in the cupboard. Which areas do you think they’ll tidy and which will
they ignore? It’s obvious, isn’t it? And yet employee reviews and incentives
are often designed the same way. That’s why, in my mind, it’s better to focus
on outcomes rather than narrow metrics – meaning, if the company is per-
forming well and individuals are contributing to that success, then they
should be rewarded accordingly.

CASE STUDY  ow big employers are overhauling performance

H
reviews

A number of big companies, such as Accenture and Deloitte, have done away
with the dreaded annual performance review and revamped their review
122 Data-driven and AI-enabled HR in practice

processes. It’s no wonder. In a survey Deloitte itself conducted, it found that

more than half of executives did not believe their employee review systems
drove employee performance or engagement.8
But what do these companies use in place of annual reviews, rankings
systems and 360-feedback models? The new systems generally focus on the
employee in his or her own role, as opposed to ranking employees against one
another or comparing performance to other employees. The focus is therefore on
generating a richer, nuanced view of every employee to facilitate better
performance.
Such systems also provide feedback much more often. Rather than a single
review once a year, they tend to conduct more frequent reviews, at the end of
each major project or every month, for example. More frequent check-ins and
reviews mean that a manager has more opportunities to steer an employee
towards his or her best performance. These more regular reviews typically take
far less time to complete. Deloitte, for example, distilled its performance reviews
down to just four questions, two of which require yes or no answers.9 This new
breed of performance evaluations also focus on looking to the future, instead of
past performance. Rather than reviewing an entire year’s performance at one go,
these shorter, more frequent reviews are designed to help employees move
forward with their careers rather than look back on past accomplishments or
failures. This means people are no longer dwelling on what happened in the past,
but instead focusing on how to improve in the future.
One major problem with standard performance reviews is that a reviewer’s
assessment of an employee’s skills often says more about the reviewer than the
employee. But these new ways of reviewing performance help to remove
subjectivity from the process. For example, to combat potential bias, Deloitte
changed their questions to ask what a manager would do with a person (promote
them, incentivize them, etc.) rather than what they think of that person.

The use of AI in performance reviews

It’s fair to say that many companies have moved away from traditional,
metric-based performance assessment in recent years. Sometimes this is be-
cause such systems have been found limiting. But sometimes it’s because
employers and managers are too easily inclined to simply ignore them, if
the findings don’t line up with their personal ‘gut feeling’ on who they like
or dislike.
 Performance monitoring and management 123

Much of the difficulty in assessing performance lies in navigating work-

place biases. These are well-documented conscious or unconscious behav-
iours that can unfairly influence an assessment of an individual’s contribution
to an organization. Race and gender are perhaps two of the most obvious
sources of individual bias (and fortunately often quite easy to spot). Others,
however, are more ephemeral, and it may not be so immediately obvious
when they are occurring.
One is known as contrast bias, meaning an assessor is inclined to com-
pare an individual’s performance to his or her peers, rather than to defined
standards of achievement. Another is recency bias, where actions in the re-
cent past are given more weight, perhaps unfairly, than actions which hap-
pened further back in time (but still within the period where performance is
being assessed). AI-driven tools can help remove bias from the equation and
build an objective picture of employee performance.
Another great thing about AI is that it won’t treat the job of performance
reviews as something to do ‘when I’ve got time’. Unlike many human man-
agers, it won’t put off assessments until the last minute – tell it you want an
ongoing, 360-degree view of your workforce’s effectiveness and (in theory)
that’s what you’ll get.
And because AI-driven assessment can happen in real time (with systems
monitoring targets, quotas and how these are affected by people’s connec-
tions), incentives and praise for good performance can be dished out imme-
diately. If targets are not being met or performance standards are slipping,
then intervention can take place before the problem grows and becomes
unmanageable.

Using generative AI to write employee performance

reviews?
At the time of writing, I’m seeing a trend emerge for using AI tools – in par-
ticular, the AI chatbot, ChatGPT – to write employee performance reviews.
The thinking behind this is to save managers’ time (because they need only
give the system brief feedback prompts, and it then writes that up into a
more comprehensive review).
While I’m in favour of using AI to analyse and boost performance, I’m
wary of using generative AI tools to actually write performance reviews.
Feedback should come from a manager, not a machine.
Besides, it’s relatively early days for tools like ChatGPT and they may not
always work in the way we intend. As an example, in one experiment using
124 Data-driven and AI-enabled HR in practice

ChatGPT to write performance reviews, the resulting reviews were found to

be ‘wildly sexist’.10 The chatbot had a tendency to impart gendered feedback
into otherwise neutral information. Kindergarten teachers and nurses, for
example, were presumed to be female while construction workers and engi-
neers were presumed to be male. This happened in some – not all – cases
when the system wasn’t given any information on the gender of the em-
ployee. When the system was told the gender of employees, it consistently
wrote longer, more critical feedback for women compared to male col-
leagues in the same job with the same feedback input.
And even when the system did deliver gender-neutral feedback, it was so
generic it wouldn’t be that helpful for employees anyway (for instance, it
was lacking in examples from the employee’s work). But, as I said, it’s early
days for using AI to generate performance reviews, and it will be very inter-
esting to see how the technology develops.

Implementing more regular, or even continual,

feedback loops
We know that short ‘pulse’ surveys can be used to gather more regular feed-
back from employees on how they feel about the company. But this process
works both ways. As well as the company benefitting from regular employee
feedback, employees themselves benefit from regular feedback. Such feed-
back, be it from a line manager, peers or a mentor, helps employees under-
stand their performance, feel recognized for their contribution and feel more
connected with the company, thereby boosting engagement.
As an example, BetterWorks’s AI-driven tool is designed to increase the
frequency of feedback. Its AI algorithms can be used to track employee goals
and progress, and provide comments, nudges and recognition where needed.
The system then prompts feedback from the relevant people, such as a line
manager. Importantly, the system also recognizes an individual’s preferences
for feedback and interactions, such as real-time feedback notifications or
batches of notifications. Solutions like this, which provide instant or very
regular feedback, could provide the ideal solution to the problem of annual
reviews based on data that’s already out of date. Managers can evaluate
performance and deliver feedback based on real-time data, and employees
can get helpful feedback and recognition also in real time.
Peer feedback is another growing aspect of performance reviews. Tools
are now available that deliver continuous (or regular), anonymous peer
feedback to employees, sometimes with recommendations to help them
 Performance monitoring and management 125

i­mprove their performance. Tools like this allow team members to commu-
nicate openly and regularly with each other, and help employees identify
their strengths and opportunities for growth and improvement. And for
managers and HR teams, these systems can help them understand wider
skills, strengths and areas for improvement.
However, employee peer review systems need to be approached with cau-
tion. When used as part of an employee ranking system, which pits employ-
ees against each other, they can be open to abuse and attempts to rig the
rankings by delivering negative feedback on peers. But, used carefully, it’s
easy to see how open, supportive feedback from peers could help individuals
improve their performance, grow as employees and achieve their potential.

Lessons from Amazon: How not to handle

people monitoring and reviews
These instances stem from a few years back, but they remain memorable
(and not in a good way) examples of how not to monitor and manage em-
ployee performance.

Tracking every move

In the UK, working conditions at Amazon’s distribution centres have made
national news in the past, with stories of workers reportedly walking up to
15 miles during a shift, having their every move monitored by GPS tracking
tags, and having just 30 minutes to walk the equivalent of nine football
pitches to get to the canteen, eat lunch and get back to the warehouse.11 The
company reportedly had the ability to monitor staff every minute they were
on site, including their precise location in the warehouse, exactly how many
items they picked or packed, and even how many bathroom breaks they
took and for how long. This delivered huge efficiencies for Amazon, but
potentially might have harmed their employer brand in the UK.

Huge workloads and secret feedback

In addition, the retailer’s feedback culture came in for significant criticism in
a 2015 New York Times article, which focused on the company’s headquar-
ters in Seattle.12 The article, which featured interviews with many ex-Amazon
126 Data-driven and AI-enabled HR in practice

employees, described a ‘bruising’ feedback culture that encourages employees

to criticize colleagues’ ideas and send secret feedback to their managers.
According to the article, every new employee had to subscribe to fourteen
leadership principles, from the not-so-unusual like ‘think big’ to the slightly
ominous sounding ‘disagree and commit’ and ‘frugality’. While many em-
ployees spoke positively of how that helped them excel, other interviewees
described a culture where huge workloads and pressure are commonplace,
with one saying she didn’t sleep for four straight days and others reporting
working at nights, on weekends and on holidays.
According to the article, the harsh performance review culture included
weekly or monthly reviews where individual employees were given lengthy
reports (sometimes 50 or 60 pages) on the various metrics that they were
being held accountable for. They were then quizzed on various aspects of the
report.
But it’s the internal feedback tool that perhaps raises the most eyebrows.
Called the Anytime Feedback Tool, this system allows employees to send
positive or negative feedback about their colleagues to management. It’s a
system open to abuse. Team members are ranked, and those scoring lowest
were (at the time of the article) reportedly eliminated each year, which meant
employees effectively competed against each other for their jobs and every-
one felt they had to outperform everyone else.

Average employee tenure is just one year

It’s perhaps no wonder that a PayScale survey ranked Amazon second on a
list of companies with high staff turnover.13 According to the data, Amazon
employees stick around on average for just one year – one of the very brief-
est tenures in the Fortune 500. At the time of the article, Amazon founder
and then CEO Jeff Bezos responded by writing a memo to Amazon staff
stating that the article didn’t reflect the culture he knew, and asked staff to
report any unfair practices to HR.14 But the high staff turnover indicates
that the internal feedback system has had a negative effect on employee
satisfaction.
For me, one of the main problems with the Amazon feedback system was
not just that employees felt driven to outperform each other, but that their
access to peer feedback filtered down only from their managers (employees
could not see the feedback sent about them whenever they wanted – it had
to be delivered via a manager). To really help employees grow and improve,
they should be able to ask for or access feedback when they need it, like the
continuous performance review systems seen earlier in this chapter.
 Performance monitoring and management 127

CASE STUDY L essons from UPS: How to drive performance

without alienating people

With vehicle sensors and GPS data, it’s possible to know exactly where delivery
drivers are, which route they’re taking or how fast they’re driving, and many
companies are routinely using this sort of data to improve driver behaviour and
optimize delivery routes. UPS, however, has taken the use of data and analytics
to a whole new level. For example, the handheld computer that drivers have
been carrying for years (those electronic boxes you sign to say you received
your parcel) is actually a sophisticated device that helps drivers make better
decisions, such as which order to deliver parcels in for the most efficient route.
But it’s the delivery trucks themselves that provide a wealth of data about
driver performance. UPS trucks are fitted with more than 200 sensors that gather
data on everything from whether the driver is wearing a seatbelt, or when the
back doors are open, to how long the vehicle spends idling as opposed to in
motion, and how many times the driver has to reverse or make a U-turn.

Big benefits from data

With so many drivers on the road, improving driver performance so they drive as
efficiently as possible means big savings for UPS. The company has said that
shaving just one minute off the time each driver spends idling as opposed to in
motion saves over $500,000 in fuel across the whole fleet. UPS has also said that
same minute adds up to operational savings of $14.6 million a year.15 One insight
gained from the sensor data was that drivers opening the truck door with a key
was slowing them down and eating up valuable time. So the company gave
drivers a key fob with a simple push-button to open the doors much quicker. Tiny
time savings like this make a huge difference across a fleet the size of UPS.
And the savings are clear. By monitoring their drivers and providing feedback
and training where needed, UPS has achieved a reduction of 8.5 million gallons
of fuel and 85 million miles per year.16 Plus, while drivers now make an average of
120 stops a day, that number used to be less than 100 – meaning the same drivers
with the same trucks are now able to deliver significantly more packages than
they used to.

Protecting and rewarding employees

This increased performance has been reflected by increased wages, with UPS
drivers now earning around twice what they did in the mid-1990s.17 The
company is widely regarded as the biggest and most efficient parcel shipper in
128 Data-driven and AI-enabled HR in practice

the world – largely thanks to its innovative use of data – and its drivers are
among the best paid in the industry. That no doubt helps support employee
buy-in for monitoring so much of what drivers do. But the company has also had
to take other steps to ensure they don’t face a huge backlash from drivers; for
example, under the terms of drivers’ contracts, UPS cannot collect data without
informing drivers of what they’re gathering. And nor can they discipline a driver
based only on what the data has told them. Sensible safeguards like this would
work for almost any type of performance data in any industry. When
implemented and properly followed, such safeguards help facilitate employee
buy-in, ensure transparency and minimize the risk of damage to morale or
employer brand.

Finding the right balance between trusting

and tracking
It should be clear by now that I’m talking about using data and AI to enable
performance, not surveil your employees. Since I wrote the first edition of
this book, I’ve noticed a clear rise in the use of productivity monitoring tools
that track employees’ computer use in incredible detail, right down to key-
board and mouse activity. This has grown enormously since the pandemic,
as companies allow more people to work remotely. (The implication being
‘If we can’t see you, how can we trust you to do the work?’)

The slippery slope to surveillance

According to The New York Times, many remote workers have even re-
ported being paid only for the minutes where they were actively working at
their computer (backed up by screenshots of the computer and photographs
of the employee sat there).18 So, minutes where they made a cup of coffee,
went to the bathroom, sat back and read an offline document related to
their job, or answered the door to sign for a package, were deducted from
their payable hours.
Incidentally, the same New York Times investigation found that eight of
the ten largest US employers were doing some form of ‘individual employee
productivity monitoring’, often in real time. And we’re not just talking
about the monitoring of employees who are low down on the corporate
 Performance monitoring and management 129

food chain – many are highly experienced, highly paid employees. At

New York’s Metropolitan Transit Authority, for instance, employees such as
engineers and lawyers were told they could work from home one day a
week, providing they agreed to full-time productivity monitoring – so, not
just on the days they were at home but all the time. While some workers
spoken to by The New York Times praised monitoring software for making
things ‘fair’ and ensuring their contribution was seen, the majority weren’t
impressed. I don’t blame them.
This level of employee monitoring is essentially surveillance. As someone
who works from home almost all of the time, I find this sort of approach at
best infantilizing, and at worst it is an infringement of personal privacy.
Consider, for a moment, what motivates you as an employee and what
would send your satisfaction and engagement plummeting. Would having
your photo taken at your desk every 10 minutes and bathroom breaks mon-
itored motivate you to perform at your best?
Personally, I’m very self-motivated – give me a goal and I’ll achieve it. If I
felt my boss was tracking my every move to make sure I achieved that goal,
I would hardly thrive. If we think of the publisher of this book as my boss
for a second, I know I wouldn’t be happy if my ‘boss’ was monitoring how
many minutes I spent typing at my computer. On paper, those minutes spent
‘idling’ and not typing would look unproductive, and yet those minutes are
critically important for research and organizing my thoughts. If my every
keystroke was monitored, I’d feel extremely demotivated and disengaged.
Not only that, while such monitoring may lead me (at least in the short
term) to produce more words per day, the quality of my output would likely
go down, not up. That’s the danger with monitoring employees – self-­
motivated employees could well be put off and you could end up with the
opposite effect to what you intended.
In other words, it’s really important to find the right balance between
trusting your employees and monitoring their performance. And this goes
for remote workers and those who are based on site. You want to enable
great performance, rather than control employees’ activities. If we refer
back to the world of sports for a moment, data is used to enable players and
athletes to perform at their best – it’s not used to micromanage and control
individual behaviour or as a tool to punish poor performance.
Sure, sometimes you do need to monitor and manage an employee a little
more closely than usual – especially when there’s been a problem with per-
formance – but, for the most part, your goal is to enable people to make
better decisions and achieve their potential.
130 Data-driven and AI-enabled HR in practice

Avoiding employee backlash

As well as the risk of demoralizing employees, there are legitimate privacy
concerns about monitoring employee performance. Employees have a right
to privacy, after all, and that must be carefully managed alongside the need
to better understand their performance. Improper use of data (see Chapter 4)
can not only tarnish your employer brand, but it could potentially land you
in legal trouble.
Say, for example, that employees wear badges that track their interac-
tions with customers and other staff. A manager could potentially use this
information to identify which member of their team went to HR with a
complaint. If the manager then fired that employee, with no other grounds,
the employee would have an excellent case for unfair dismissal. Or, if em-
ployees are wearing fitness tracking devices, for example, there is a danger
that health data could be used to discriminate against those who are less
physically healthy, regardless of how well they perform in the job.
One real-life example of employee monitoring backfiring came from
British newspaper the Daily Telegraph back in 2016. Journalists reportedly
arrived at work one morning to find motion sensors had been installed
under their desks, without any warning or explanation whatsoever.19 The
employees’ union got involved and the newspaper’s management quickly
removed the devices. I find it pretty shocking that this measure wasn’t com-
municated to staff prior to the installation – if it had been handled better,
they could have avoided all the uproar.
As well as communicating performance-monitoring measures to employ-
ees, it’s also really important to remember that improved performance and
increased output aren’t necessarily the same thing. If all you’re doing is try-
ing to increase output, with no consideration of employee wellbeing and
engagement, it’s likely to backfire and lead to increased stress. And that’s
especially true in today’s fast-paced working environment, where 77 per
cent of people say they have experienced burnout on at least one occasion.20

Surveillance can have other negative consequences

Alienating your employees and increasing the risk of burnout are two poten-
tial downsides of surveilling employees. But that’s not all. One study found
that employee surveillance can, surprisingly, lead employees to break the
rules, not follow them. The researchers found that US employees who were
under surveillance took more unapproved breaks, intentionally worked
more slowly and even stole more office equipment than colleagues who
 Performance monitoring and management 131

weren’t under surveillance.21 To test their theory and prove causation, rather
than correlation, the researchers carried out a second study in which work-
ers were given a set of tasks to complete as well as the opportunity to cheat.
And it turned out, people who knew they were under surveillance were
more likely to cheat on their tasks. Why? Those who took part in the study
reported feeling a lack of agency and personal responsibility. Monitoring
employees too closely can remove people’s sense of agency and even create
the very behaviours you don’t want to see in the workplace.

So, how can you strike the right balance?

Here are six simple best practice guidelines to help you walk the fine line
between legitimately useful performance measurement and downright
­surveillance:

1 Be transparent. Be transparent with your employees about exactly what

data you collect and how you intend to use it. Be specific on how this will
benefit them and help improve company performance overall. Make it
clear it’s about looking at performance, not watching over every little
thing employees do, and that it would never be used to punish individuals.
2 Minimize the data you collect. Practise thoughtful data minimization
and only collect the data you need to have a genuine impact on
performance. You should always be able to justify exactly why you need
certain data. If there’s no good business case for collecting it, don’t do it.
3 Get consent. You must ask your employees for consent to use their
performance data. And, once you’ve got consent, only use data for the
purpose for which employees have given consent.
4 Work with the union. If your employees are members of a union, like
UPS drivers, you’ll need to consult with the union and gain agreement on
performance measurement practices before you implement any measures.
5 Maintain dialogue. Keep your employees informed when you make any
changes as to what data is gathered and how it’s used. Just because you
got their buy-in once, doesn’t give you carte blanche to monitor anything
you like in future.
6 Demonstrate clear benefits from the data. Be vocal about successes and
show how data-driven performance measuring and reviewing improves
the bottom line and helps the company meet its goals – just as in the UPS
example. And when improved employee performance delivers better
financial performance for the company as a whole, reward your people
accordingly.
132 Data-driven and AI-enabled HR in practice

Key takeaways

Clearly, this is one of the trickier areas of data-driven HR, and a lot of
careful thought is needed on what’s right for your organization and what
will enable your employees to perform at their best. Here’s what we’ve
learned in this chapter:

●● There’s a wafer-thin line between performance improvement and

employee surveillance, and companies that have overstepped this mark
have faced huge backlashes. Done badly, measuring performance can
destroy employee motivation.
●● You should be looking to uncover which factors motivate employees,
what stops them performing at their best, where people are dissatisfied
in the organization and so on, rather than metrics like how long
employees spend in the bathroom.
●● The idea is to help individuals and the company as a whole perform
better, not to punish individuals who aren’t performing well. If someone
isn’t performing a task as well as expected, there may be a very good
reason, such as fatigue or stress.
●● The IoT and connected devices are playing an increasingly important
role in driving employee performance. (Remember, connected workers
are generally happier and more productive.) Likewise, AI is being used to
create a more objective approach to performance management, and
even predict performance.
●● In today’s fast-paced, technology-driven workplaces, annual
performance appraisals just aren’t working any more. Companies are
moving away from annual reviews, and instead are generating more
regular discussions and looking to the future more.
●● It’s really important to find a balance between monitoring employees and
treating them like trusted adults. You want to enable performance, not
micromanage or control people. Follow my best practice tips for using
performance data in a fair, ethical way.

When it comes to enabling performance, we can’t ignore the importance of

training and developing your people. So, in the next chapter, let’s explore
how data and AI can help you provide a better learning experience for
­employees – and prepare the organization for the future.
 Performance monitoring and management 133

Notes
1 E Singer. Extra sleep boosts basketball players’ prowess, MIT Technology
Review, 7 July 2011. www.technologyreview.com/2011/07/07/193127/
extra-sleep-boosts-basketball-players-prowess (archived at https://perma.
cc/4TBE-A4QK)
2 H James Wilson. Wearables in the workplace, Harvard Business Review,
September 2013. hbr.org/2013/09/wearables-in-the-workplace (archived at
https://perma.cc/748X-TP36)
3 S Frankel. Employers are using workplace wearables to find out how happy
and productive we are, Quartz, 11 August 2016. qz.com/754989/employers-
are-using-workplace-wearables-to-find-out-how-happy-and-productive-we-are
(archived at https://perma.cc/T5AP-442V)
4 Accenture. Airbus soars with wearables, Accenture, nd. www.accenture.com/
bg-en/case-studies/aerospace-defense/airbus-wearable-technology (archived at
https://perma.cc/A3LD-HZWW)
5 T Greenwald. How AI is transforming the workplace, The Wall Street Journal,
12 March 2017. www.wsj.com/articles/how-ai-is-transforming-the-­
workplace-1489371060 (archived at https://perma.cc/5549-K5Y4)
6 B Marr. The future of performance management: How AI and big data combat
workplace bias, Forbes, 17 January 2017. www.forbes.com/sites/­bernardmarr/
2017/01/17/the-future-of-performance-management-how-ai-and-big-data-
combat-workplace-bias (archived at https://perma.cc/8MVR-E9MV)
7 S Shekhawat. Bots and artificial intelligence – next wave of disruption in HR,
Your Story, 15 December 2016. yourstory.com/2016/12/bots-artificial-­
intelligence-hr (archived at https://perma.cc/AT8Y-V6NB)
8 S Garr. Performance management is broken, Deloitte, 5 March 2014. www2.
deloitte.com/us/en/insights/focus/human-capital-trends/2014/hc-trends-
2014-performance-management.html (archived at https://perma.cc/XPL8-
E4NP)
9 J McGregor. What if you could replace performance evaluations with
four simple questions? The Washington Post, 17 March 2015. www.­
washingtonpost.com/news/on-leadership/wp/2015/03/17/deloitte-ditches-
performance-rankings-and-instead-will-ask-four-simple-questions (archived at
https://perma.cc/G59E-6CX4)
10 K Snyder. We asked ChatGPT to write performance reviews and they are
wildly sexist (and racist), Fast Company, 2 March 2023. www.fastcompany.
com/90844066/chatgpt-write-performance-reviews-sexist-and-racist (archived
at https://perma.cc/4KE5-FKGC)
11 M Ledwith. Tagged by their bosses, zero-hour Amazon workers, Daily Mail,
2 August 2013. www.dailymail.co.uk/news/article-2382800/Tagged-bosses-
zero-hour-Amazon-workers-Employees-guaranteed-income.html (archived at
https://perma.cc/7TCB-ZS3L)
134 Data-driven and AI-enabled HR in practice

12 J Kantor and D Stretifield. Inside Amazon: Wrestling big ideas in a bruising

workplace, The New York Times, 16 August 2015. www.nytimes.com/
2015/08/16/technology/inside-amazon-wrestling-big-ideas-in-a-bruising-
workplace.html (archived at https://perma.cc/7UJA-9MRY)
13 Payscale. Companies with the most and least loyal employees, Payscale, nd.
www.payscale.com/data-packages/employee-loyalty/least-loyal-employees
(archived at https://perma.cc/RE8D-8QGH)
14 J Cook. Jeff Bezos responds to brutal NYT story, says it doesn’t represent the
Amazon he leads, GeekWire, 16 August 2015. www.geekwire.com/2015/
full-memo-jeff-bezos-responds-to-cutting-nyt-expose-says-tolerance-for-lack-
of-empathy-needs-to-be-zero (archived at https://perma.cc/5V8G-JJWT)
15 J Goldstein. The future of work looks like a UPS truck, NPR, 8 June 2016.
www.npr.org/transcripts/481295201 (archived at https://perma.cc/DAW4-
EKKL)
16 B Marr. Five inspiring ways organizations are using HR data, Forbes, 11 May
2018. www.forbes.com/sites/bernardmarr/2018/05/11/5-inspiring-ways-­
organizations-are-using-hr-data (archived at https://perma.cc/C95S-KYBZ)
17 J Goldstein. To increase productivity, UPS monitors drivers’ every move, NPR,;
11 May 2018. www.npr.org/sections/money/2014/04/17/303770907/to-
increase-productivity-ups-monitors-drivers-every-move (archived at https://
perma.cc/4QSV-2KQG)
18 The New York Times. The rise of workplace surveillance, The New York
Times, 24 August 2022. www.nytimes.com/2022/08/24/podcasts/the-daily/
workplace-surveillance-productivity-tracking.html (archived at https://perma.
cc/2Y8F-9Q8R)
19 J Waterson. Daily Telegraph installs workplace monitors on journalists’ desks,
Buzzfeed, 11 January 2016. www.buzzfeed.com/jimwaterson/telegraph-­
workplace-sensors (archived at https://perma.cc/Z2XP-LRDN)
20 Deloitte. Workplace burnout survey, Deloitte, 2015. www2.deloitte.com/us/en/
pages/about-deloitte/articles/burnout-survey.html (archived at https://perma.cc/
AZ77-9EGX)
21 K Morgan and D Nolan. How worker surveillance is backfiring on employers,
BBC, 30 January 2023. www.bbc.com/worklife/article/20230127-how-worker-
surveillance-is-backfiring-on-employers (archived at https://perma.cc/L2TH-
QC6Z)
 135

Data-driven 10
employee
training and
development
A quick glimpse at the digital transformation happening in the world of
education, particularly since the pandemic, shows us how technology can
facilitate learning at all levels, from schools and universities to corporate
learning. Today, everything in education can be measured, from how well a
student performs in tests, to how well they engage with and understand the
pages in an online course. For example, data has been used extensively in
education, even in primary schools, to give a better understanding of skill
levels, thereby helping to identify those who may be struggling and need
extra support.
Developments like this can feed into a corporate training and develop-
ment programme that is intelligently designed around the employees’ and
the organization’s needs. An approach to learning that’s better suited to
today’s rapidly changing workplace, where skills quickly become outdated
and more tasks become automated. In this era of rapid transformation,
upskilling and reskilling the workplace will become an even more critical
part of what HR teams do. It’s up to HR teams to ensure the workforce
stays relevant, and that the organization finds the right balance between
machine skills and human skills. This will be especially important as new
advances like artificial intelligence and the metaverse play a larger role in
the world of work.
Preparing the workforce for this data-driven, intelligent future is cer-
tainly a challenge for HR teams – but, as we’ll see in this chapter, data and
AI can be used to optimize employee training and development pro-
grammes, providing HR teams with insights into employee learning needs
and preferences. By leveraging data and AI, organizations can create more
136 Data-driven and AI-enabled HR in practice

effective and engaging training programmes, enhance employee skills and

­competencies and ultimately drive better business outcomes.

How data and AI are positively disrupting

the education sector
I started this chapter by saying organizations can learn a lot from the data-
driven revolution taking place in schools and universities. So let’s explore
some of these changes.
With more learning now coordinated online and often taking place via a
laptop or tablet, even when the student is in a traditional classroom environ-
ment, increasingly large amounts of data are being generated about how
students learn. Innovators working with educational establishments are
turning this data into insights that can identify better teaching strategies,
highlight areas where students may not be learning efficiently and transform
the delivery of education.

Learning that’s tailored to individual students

Education has always fundamentally been about feedback loops. A teacher
presents a problem and the student attempts to solve it. From that attempt,
the teacher can learn what the student understands and doesn’t understand,
and can adjust his or her instruction accordingly. Likewise, the student un-
derstands more about the problem he or she attempted. When a teacher is
faced with a classroom overflowing with students, data and AI help to fa-
cilitate this feedback process.
Any teacher can walk students through a course. But to pinpoint and
develop the specific problem areas of each student in a classroom of many is
a tough undertaking. This is why numerous adaptive learning companies
like Knewton have sprung up, offering services that analyse the progress of
students, from the nursery class to university level, to create better test ques-
tions and personalized learning materials. Crucially, these data-driven
courses adapt to each individual student. Technology now makes it possible
to assess, in real time, whether a section is too easy, too hard, or just right
for that student, and adjust the remaining course materials accordingly.
Personalized learning like this also allows students to learn at their own
pace, regardless of what the other students around them are doing. Then,
 Data-driven employee training and development 137

the teacher can receive that information and understand where any one stu-
dent might be struggling, or analyse the performance of a class as a whole.

The impact of AI in schools

The best teachers go into the profession because they’re passionate about
educating young people and they thrive on seeing a student’s eyes light up
when they understand a subject. The idea of effectively becoming a data
administrator may not appeal to most teachers. It’s the classic human vs.
machine scenario: as AI gets better at teaching and providing educational
assistance, the question inevitably turns to whether human teachers will be
replaced by computers.
Now, I don’t believe cyborgs are going to take over our classrooms.
Instead, teachers and AI systems will team up to provide stronger, better
educational experiences for students at every level. Here are just some of the
ways AI is positively disrupting education:

●● AI can automate basic, repetitive activities like marking multiple-choice

papers and grading fill-in-the-blank-style homework.
●● Educational programmes can adjust the speed at which individual
students go through coursework, provide additional help when a student
is getting stuck, or provide additional enrichment when a student is
working ahead of the rest of the class.
●● AI can help the teacher provide better learning experiences. For example,
if the software notes that a large percentage of students are missing a
particular question, it can provide important feedback to the teacher that
his or her lesson may need additional details.
●● AI systems can provide valuable feedback to parents, educators and
administrators. This could reduce the need for separate standardized
testing and provide a level playing field for helping to assess teacher and
school performance.

Real-world examples from the education sector

There are numerous examples of how technology is helping both teachers
and students get the most out of their school days. In Wisconsin’s Menomonee
Falls School District, for example, data has been put to use for everything
from improving classroom cleanliness to planning school bus routes, after
138 Data-driven and AI-enabled HR in practice

department leaders were encouraged to attend classes themselves on how to

gain insights from data and analytics.1
Data has also been used to help improve student behaviour. One US mid-
dle school found that, for some reason, the number of pupils being sent to
the principal’s office for disciplinary reasons had grown by a worrying
amount. On examining the data, they realized that this had coincided with
a reduction in school excursions such as ice skating and sledding trips. When
these were reinstated, behaviour among students improved, leading to a no-
ticeable reduction in the number being sent to see the principal.2
Schools are also finding themselves armed with new technologies aimed
at cutting down on exam cheating and plagiarism among students. The
Proctortrack system aims to prevent cheating by using webcams and micro-
phones to monitor students while they sit for online exams. By building
profiles of cheating behaviour, it is able to recognize and flag suspicious ac-
tivity. Proctortrack also uses facial recognition to ensure that the correct
student is taking the test, monitors computer activity to make sure that
­unauthorized sources aren’t being consulted, and even tracks eyeball move-
ment during the assessment. The system can be used for tests taking place in
traditional exam room environments as well as remote learning.
Of course, not all education takes place in the first two decades of life or
in a traditional classroom setting. Online courses mean people of all ages
can grow their skills and knowledge, regardless of their geographical loca-
tion, income level and other factors. These massive open online courses
(MOOCs) – which deliver all of the learning materials and exams via a
computer or tablet – are providing a wealth of insights into the ways that
people learn.
As we’ll see in this chapter, trends from the education sector – such as the
ability to deliver more personalized learning, and to track learners’ pro-
gress – are now filtering into the world of workplace learning. But before
you can optimize your workplace learning offering, you may first need to
identify where there are skills gaps in your organization.

What skills does your organization need?

We know from Chapter 9 that data can significantly improve a company’s
ability to assess performance and pinpoint exactly where employees are per-
forming well and where they may need some extra assistance. In this way,
data helps HR teams identify gaps in learning, so that they can plug those
gaps through appropriate training.
 Data-driven employee training and development 139

Clearly, with data, analytics and automation developing at the pace they
are, and with no sign of that exhausting pace letting up, one major function
of HR is to help fill the digital skills gap. HR teams have a responsibility to
ensure more people in the organization have the necessary skills to prepare
for the data-driven transformation of business. There’s no doubt in my mind
that the ability to tap into and nurture digital skills is going to be critical to
most businesses’ success in the future. Yet, more than 12 million employees
in the UK do not have the necessary digital skills.3
That said, the key skills for success aren’t just centred around digital
skills. Far from it. Certainly, digital skills will be important for the work-
force, but it would be a mistake to overlook the importance of soft skills.
The things that machines can’t do, basically.
In my book Future Skills: The 20 skills and competencies everyone needs
to succeed in a digital world I set out the skills that I believe will help or-
ganizations – and individuals – surf the wave of digital transformation. I’ll
briefly summarize those skills below.
The essential future skills are:

Essential future skills

1 Digital literacy: The digital skills needed to learn, work and navigate
everyday life, including using devices, software and apps with
confidence; communicating and collaborating via digital tools; and
keeping abreast of new technologies.
2 Data literacy: Being able to access appropriate data, work with data
confidently (creating/gathering data, keeping it up to date, etc.), extract
meaning from data and communicate those data-based insights to
others.
3 Technical skills: The huge variety of ‘hard’ skills that are necessary for
many jobs (such as accountancy skills for accountants, coding skills for
software developers, and so on).
4 Digital threat awareness: Being aware of the dangers of being online or
using digital devices, including hacking, phishing, privacy violations,
cyberbullying, digital addiction, etc. And knowing how to keep yourself
(and your organization) safe.
5 Critical thinking: The ability to think objectively, and to analyse issues or
situations based on evidence rather than personal opinions or biases.
140 Data-driven and AI-enabled HR in practice

6 Judgement and complex decision making: For me, the ability to make
sound decisions also includes recognizing the impact that personal
preferences, values and beliefs have on our judgement.
7 Emotional intelligence and empathy: So long as there are humans in the
workplace and human-to-human relationships, we will need emotional
intelligence (the ability to be aware of, express and control our
emotions) and empathy (the ability to see the world from someone else’s
perspective).
8 Creativity: The act of turning imaginative ideas into reality absolutely
belongs in the workplace, especially as we give more and more routine
tasks over to machines. Creativity enables creative thinking and
problem solving.
9 Collaboration and working in teams: As the nature of collaboration and
teamwork evolves to include hybrid workers, fully remote workers and
gig workers, we will need collaborative skills more than ever.
10 Interpersonal communication: Communication in all its forms (oral,
written, non-verbal and, importantly, listening) remains a vital skill for
workplace success.
11 Working in gigs: Because we must all prepare for a future in which
more people work as ‘free agents’.
12 Adaptability and flexibility: In the workplaces of the future, change is
going to be even more of a driving factor than it is today. We must
therefore all develop the mental resilience to thrive amidst constant
change.
13 Cultural intelligence and diversity consciousness: Which means
incorporating a basic awareness of diversity (and the recognition that
diversity is a good thing), and the ability to relate to others from different
backgrounds.
14 Ethical awareness: Because digital transformation has given rise to a
whole new set of ethical challenges to overcome (such as the dilemmas
surrounding artificial intelligence or the use of people’s personal data).
15 Leadership skills: The combination of factors that will shape 21st-
century work – distributed teams, increasing diversity, humans
transitioning to more creative tasks, the gig economy, fluid
organizational structures, and so on – means that leadership skills will
be important not just for traditional leadership roles, but for a wide
range of roles across the organization.
 Data-driven employee training and development 141

16 Brand of ‘you’ and networking: The ability to build and maintain a

personal brand, establish your expertise, grow your network, make new
connections and enhance your career.
17 Time management: By which I mean the ability to work smarter rather
than working longer or harder.
18 Curiosity and continuous learning: One of the most important skills on
the list (if not the most important). Because curiosity and continual
learning is fundamental to being able (and willing) to embrace change.
Which brings me to…
19 Embracing and celebrating change: I’ve already mentioned adaptability
(the mental resilience to thrive amidst change) on this list, but
individuals and organizations will also need the practical skills to
embrace and manage change successfully.
20 Looking after yourself: By which I mean creating a more balanced life,
and taking good care of your mental and physical health.

Do keep these skills in mind as you read this chapter, and consider how your
organization can build and maintain them. But also keep in mind that the
jobs that people do will undoubtedly change in the future – many jobs will
evolve, many will become obsolete, and many new jobs will emerge. So
when you think about the essential skills that will drive your company for-
ward, try to consider not just the jobs that exist today in your organization,
but also the sorts of jobs that may exist in the future.

The major trends in data-driven workplace

learning
Workplace learning is moving away from traditional models where partici-
pants go to a specific place for a set duration of time to learn at a pre-defined
pace. Now, for workers, learning is becoming something to dip into much
more frequently, perhaps in more bite-size pieces, and at their own pace.
Learning is essentially becoming a core part of the day-to-day job – a more
continual approach to learning than we’ve seen in the past.
So how can data and AI aid this process? By leveraging data from
various sources such as employee performance data, surveys and ­feedback,
142 Data-driven and AI-enabled HR in practice

AI algorithms can identify knowledge gaps and recommend personalized

training programmes that align with employees’ learning styles and ca-
reer goals. AI-powered learning platforms can also provide employees
with real-time feedback and coaching, creating a more engaging and in-
teractive learning experience. What’s more, data and AI can be used to
measure the impact and effectiveness of training programmes, allowing
HR teams to continuously improve and optimize their learning and devel-
opment initiatives.

Learning that’s adapted to individual employees

Thanks to online learning, data and analytics, training and development is
becoming increasingly personalized to individual learners. ‘Adaptive’ learning
technology allows courses, segments of courses, activities and test questions to
be personalized to suit the learner’s preference, pace of learning and best way
of learning. As well as allowing individuals to learn at their own pace, online
learning also offers the same big advantage seen in the education sector: the
ability to measure how individual participants are progressing, how well
they’re retaining the information and where additional guidance or informa-
tion might be required. Individual, self-paced online learning is also arguably
more cost effective than pulling employees out of their job for a day or week
to send them on expensive training courses. Self-directed learning like this also
helps integrate ongoing development into workers’ everyday routines.
Danone’s online Campus X is one example of this in action. The food giant
has created a user-friendly online platform where e­ mployees can boost their
development and share best practice and k ­ nowledge with other staff.

The use of AI in personalized learning

Online learning allows for much greater measurement, because learners leave
behind digital traces of everything they do within the parameters of the course
they’re taking – and these traces include how quickly the learner moved
through a particular element of the course, where they paused, where they got
a test question wrong, which material they revisited and even, potentially,
what time of day the learner best assimilates information. Learning manage-
ment systems allow providers to track this data and use those insights to tailor
courses to individuals’ needs and, therefore, make them much more engaging.
 Data-driven employee training and development 143

AI is critical in this ability to provide adaptive, personalized learning. It is

AI (particularly machine learning) that allows providers to identify where a
learner might be struggling and what areas need extra emphasis for that
individual. Analytics company Zoomi, for example, uses AI capabilities to
analyse each learner’s behaviour, performance, engagement and comprehen-
sion to improve learning content and create a uniquely individual learning
experience. Zoomi claims its solution can increase knowledge transfer by
50 per cent and improve business outcomes by 60 per cent.4
The evidence for an AI-based approach is clear. One paper published by
Pearson and UCL set out how AI creates learning programmes that are more
flexible, efficient and inclusive.5 In particular, the authors cited how AI ef-
fectively allows personalized, one-on-one learning to be provided on a large
scale, which is especially beneficially for larger companies with a diverse
network of employees with different training needs.
Naturally, this works both ways. Not only can personalized learning
track how individuals are progressing, it also allows learners to provide in-
stant feedback on course content and features.

Micro, mobile and blended learning

Building on the idea of employees learning when it best suits them, ‘micro
learning’ has become a bit of a buzz phrase in training. Micro learning in-
volves very short bursts of learning, often delivered through short videos of
just a couple of minutes. These are typically delivered as part of a wider
course, and are used to help learners absorb information more quickly and
easily. We all know how information is easier to absorb in small chunks at a
time rather than in one massive deluge. Micro learning capitalizes on this.
Mobile learning is another trend in learning and development, as more
and more providers support mobile devices in their programmes. Mobile
access to learning content allows employees the flexibility to learn when and
where it suits them, for example when there are few distractions around. It
also fits with the increase in remote working.
Finally, ‘blended learning’, a phrase commonly used to describe the mar-
riage of online learning and classroom learning, is proving very popular as
companies transition away from traditional training models. So what works
for your company may indeed be a blend of traditional training courses and
self-directed learning.
144 Data-driven and AI-enabled HR in practice

Remote learning and the virtual classroom

One of my favourite examples of just how much learning has transformed
comes from Harvard Business School (HBS), which created a digital class-
room as a template of the classroom of the future.6 In this virtual classroom,
the lecturer teaches in a specially designed broadcast studio. Cameras follow
the lecturer as they address a huge screen made up of faces – live-feed videos
(with audio) of all the students participating in the lecture, connected to the
class simply by their laptops. This virtual classroom, which HBS has called
HBX Live, took three years of planning and development, and is designed to
integrate seamlessly with HBS’s online teaching programmes.
One thing that’s really striking about this virtual classroom is the extent
to which it feels like a real classroom experience to the lecturer and students
participating. The system was designed so that all microphones would be on
all the time, including the students’, and no one was muted. This makes for
a much more collaborative, authentic learning experience, where students
can laugh along if the professor makes a joke and agree verbally when some-
one makes a good point. When a student wants to interject with a question
or point, rather than raising their hands, they simply click a button on their
computer and their nameplate on the screen turns red, letting the lecturer
know that they have something to say. Students can also type comments via
a chat bar, and the comments then scroll along the bottom of the huge video
screen like a news ticker. And, even more impressive, up to 60 students can
participate in these virtual classes at a time, which is quite a feat when you
consider that’s 60 separate video feeds being managed in real time without
any delay. The system also allows for up to 1,000 additional students to
observe the virtual lesson with a time delay. In the future, we could see em-
ployers develop their own version of the Harvard classroom – or, of course,
tap into services provided by forward-thinking schools like Harvard. There’s
a great example from Walmart coming up later in the chapter.
What’s more, some providers are going a step further to create immersive
courses that incorporate virtual reality and augmented reality (AR). But
we’ll talk more about that later in the chapter.

Making use of MOOCs

Because of their vastness (‘massive’ being the big clue in the name), MOOCs
provide a unique opportunity for data. Huge amounts of data can be gath-
ered both on individuals but also across multiple learners to pick up broader
 Data-driven employee training and development 145

patterns and insights. These courses allow providers to map an individual’s

learning trajectory, identify trouble spots and provide targeted interventions
where needed.
MOOCs from the likes of Coursera and Khan Academy provide accessi-
ble learning opportunities for millions of people around the world, covering
anything from vocational learning to degree-level courses. In MOOCs,
learners undertake self-directed learning when it suits them, engage with
bite-size micro-learning content like short videos and participate in collabo-
rative discussions with other learners.
Today, many corporate learning programmes are making use of MOOCs
to deliver training to their employees. Companies like Microsoft are creating
their own custom MOOCs for employees. Global steel manufacturer Tenaris
has created MOOCs on various topics, including technical topics like
‘Introduction to steel’. Tenaris is now offering the MOOCs externally to at-
tract university students and boost its employer brand. Other organizations,
such as Bank of America, are leveraging content from existing MOOCs to
deliver training on core competencies. This strategy allows businesses of all
sizes to curate a wide range of content that suits their needs in a simple, cost-
effective way. Both strategies are innovative ways of rethinking your learning
offering and capitalizing on the latest advances in learning technology.

Measuring how learners are doing – and how effective

your training is
When learners work through the content in a digital course, they leave a
digital trace of all their actions. This ability to track learners’ journeys gives
training providers and HR professionals the opportunity to understand a
great deal about the learning experience. Indeed, most learning providers
incorporate some sort of learning management system that tracks how
learners progress and provides insights that can help both individual learn-
ers and the company-wide education programme.
Learning analytics should therefore underpin every aspect of employee
learning, from developing better learning programmes, to delivering them in
the most engaging way, to tracking how employees interact with the train-
ing. Data and analytics can also dramatically improve the measurement of
your training offering by showing how effective (or not) it is in practice.
Typically, the effectiveness of corporate training is assessed by employees
filling out a basic evaluation questionnaire after the course. But data allows
us to go so much further, and pinpoint exactly what’s working and what
146 Data-driven and AI-enabled HR in practice

isn’t. Two specific data points that every company should be measuring are
employee comprehension (e.g. are people struggling with various aspects of
the content?), and employee engagement with content (e.g. are they taking
up opportunities for learning, and are they then participating in courses all
the way through or are they ignoring various aspects?).
Data also allows HR teams to create clear, evidence-based links between
training and performance, which is helpful for improving future training
and development, establishing ROI, and securing leadership buy-in for
training programmes.

The cutting edge: Incorporating VR, AR,

digital twins and the metaverse
Interestingly, VR and AR are becoming more common tools in corporate
learning – in particular, many vendors are now offering VR- and AR-enabled
training programmes. Beyond that, digital twin technology and the metaverse
may play an increasingly important role.
Let’s explore the cutting edge of workplace learning.

Using VR to create immersive training scenarios

VR creates an interactive environment by generating realistic images, sounds
and other sensations to fully immerse the user in that environment. It’s easy
to imagine how this technology could be used to provide immersive training
experiences in fields as diverse as medicine, the armed forces, engineering
and many more. This isn’t a new idea – just think of flight simulators, for
example – but the ever-decreasing cost of VR hardware (headsets, gloves,
etc.) has made the technology more accessible to a far wider audience. VR
technology can even be used on smartphones, although that tends to be less
immersive than using a VR headset.
VR brings enormous benefits to workplace learning because it allows
learners to learn by doing and seeing. And, for most of us, learning by doing
or seeing is a much better way of understanding and retaining information
compared to, say, listening to a trainer explain a concept, or reading infor-
mation onscreen in a PowerPoint presentation. In particular, VR has huge
potential when it comes to training for scenarios that are too difficult, dan-
gerous or expensive to recreate in real life – such as firefighter and law
­enforcement training.
 Data-driven employee training and development 147

BP is one company that’s invested in VR-enabled training for employees

working in dangerous situations. To train employees in start-up and emer-
gency exit procedures at BP’s oil refinery in Hull, UK, the company ­partnered
with Igloo Vision – known for creating immersive shared VR spaces. When
you work in an oil refinery, mistakes can be fatal, but this virtual training
allowed employees to learn from their mistakes safely. How they pulled off
this training is particularly interesting; rather than trainees each wearing
their own VR headset, Igloo Vision built a six-meter igloo at the Hull refin-
ery.7 Inside the igloo, employees could experience an extremely detailed rep-
lica of the plant and practise critical safety tasks, all in a safe virtual setting.
What’s great about this is it provides the opportunity to assess whole shift
teams at a time, rather than immersing each individual in their own simula-
tion. This could be the future of critical team-based training exercises and
assessments.
But it’s not just high-risk jobs that benefit from VR training. Athletes are
also beginning to make greater use of this technology. Any sports training
regimen relies on repetition, whether it’s on a football field, tennis court or
wherever. But getting those reps in can be a challenge – particularly if the
weather is bad, or you’re travelling, or even injured. Immersive learning
specialists Strivr set out to change all that with their immersive learning
technology. Although the technology was originally developed for sports
teams, Strivr now works with a wide range of corporate partners to deliver
immersive enterprise learning – including companies like Verizon, MGM
Resorts, Sprouts, JetBlue and Stanford Children’s Hospital.
One Strivr partner is Walmart. And when the US’s largest employer gets
behind VR training, you know other organizations are bound to follow suit.
Walmart is using VR to train associates and the next generation of managers
via its Walmart Academy. Originally, the Walmart Academy comprised
physical classrooms in stores around the US. But now, thanks to its partner-
ship with Strivr, training can be efficiently delivered to Walmart employees
around the world. Employees can access the training via VR headsets and
iPads (provided for training purposes in Walmart stores) or by using the
Walmart Academy app, which is great for bite-size on-demand training ses-
sions. This VR-enabled evolution of the Walmart Academy will deliver
­job-specific training, future skills training, wellness courses and leadership
training. Walmart first began using VR-based training in 2018, so it is no
stranger to the technology, and the company reports that VR has helped
Walmart employees ‘feel more confident and prepared to perform their
job’.8 Interestingly, Walmart also offers immersive classes via video
148 Data-driven and AI-enabled HR in practice

c­ onferencing, using software that’s similar to the Harvard virtual classroom

I mentioned earlier in the chapter.
Clearly, VR can play a key role in a variety of job-specific training. But even
soft business skills can be learned with VR technology. As an example,
VirtualSpeech’s app makes use of Google Cardboard VR smartphone technol-
ogy to allow users to practise their public speaking and interpersonal com-
munication skills. Combining practical experience, online reading materials
and videos that teach basic skills, and instant feedback, the course aims to
help people develop quickly and build their confidence in a safe environment.
It’s easy to see how VR technology like this could be incorporated into train-
ing and development programmes to boost employees’ presentation skills.

It’s reality, Jim, but not as we know it – tapping

into AR
While VR plunges the user into a simulated world, AR is rooted firmly in
reality, and adds an extra layer of information to the real world that the user
sees in front of them. Basically, AR is used to project digital elements – such
as text or images – onto the real world around us, just as the Pokémon GO
mobile game projects Pokémon characters onto the real world via your
phone camera. AR is also referred to as mixed reality or hybrid reality.
Typically, AR is used with a simple smartphone, but there are also specific
AR glasses and headsets available, such as Google Glass and the HoloLens
headset.
Even though it is less immersive, AR still bring many advantages to work-
place learning. In one example, multinational engineering conglomerate
Honeywell has been using AR to address the problem of knowledge ‘leak-
age’, where older workers retire and take their knowledge with them.
Traditionally, retirees were asked to put their knowledge into PowerPoint
slides or Word documents that could be shared in classroom-like spaces
with new hires. But Honeywell found that this passive learning experience
led to an information retention rate of just 20–30 per cent. So they equipped
both departing workers and new hires with a HoloLens mixed reality head-
set, allowing retirees to record exactly what they were doing in their work
so that new workers could see this information overlaid onto their own
work activities in front of them. This more active form of training boosted
the level of information retained from 30 per cent (at best) to 80 per cent.9
How impressive is that?
 Data-driven employee training and development 149

Creating digital twins

Digital twin technology is closely related to AR because it pairs the virtual
and physical worlds. Quite simply, a digital twin is a digital model of a pro-
cess, product or service that allows analysis of data and monitoring of sys-
tems to identify problems before they even occur, prevent downtime and
even plan for the future by using simulations.
How do digital twins work? First, smart components that use sensors to
gather data about real-time status, working condition or position are inte-
grated with a physical item (such as a factory machine). The components are
connected to a cloud-based system that receives and processes all the data
the sensors monitor. This input is then analysed, and lessons are learned and
opportunities are uncovered within the virtual environment that can be ap-
plied to the physical world. One example of this in action comes from
General Electric’s (GE’s) ‘digital wind farm’. GE uses digital twin technology
to inform the configuration of each wind turbine prior to construction in
order to generate efficiency gains.10
While digital twins are largely used to drive performance and efficiency,
it’s not a huge leap to imagine how this technology could be used to enhance
training for a wide range of employees, particularly in the field of ­engineering.

Using the metaverse in your training

Virtual reality, augmented reality and digital twin technologies are all closely
related to – and enablers of – the metaverse. So it’s fair to expect the
metaverse will play a part in workplace learning in future. After all, the
metaverse is all about creating immersive experiences and environments,
usually accessed through VR or AR technologies. So the next step in VR-
and AR-enabled training may involve employers delivering immersive
­training sessions in their own little corner of metaverse.
Indeed, providers are already emerging that provide metaverse training
offerings. One example is Virbela, which offers a platform for virtual class-
rooms, allowing learners and teachers to interact in a virtual, 3D, immersive
space. Another option for metaverse training is immersive simulations of
specific scenarios. If you think about it, in the metaverse, you could create
whatever training scenarios you wanted – such as customer service training
in a virtual store with virtual customers, or engineering training that takes
the user ‘inside’ the workings of a machine or component. Medical training
is already moving in this direction with platforms like Fundamental Surgery,
150 Data-driven and AI-enabled HR in practice

which allows medical professionals to practise surgical procedures virtually.

There are also platforms that provide immersive virtual spaces for compa-
nies to host their team-building events (one example being Teamflow).
Solutions like this are great when you want to bring team members who are
scattered in multiple locations together in one place, without the time and
expense associated with travel. The metaverse also allows great scope for
gamification, with companies like Classcraft creating learning experiences
that provide students with interactive games and challenges.

CASE STUDY Even more cutting edge: The use of generative

AI in training and development

Another cutting-edge area of workplace learning and development involves

using generative AI to automatically create educational content. From written
text and videos to interactive quizzes, generative AI tools (such as the GPT
chatbot we’ve already mentioned in this book) can be used to create a wide
range of educational content – saving trainers a lot of time and effort. (And
saving cost, especially if you currently outsource your content creation.)
It’s now possible, for example, to convert a video lecture or podcast into
written course content through AI. Conversely, your written content can easily
be converted into videos, complete with virtual presenters, thanks to tools like
Synthesia. This content can be incredibly realistic and engaging. Take it from
me – because I use Synthesia’s generative AI tool to create my own video
content. I simply write a script and the tool turns my text into a video featuring
my virtual avatar (I call him ‘Bernard Marr 2.0’). This allows me to create videos
incredibly quickly, without the hassle of setting up cameras, recording myself
and editing footage.
You don’t have to create an avatar of yourself, if that doesn’t appeal. Synthesia
has over 100 stock avatar presenters for you to choose from (great for ensuring
your videos are diverse and inclusive). And you can type your text in over 120
languages. Synthesia reckons its tool has saved companies $5,000 per video
versus traditional course creation methods, helped cut video production time by
50 per cent and helped employers increase engagement by over 30 per cent.11

Already, the Synthesia tool is very impressive. But it’s also important to re-
member that technology like this will only get better and better in future,
allowing employers to create even more immersive and engaging learning
 Data-driven employee training and development 151

materials, quickly and easily. Boring PowerPoint training sessions might

soon be a quaint thing of the past!
But what else can generative AI do besides generating course content?
Well, generative AI tools can help you design and structure course content –
which is great when you’re developing a new training course from scratch.
And, of course, the AI element means learning can be highly personalized to
individual learners, if required.
AI-powered chatbots can also provide learners with round-the-clock sup-
port, answering questions they may have as they complete courses. Many
questions are common across all learners, so it makes sense that chatbots
could handle these FAQs. It’s a bit like having a personalized digital tutor on
hand at every stage of learning.
Established learning providers are already using generative AI to enhance
and create educational content. One example comes from language learning
app Duolingo, which is using Open AI’s GPT-4 generative AI tool to deliver
a more personalized experience to language learners.
Already a pioneer in AI-enhanced education, Duolingo is using GPT-4
as part of its mission to create the ultimate virtual language tutor.12 At the
time of writing, the Duolingo app has harnessed GPT-4 in two ways:
firstly, a role-playing feature that allows users to interact with an AI-
powered persona and carry out various learning-focused tasks (for exam-
ple, conversing with an AI barista in a Parisian café); and secondly, in a
feature called ‘Explain My Answer’, which gives users a detailed, person-
alized explanation of why an answer they have given is correct or incor-
rect. Thanks to generative AI, Duolingo’s app feels more like a human
language tutor than ever. In the future, Duolingo aims to use GPT-4 to
generate new course content, which has traditionally been a bottleneck
area for the company.
Of course, as with other uses of AI, there are challenges in using genera-
tive AI – bias being a potential pitfall. What other challenges lie ahead for
organizations using data and AI to enhance their training?

A quick look at the downsides of data-

driven training and development
Naturally, there are ethical and practical concerns around working with in-
dividuals’ learning data, particularly when it comes to data privacy and se-
curity. (Circle back to Chapter 4 for more on this.) Data breaches are always
a legitimate concern, and rightly so. In one example, a school district in
152 Data-driven and AI-enabled HR in practice

Tennessee inadvertently left the names, addresses, birth dates and full social
security numbers of 18,000 students on an unsecured server for months!13
Ensuring your employees’ data is private and secure is a critical concern for
any HR team. These days, it’s incredibly naïve to think you don’t have to
worry about protecting your employees’ data. Where possible, anonymizing
employees’ training-related data will help. Where anonymizing data isn’t
possible, you will need to ensure the data is kept secure.
Good practices of data minimization and transparency are also essential.
There is no point in gathering data for data’s sake. Therefore, if you don’t
intend to use training data to make improvements, then don’t gather it. It’s
as simple as that. And when you do intend to gather it, make sure you’re
upfront with your employees about what information you’re gathering and
why. If it’s clear this data is being analysed to help improve the delivery of
learning programmes in the future, and to facilitate individuals’ develop-
ment within the company, staff are much more likely to get on board.

Key takeaways

We’ve covered a lot of ground in this chapter, so let’s recap the main
takeaways:

●● Upskilling and reskilling the workforce – basically, preparing the

organization for the intelligence revolution by developing future skills – is
one of the most important and challenging jobs for HR teams in the
future. Data and AI have a huge role to play in this.
●● Workplace learning is undergoing a massive digital transformation, with
key trends being adaptive and personalized learning, micro learning,
remote and blended learning and MOOCs. Data and AI are also being
used to track how employees engage with training programmes.
●● VR and AR are becoming more common tools in workplace learning,
allowing employers to simulate a wide range of scenarios and deliver a
more immersive, engaging training experience. We can also expect the
metaverse to play an increasing role in training and development.
●● And at the very cutting edge, generative AI is being used to
automatically generate course content and support learners’ needs. A
good example is using generative AI to create video content.
●● Do ensure you take necessary steps to protect your employees’ learning
data and minimize data collection wherever possible.
 Data-driven employee training and development 153

Now let’s turn to the final chapter in this part and see how data and AI can
improve the work of HR in ensuring employee safety and wellbeing.

Notes
1 M Rich. Some schools embrace demands for education data, The New York
Times, 11 May 2015. www.nytimes.com/2015/05/12/us/school-districts-­
embrace-business-model-of-data-collection.html (archived at https://perma.
cc/84HS-8588)
2 K Pal. How big data can revolutionize education, Techopedia, 1 May 2018.
www.techopedia.com/2/31725/trends/how-big-data-can-revolutionize-­
education (archived at https://perma.cc/Y86U-SXVE)
3 R Cellan-Jones. More than 12 million fall into UK digital skills gap, BBC
News, 19 October 2015. www.bbc.com/news/technology-34570344 (archived at
https://perma.cc/2PC5-6BZA)
4 Zoomi. Next generation learning, Zoomi, nd. zoomi.ai/next-generation-­
learning (archived at https://perma.cc/V9SH-8CTX)
5 UCL. Why we should take artificial intelligence in education more seriously,
UCL, 13 April 2016. www.ucl.ac.uk/ioe/news-events/news-pub/april-2016/
New-paper-published-by-pearson-makes-the-case-for-why-we-must-take-
artificial-intelligence-in-education-more-seriously (archived at https://perma.
cc/9RVQ-LJY8)
6 J A Byrne. Harvard Business School really has created the classroom of the
future, Fortune, 25 August 2015. fortune.com/2015/08/25/harvard-business-
school-hbx (archived at https://perma.cc/7YRG-TCMQ)
7 Igloo Vision. BP training, Igloo Vision, nd. www.igloovision.com/case-studies/
bp-training (archived at https://perma.cc/AQ79-GKXY)
8 P Albinus. What HR tech is behind Walmart’s new Global Upskilling
Academy? HR Executive, 9 June 2022. hrexecutive.com/what-hr-tech-is-­
behind-walmarts-new-global-upskilling-academy (archived at https://perma.cc/
NL33-E9XT)
9 B Marr. The amazing ways Honeywell is using virtual and augmented reality to
transfer skills to millennials, Forbes, 7 March 2018. www.forbes.com/sites/
bernardmarr/2018/03/07/the-amazing-ways-honeywell-is-using-virtual-and-
augmented-reality-to-transfer-skills-to-millennials (archived at https://perma.cc/
T8JS-BQM4)
10 GE. Meet the digital wind farm, GE, nd. www.ge.com/renewableenergy/stories/
meet-the-digital-wind-farm (archived at https://perma.cc/9P4R-P9DB)
11 Synthesia. www.synthesia.io (archived at https://perma.cc/L8ET-5KAA)
154 Data-driven and AI-enabled HR in practice

12 B Marr. The amazing ways Duolingo is using AI and GPT-4, Forbes, 28 April
2023. www.forbes.com/sites/bernardmarr/2023/04/28/the-amazing-ways-­
duolingo-is-using-ai-and-gpt-4 (archived at https://perma.cc/BKM7-3TZ8)
13 B Marr. Big data in the classroom, LinkedIn, 31 July 2014. www.linkedin.com/
pulse/20140731081214-64875646-big-data-in-the-classroom-why-learning-
will-never-be-the-same (archived at https://perma.cc/2RQF-JNWD)
 155

Employee safety 11
and wellbeing
Employee safety and wellbeing are critical areas of any HR team’s work.
Intelligent, data-driven HR is about using data and AI to better manage
employee safety and boost employee wellbeing and wellness. Technology,
and particularly sensors, has helped to make the work environment safer for
a long time now – think smoke alarms, gas sensors, security and entry sys-
tems and the like – but as we’ll see in this chapter, the emergence of data-
and AI-driven tools has taken this to a completely new level.
By leveraging data from various sources, such as wearable devices and
environmental sensors, AI algorithms can provide real-time insights into
employee health and safety, identifying potential hazards and providing per-
sonalized recommendations for prevention. And thanks to developments in
virtual reality and the metaverse, employers can now create more immersive
and engaging virtual experiences that help employees to identify and man-
age potential risks and improve their overall wellbeing. By using data, AI
and the metaverse in this way, organizations can create a more proactive,
immersive and personalized approach to employee safety and wellbeing,
enhancing employee engagement, retention and overall quality of life.

Improving employee safety with data

and analytics
Making sure people are safe at work is a critically important role of data
and AI. Obviously, there is a sliding scale of technology, from completely
automated robotics-driven factories at one end of the scale, to the more re-
alistic (for most businesses, at least) end of the scale where sensors and other
technology are deployed as part of a health and safety programme. This
chapter assumes that your business sits at the latter end of the scale, where
humans and machines work together to improve employee safety.
156 Data-driven and AI-enabled HR in practice

Indeed, it’s incredibly powerful when workplace systems are aware of the
people in the workplace – what they’re doing, how they’re performing and
how they’re feeling – and this is perhaps one of the main driving forces in
making our workplaces safer (and employees happier).

Embracing technology, not abdicating responsibility

to technology
In this way, I’m not talking about companies handing all responsibility for
employee safety over to machines. There was a Georgia Tech study a few
years ago in which participants were found to trust safety robots over their
own common sense – even when it was obvious the robot was leading them
into a dangerous situation.1 In the experiment, people were guided to a
room by a clearly faulty robot; it either took them an obviously inefficient
route to the room in question, or broke down in the process – all of which
was set up by researcher. Once the participants were settled in the room, the
smoke alarm was triggered and the unreliable robot then guided them
through corridors filled with artificial smoke. Here’s the scary part: even
though the robot was clearly leading people the wrong way, away from
emergency exit signs, most people in the study still chose to follow the robot.
A few even followed the robot into a dark room blocked by furniture –
again all set up by the researcher. People trusted the robot, despite the fact
that it had proven itself faulty or unreliable at the start of the experiment.
This crazy outcome shows how we need to marry technology with human
experience and common sense, rather than just turn over all responsibility
for our safety to machines.
In today’s data-driven world, almost everything can be measured, and it’s
now possible for companies to measure a great deal about what their em-
ployees are doing and how they’re feeling. The world of healthcare shows us
just how much can be measured these days. Cloud-based health monitoring,
which is at the cutting edge of modern medicine, enables healthcare profes-
sionals to monitor people’s health from afar and provide help or advice
when needed. Even the Apple Watch can monitor your heart health and
perform an ECG. As we’ll see in this chapter, wearable technologies can play
a huge role in employee safety and wellbeing, and the technology is advanc-
ing all the time.
 Employee safety and wellbeing 157

How the Internet of Things is making workplaces safer

Of course employers want their workplaces to be safe environments where no
one gets hurt. Yet workplace accidents and work-related health issues remain
a problem. In the UK, for example, the Health and Safety Executive reports
that 565,000 workers sustained an injury at work in the year 2021–22, and
that the cost of injuries and ill health due to working conditions is more than
£18 billion.2 The impact of work-related accidents and health problems is
huge, both for the individuals and their families, but also for the employer in
question and their reputation. Clearly, something needs to change.

Changing employee behaviour

In recent years, the IoT – smart devices, wearables and sensors that are con-
nected to the internet – has transformed the way we think about and deliver
employee safety. One of the challenges in workplace safety is getting em-
ployees to change their behaviour in line with existing company safety rules
or to adopt new safety initiatives. And this can be particularly helpful in
industries or companies that rely on contract or temporary employees, like
construction. The IoT helps encourage employee adoption of safety initia-
tives by providing much clearer monitoring and insights into safety-related
behaviour. IoT devices, particularly wearables but also sensors, can now
generate a mountain of real-time data on workplace safety and employee
activities. Not only can this data show whether safety rules and initiatives
are being properly adopted, it can also lead to insights that help improve
safety programmes in the future. And the more this data improves safety
programmes, the greater employee buy-in is, and the more likely employees
are to adopt new or improved safety initiatives in future.
Crucially, because IoT devices can be used to transmit real-time data for
on-the-fly analysis, managers can then be alerted when unsafe practices are
taking place and take appropriate action. For example, video data can be
used to detect when an employee is not wearing the appropriate safety gear,
prompting a notification to be sent to the employee’s supervisor. Analytics
like this can help to significantly reduce workplace accidents and injuries.
Indeed, our ability to predict workplace accidents is improving all the time.
Researchers at Carnegie Mellon University, for example, used real-world
data to create predictive safety models with accuracy rates of between 80
and 97 per cent.3 The model takes workplace safety inspection data and uses
this to predict not just the number, but also the location of safety incidents
over the next month.
158 Data-driven and AI-enabled HR in practice

The role of wearables

Wearable IoT devices, such as sensors, tracking bands and smart helmets,
are now being used across a variety of industries. Clearly, wearable technol-
ogy will have a huge impact on the field of employee safety, and the vision
of a ‘connected worker’ is becoming reality in many different industries. The
beauty of IoT devices is that they make employees (and their supervisors)
more aware of what they’re doing and the environment around them,
whether that means alerting someone when they’re close to over-exerting
themselves and need to take a break, or raising the alarm when the proper
safety equipment is not being used. This awareness in itself can dramatically
improve safety, because more aware workers are likely to behave in a much
more safety-conscious manner. What is really exciting is the ability not only
to monitor individuals in real time, but also to personalize advice and ac-
tions to individuals based on what the data is telling us.
Sensor technology can also be used to monitor the environment that
someone is working in. Data can be gathered on temperature, noise levels,
humidity, light levels, toxic gases and radiation. Robots can effectively
‘smell’ now, and can use sensors to detect chemical signatures like blood or
alcohol in the air. Blanca Lorena Villarreal, a researcher from the Tecnológico
de Monterrey in Mexico, has developed an ‘electronic nose’ that can be built
into robotic devices.4 For me, there’s no question that the IoT is the future
of employee safety.

Making driving safer

Driving remains one of the most dangerous things humans do, whether it’s
simply driving to and from work, taking to the road to visit clients, or driv-
ing machinery as part of the job.
Driver fatigue is a huge issue that contributes to a significant portion of
accidents. So if your employees are driving vehicles as part of their job, it
pays to make sure they’re not struggling due to fatigue. And this doesn’t just
apply to transportation companies or individual employees taking to the
roads in cars. Driver fatigue can be an issue when driving any kind of vehi-
cle, such as diggers and bulldozers.
Which is why Caterpillar, in association with Australian company Seeing
Machines, rolled out an eye- and face-tracking system that can detect driver
fatigue. The system – originally designed for vehicles used in mining but
since used in regular trucks – incorporates a camera, GPS and accelerometer.
It tracks eye and eyelid movement, such as how often a driver blinks, how
 Employee safety and wellbeing 159

long those blinks last, and how slowly the eyelids are moving – and it can do
all this even if the driver is wearing sunglasses. It can even analyse the posi-
tion of the driver’s head and whether it has started to drop. When a driver
closes their eyes for longer than 1.6 seconds, an alarm is triggered inside the
truck – both a noise and a vibration within the seat. Then, if the alarm is
triggered for a second time, a dispatcher or supervisor will be contacted, so
that they can make contact with the driver via radio. If a third alarm is trig-
gered, the driver would generally be taken off driving duty. The system has
been shown to reduce fatigue events by up to 90 per cent.5 Interestingly, the
system can also be used to detect when a driver is distracted and taking their
eyes off where they should be – again, triggering an alarm in the cab.

Making industrial and manufacturing settings safer

This vision of a connected worker may soon become standard practice in
many industrial and manufacturing settings. Companies such as Honeywell
now offer a range of wearable technologies that can improve employee safety.
Such devices may gather data on heart rate, breathing, motion, posture and
even the presence of toxic gases, with the information pulled together into a
dashboard display that gives supervisors and safety professionals an accurate
picture of what employees are experiencing in real time.
Of course, most industrial and manufacturing settings involve humans
working with machinery – and sensors can also play a vital role in increasing
machine safety and efficiency. Sensors can be used to assess machinery com-
pliance, safety anomalies, machine stoppages (and their causes) and much
more – all of which helps companies better understand what’s going on in
real time on the floor and the safety risks, accurately pinpoint machinery
misuse and reduce safety-related stoppages. Despite the fact that technology
plays a critical role in most industrial and manufacturing settings, safety
management has traditionally relied on rather dated methods and informa-
tion, often based on what’s happened in the past or at other locations. The
ability to gather real-time insights therefore makes a huge d ­ ifference.
One way data is proving particularly valuable is in identifying discrepan-
cies between the way machinery and safety systems are designed to be used,
and the way they are actually used in practice. For example, data may high-
light that emergency stop buttons on machines are not in fact being used in
emergencies, but to clear routine jams. This misuse could reduce the ­efficiency
of the safety system and cause it to fail when it’s really needed, thus putting
people at risk. Insights like this highlight when additional safety training is
160 Data-driven and AI-enabled HR in practice

needed for employees. Plus, with the predictive capabilities of analytics, ma-
.

chine safety systems can predict risk through a detailed risk calculator.

Making construction sites safer

Despite strict safety measures and regulations, construction sites remain
some of the most dangerous places to work in the world; in the UK, con-
struction is the third most dangerous industry to work in, and the rate of
fatal injuries is around four times higher than the average across all indus-
tries.6 Construction sites are dynamic places, constantly in flux, and each
site is different, which makes identifying every possible risk and hazard al-
most impossible.
However, wearable devices can help construction companies monitor
what employees are exposed to and provide safer working environments.
Even hardhats are now being made ‘smart’, and fitted with sensors that detect
fatigue in those operating machinery. In another example from the construc-
tion industry, sensor data proved that one particular mortar board (those
boards with a handle underneath used by bricklayers) actually reduced the
risk of bricklayers suffering lower back injury. The EcoSpot mortar board
system reduced the amount of time workers spent with their backs bent more
than 20 degrees by as much as 85 per cent. Not only that, it led to a 17 per cent
increase in productivity (i.e. the number of bricks laid per minute).7
Virtual reality and augmented reality are also proving useful safety tools
in the construction industry. In site planning, for example, research has
shown that VR can help to improve safety by allowing professionals to visu-
alize conditions and identify possible hazards before construction even
starts.8 Crucially, these more immersive tools can simulate actual site condi-
tions in a virtual environment, which makes them more effective than look-
ing at standard two-dimensional drawings. The same research also found
that visualization technologies were very effective in immersive construction
safety training and education, which makes a lot of sense.
And when construction is underway, AR comes into its own as a safety
visualization tool. Naturally, construction workers can’t be immersed in VR
headsets and blocking out everything around them on-site. Because AR
glasses are transparent, the worker stays aware of and engaged with their
surroundings, while being fed additional data that’s being superimposed
over the real world. So, graphics can be overlayed that show where the wir-
ing is, for example. Or, when a worker sees a real-life hazard sign some-
where on the site, AR glasses can display text to explain what the hazard is
and what safety measures need to be taken.
 Employee safety and wellbeing 161

Keeping people safe in the heat

Particularly for those who work outdoors or do very physical jobs, extreme
heat can present serious risks. Remember the smart hardhats I mentioned?
Engineering company Laing O’Rourke, which operates in the Australian
outback, has used such technology to keep its employees safe in extreme
conditions – smart hardhats fitted with a sweatband sensor that measures
the heart rate and temperature of wearers, as well as the external tempera-
ture around them.9 The data is uploaded from the hardhats to the cloud,
where it’s analysed to look for patterns that suggest a worker might be at
risk of heatstroke. If an employee is in danger, the hardhat itself receives a
sound and vibration alarm that alerts the worker to take a break in the
shade. Such heat sensors could be used in a number of different settings,
from fruit farms or vineyards to construction sites.

Looking at the link between connectivity, employee

safety and productivity
While it’s clear that increased connectivity of both workers and machines
can help to dramatically increase safety, it can also significantly boost pro-
ductivity. Think about the factory machinery example outlined earlier in the
chapter – if machinery or safety systems are not being used in the way
they’re intended, this can lead to earlier failure or extended shutdown for
unscheduled maintenance, which obviously impacts productivity. Detailed
insights on safety-related issues can improve troubleshooting and resolve
downtime issues much faster, and even prevent them from happening in the
first place through improved staff training.
The same is true of connected individual workers. We’ve already seen in
this chapter how connectivity helped improve bricklayers’ productivity by
17 per cent. Other studies have shown that connected workers are typically
around 8 per cent more productive.10

The cutting edge of workplace safety:

VR and AR
We caught a quick glimpse of how VR and AR can be used to enhance safety
on construction sites. But how else are these technologies being used to im-
prove employee safety?
162 Data-driven and AI-enabled HR in practice

One promising area is the use of VR to provide immersive, highly realistic

training simulations for employees. Such training could be used to help
workers identify hazards, safely operate machinery and even respond to
emergency scenarios (as we saw in the BP example from Chapter 10). In one
example, Boeing developed a VR training programme that allows employees
to experience certain emergency scenarios, such as a cabin decompression –
and the immersive training has reportedly helped the company reduce train-
ing time by up to 50 per cent.11
But safety training doesn’t have to be fully immersive to be effective. AR,
which overlays information over whatever the user is seeing in the real
world, can also prove very useful in training – by providing employees with
safety instructions in real time as they go about their work. The Nsflow
platform, for example, works in conjunction with QR scanners, allowing
employees to access AR instructions whenever they need them. And if you
think VR and AR training only applies to those working in hazardous or
highly technical scenarios, think again. Nsflow says its AR training can be
applied in all sorts of settings, including offices, clinics, educational insti-
tutes and warehouses.
Think of AR training as, essentially, 3D training – a more interactive way
of learning safety protocols step by step – and it’s easy to see how the tech-
nologies could be applied in a wide range of settings. That’s the idea behind
Vection Technologies’ EnWorks AR training solution, which creates three-
dimensional training manuals that are designed to replace paper manuals.
EnWorks can also be used to create mock-ups of the work environment,
allowing employees to practise safety procedures. And when carrying out
their jobs, Vection Technologies says EnWorks can provide real-time guid-
ance and reminders, such as identifying potential hazards and displaying
instructions.
Circle back to Chapter 10 for more on the use of AR and VR in training.

Improving employee wellbeing

and wellness
As well as ensuring working environments are safe, sensors are also com-
monly used to ensure workplaces are pleasant environments to be in – think
temperature sensors, windows that open automatically to control ventila-
tion, and so on. This sort of technology is commonplace in many organiza-
tions, so in this section we’ll focus on some of the newer ways in which
 Employee safety and wellbeing 163

companies are looking after their employees. Much of this focuses on em-
ployee health or wellness, and how many organizations are providing data-
driven employee wellness programmes.

Why is wellness important?

It makes sense that the healthier employees are and the happier they are, the
better they will perform for the company. Some of the most common work-
related illnesses include mental health issues (like stress and anxiety) and
musculoskeletal problems (such as back pain), and these health issues can
cost companies dearly through employee absence and lost productivity.
In this landscape, wellness programmes are becoming more and more
popular among employers, in an effort to encourage employees to be health-
ier – and therefore happier. Such programmes are not just about reducing
absence, though; wellness programmes have also been shown to boost em-
ployee engagement, performance and retention.12
At an individual level, wearable fitness tracking bands such as Fitbit are
increasingly being offered to employees either for free or at a subsidized rate
in order to help them monitor their activity levels and encourage them to be
more active. There’s more on this later in the chapter.
And at the organizational level, HR analytics (see Chapter 6) allows em-
ployers to analyse data on their wellness programmes – such as data from
wearable devices or responses to pulse surveys – to help them better design
and manage aspects of employee wellness. For example, if pulse surveys
highlight that one aspect of a wellness programme has a take-up that’s lower
than expected, the company can either modify and improve that part of the
programme or replace it with something new.

Looking after your employees’ mental health

AI is also playing an increasing role, particularly when it comes to employ-
ees’ mental health – a good example being the use of pulse surveys and
sentiment analysis to identify how employees are feeling. This sort of tech-
nology, particularly sentiment analysis, can even pinpoint signs of stress,
depression or anxiety in employees.
With 83 per cent of US workers suffering from workplace stress, and one
million Americans missing work every day due to stress, this should be an
especially big concern for employers.13 Stress costs companies dearly in
terms of absenteeism, lost productivity and employee engagement.
164 Data-driven and AI-enabled HR in practice

Again, AI analytics of data from pulse surveys can play a huge role in
enhancing employees’ mental health, by helping you gauge how happy and
satisfied your workforce is. Not only can this help to inform your workplace
culture, it can also help you design better, more thoughtful wellness pro-
grammes.
In one project in Finland, AI is being used to boost employee wellbeing
through an app that asks employees questions about their wellbeing. The
answers are then anonymized, meaning employers can use the data to create
targeted, preventative wellbeing tools – and employees feel comfortable giv-
ing honest answers. One employer participating in the project said the tool
resulted in employees taking fewer sick days.14
As we’ll see later in the chapter, AI is even capable of predicting potential
mental health issues.

Improving physical health with the IoT

As well as identifying when employees are in physical danger, or suffering
from stress or anxiety, technology is now able to help employees lead health-
ier, more active lives. The IoT has played a huge role in this – for example, how
many people do you know who wear a fitness tracking band or use an app on
their phone to track their activity or number of steps a day? A lot, I bet.

Reducing the risk of back problems

I was shocked to learn that 12.5 per cent of all sick days in the UK are down
to back pain.15 Increasingly, this can be attributed to how many of us sit at
a desk all day. We’re simply not designed to sit for seven hours a day – even
if we do sit correctly with perfect posture the whole time. Most of us strug-
gle to maintain great posture all day – come to think of it, most of us aren’t
even aware of our posture a lot of the time. Yet poor posture can have seri-
ous long-term health effects and shouldn’t be overlooked.
While workstation risk assessments and ergonomic products like back
supports and footrests go some way to protect employees against back
problems, it’s clear that more could be done. Part of the solution may lie in
IoT. We’ve already seen many examples of how products and people are
becoming increasingly connected – now even your office chair has under-
gone an IoT makeover. Even with a super fancy ergonomic chair, it’s still
possible to sit badly because we’re generally not aware of our posture while
we’re busy working away. Now, smart office chairs are being designed with
sensors that register the user’s posture and make them aware (through
 Employee safety and wellbeing 165

v­ ibrations) when they’re sitting incorrectly. The idea is that, by increasing

awareness of posture throughout the day, employees can modify their pos-
ture as needed and avoid back problems in the future.

The role of fitness tracking bands and smart watches

The IoT is also impacting employee health and wellness in more obvious
ways: wearable fitness tracking bands and smart watches. These are increas-
ingly becoming part of corporate wellness programmes around the world,
including at companies like BP, Bank of America and Target. Fitbit’s corpo-
rate wellness offering, for example, includes a suite of tools and resources
for employers, including dashboards to monitor how employees are doing.
And the trackers themselves do far more than encourage staff to get up and
walk more; Fitbit claims they also increase engagement in wellness pro-
grammes and improve health outcomes. And, particularly in the US, these
trackers are also being used to reduce health insurance costs by allowing
employers to leverage employee health and activity data to negotiate with
insurers.
This isn’t to say that you must rush out and buy hundreds of fitness track-
ers or smart watches for your staff, but it does point to how employees are
willing to engage with IoT-enabled wellness programmes. As more and more
people are investing in their own fitness tracking bands, and as mobile apps
are increasingly offering similar capabilities to track activity and other
health metrics, more and more wellness programmes will leverage these de-
velopments to their advantage.

Delivering personalized recommendations

for employees
As we’ve seen throughout this book, AI is great for providing personalized
recommendations and solutions. Employers are now beginning to use this to
deliver recommendations regarding employees’ health and wellness. Indeed,
AI has the ability to design personalized wellness programmes that are tai-
lored to individual employees – for example, by categorizing employees ac-
cording to their wellness preferences, and helping people set and achieve
their own wellness goals.
This can even extend to recommendations regarding healthcare. In
Chapter 5 I briefly mentioned that Walmart has begun using AI to help em-
ployees find the right medical provider. How does this work? Walmart has
166 Data-driven and AI-enabled HR in practice

partnered with AI company Health at Scale, which uses AI to give patients

personalized recommendations on which provider is likely to provide the
best outcomes for them, based on their health needs and medical history.16
Essentially, the system uses AI to identify healthcare providers who have
successfully treated patients with similar needs. Walmart has rolled out the
system to workers (and their families) in certain locations, with the tech in-
tegrating with Walmart’s existing health plan administrator’s search engine
and virtual care referrals.

The role of predictive analytics in wellness

The next level in AI-enabled employee wellness is using predictive analytics
to pre-empt physical and mental health conditions, and help employees get
the help they need.
A great example of pre-emptive mental healthcare comes from the AI app
Kintsugi, which uses vocal biomarkers to detect signs of depression and
anxiety based on just 20 seconds of free-form speech.17 The app – named for
the Japanese artform of repairing broken pieces of ceramic with gold
enamel – is language-agnostic, which means it can be used globally, irrespec-
tive of the language, dialect or accent. The app aims to offer a diagnosis,
where appropriate, plus a treatment plan.
Depending on where you are in the world, accessing mental health ser-
vices can be a real struggle, so apps like this – offered as part of a corporate
wellness offering – could be a lifeline for employees who want to better
manage their mental health. That said, there are pitfalls around predicting
health issues. More on that coming up…

The cutting edge of wellness: Tapping

into VR and the metaverse
I believe we will increasingly see employers use immersive technologies to
aid employee wellness. It makes sense when you think about it, since VR can
be used to create immersive meditation, mindfulness and relaxation experi-
ences. So why not incorporate this into a wellness offering to help employ-
ees reduce stress and feel calmer?
This could prove especially valuable for those working in fast-paced,
high-stress jobs. In one study, clinicians working in the UK’s National Health
Service were offered the chance to participate in a relaxing VR session
 Employee safety and wellbeing 167

­ uring the working day. And those who did reported significantly increased
d
feelings of happiness and relaxation. It even helped to lower participants’
heart rate.18
VR seems a natural fit with wellness. But what about the metaverse? As
more and more work activities take place in immersive metaverse environ-
ments (like onboarding new employees in the metaverse, see Chapter 8), it’s
worth considering the potential applications of the metaverse in employee
wellness.
In particular, virtual spaces can help employees – especially those who
work remotely – feel more connected to their colleagues. And this can have
a huge impact on employee satisfaction and happiness. So, in the future,
employers might create virtual hangout spaces where employees can gather
and share those little ‘water cooler’ moments. Social events can be held in
the metaverse to help foster team spirit. And regular team meetings can be
held in immersive virtual spaces, as opposed to a Zoom call. Imagine, for
example, gathering your team in a luxurious virtual meeting room with
stunning views over a green valley or dramatic mountains. By making eve-
ryday aspects of work more immersive, the metaverse may help to enhance
employee wellbeing.

The potential downsides of data-driven

employee safety and wellness
It’s clear that we can monitor increasing amounts of data about employees’
activities and health. The question is, how much monitoring is too much?
Particularly when it comes to employees’ health data, this is obviously highly
sensitive and personal data and employers need to tread carefully and act in
an open, transparent way (see Chapter 4).

Remember, health data is valuable data

Shockingly, health data is reportedly 10 times more valuable on the black
market than credit card data.19 Large batches of personal health data are
especially valuable because they can be used for medical fraud – and because
medical fraud is typically slower to detect than, say, banking fraud, it’s a far
more tempting prospect for criminals than your bank details. Plus, health
data is frequently not as well protected as credit card data or other obvious
sources of fraud, which makes it an easier target. While this is more of an
168 Data-driven and AI-enabled HR in practice

issue for healthcare providers – who are often operating on old legacy com-
puter systems in desperate need of an update – it still needs to be considered
by employers who are working with employee health data. It’s imperative
you guard this valuable employee data with the same level of protection as
you would your customer data.

How much should employers know about their

employees?
We know that tracking employee data can have a whiff of Big Brother about
it, and sometimes employee scepticism is warranted. One unsavoury report
showed how employers were already using data to identify when employees
might be pregnant or considering becoming pregnant, before those employ-
ees were ready to divulge the news to their employer.20 Not only is this an
invasion of privacy, it also opens up the potential for employers to slyly
discriminate against pregnant or soon-to-be-pregnant women (such as over-
looking them for promotion) before they’ve been officially informed of the
pregnancy. The argument in favour of such systems is that it can help em-
ployers identify when an employee may be at risk of health issues.
But the thing is, employees may not be comfortable with their bosses
knowing how fit (or not) they are, or having the ability to identify when they
might be at risk of health issues. Back in 2012, Ohio healthcare provider
The Cleveland Clinical announced that employees who were overweight or
at risk in other ways (such as being a smoker) who didn’t join the company
wellness programme would have to pay more for their health insurance –
over 20 per cent more, in fact. And those that did join the programme but
didn’t meet health targets set by programme administrators for them also
saw their premiums rise by almost 10 per cent.21 Practices like this are likely
to alienate employees and send morale plummeting.

Navigating these challenges

The challenge for HR teams is therefore to encourage participation in well-
ness programmes and use data to help employees live healthier lives (which,
in turn, financially benefits the company) without making employees un-
comfortable.
One option is to only use aggregated data, not the ability to drill down to
into individual activities. Another is to ensure wellness programmes are vol-
untary – because, when participation in wellness programmes becomes
 Employee safety and wellbeing 169

mandatory, or when employees feel their health data may be used to punish
them in some way, buy-in for wellness programmes take a nose-dive. And
rather than punishing employees who don’t participate – by, for example,
putting up their health insurance premiums – it’s far better to offer employ-
ees lower premiums for taking part. In other words, offering an incentive for
employees to take part, but not disadvantaging employees who don’t wish
to participate.
Finally, we should acknowledge that there are many non-data ways to
facilitate employee wellbeing, such as providing on-site exercise facilities or
discounted (or free) gym membership, and serving up healthy food in the
canteen. Data and AI should never replace good practices like these. But,
used well, data and AI can give HR teams precious insights into how to
manage and improve employee wellbeing and safety.

Key takeaways

I think employee safety and wellbeing is one of the most exciting and
fast-developing areas of data-driven HR. Key points from this chapter are:

●● Data- and AI-driven technologies, especially wearable technologies and

sensors, are making workplaces safer and more comfortable places to
be – from construction sites and factories to regular offices.
●● Because IoT devices can transmit real-time data for on-the-fly analysis,
managers can be alerted when unsafe practices are taking place and
take appropriate action.
●● While increased connectivity of both workers and machines can help
dramatically increase safety, it can also significantly boost productivity.
●● VR and AR also have an important role to play in workplace safety,
particularly when it comes to delivering better, more immersive, more
engaging safety training – or even step-by-step safety instructions in
real time.
●● When it comes to wellness, data and AI can play a role in improving
employees’ mental and physical health – for example, by creating better
feedback loops, offering employees wearable fitness trackers, and
delivering AI-driven personalized health recommendations.
●● VR may also help to enhance your wellness offering, especially in terms
of lowering stress.
170 Data-driven and AI-enabled HR in practice

●● It’s really important to take proper precautions to protect employee

health and wellness data. Also, remember that employees may be
uncomfortable with their bosses knowing how fit (or not) they are. The
challenge for HR teams is to encourage participation in wellness
programmes and use data to help employees live healthier lives without
making those employees uncomfortable.

That brings us to the end of Part Two. Now that we’ve got a good idea of
how organizations can use data and AI to deliver more thoughtful, efficient
HR services, how can you put these ideas into practice? Turn to Part Three
to find out…

Notes
1 D Hardawar. Humans trust this emergency robot more than common sense,
Engadget, 29 February 2016. www.engadget.com/2016-02-29-humans-trust-
this-emergency-robot-more-than-common-sense.html (archived at https://perma.
cc/4CKQ-JW99)
2 Health and Safety Executive. Health and safety statistics: Key figures for Great
Britain (2021/2022), Health and Safety Executive. www.hse.gov.uk/statistics
(archived at https://perma.cc/NSK5-RXCA)
3 G Schultz. The Holy Grail of safety, EHS Today, 6 February 2014. www.
ehstoday.com/safety-leadership/article/21916109/the-holy-grail-of-safety-a-
single-allencompassing-safety-leading-indicator (archived at https://perma.cc/
L2QY-ZR6X)
4 Science Daily. Electronic nose could aid in rescue missions, Science Daily,
23 July 2014. www.sciencedaily.com/releases/2014/07/140723110403.htm
(archived at https://perma.cc/BG8T-ZG8C)
5 L Kelion. Caterpillar backs eye tracker to combat driver fatigue, BBC News, 28
May 2013. www.bbc.com/news/technology-22640279 (archived at https://
perma.cc/94ZS-74C9)
6 Health and Safety Executive. Work-related fatal injuries in Great Britain 2022,
Health and Safety Executive, 23 November 2022. www.hse.gov.uk/statistics/pdf/
fatalinjuries.pdf (archived at https://perma.cc/9F7A-QYX6)
7 S Smith. IOT: Reducing back injuries and costs, improving productivity, EHS
Today, 18 April 2016. ehstoday.com/construction/iot-reducing-back-injuries-
and-costs-improving-productivity (archived at https://perma.cc/CU7C-RQFE)
 Employee safety and wellbeing 171

8 S Azhar. Role of visualization technologies in safety planning and management

at construction jobsites, Procedia Engineering, 2017, 215–26, doi.org/10.1016/
j.proeng.2017.01.329 (archived at https://perma.cc/4835-NA6X)
9 A Chanthadavong. Laing O’Rourke monitors workers’ safety with a smart
hardhat, ZDNet, 17 November 2015. www.zdnet.com/article/laing-orourke-
monitors-workers-safety-with-a-smart-hardhat (archived at https://perma.cc/
ND5P-NTLZ)
10 Toppan. The role of wearable technology in the workplace, Toppan, nd.
toppandigital.com/translation-blog/role-wearable-technology-workplace
(archived at https://perma.cc/CX5M-JULK)
11 Ironyun. How AI can improve workplace safety, Ironyun, 20 May 2022. www.
ironyun.com/blog/how-ai-can-improve-workplace-safety (archived at https://
perma.cc/8RYQ-6S25)
12 S Miller. Wellness programs as an employee retention tool, SHRM, 20 January
2010. www.shrm.org/resourcesandtools/hr-topics/benefits/pages/wellness_
employeeretention.aspx (archived at https://perma.cc/4YLR-8YFV)
13 The American Institute of Stress. Workplace stress, The American Institute of
Stress, nd. www.stress.org/workplace-stress (archived at https://perma.cc/
KT6W-VDDD)
14 European Commission. Using AI to boost productivity – and employee
wellbeing, European Commission, European Social Fund Plus, 28 April 2022.
ec.europa.eu/european-social-fund-plus/en/projects/ai-boosts-productivity-and-
employee-wellbeing (archived at https://perma.cc/CS3B-YHGY)
15 G Wynne-Jones et al. Absence from work and return to work for people with
back pain, Occupational and Environmental Medicine, 2014, 71, 448–56.
oem.bmj.com/content/71/6/448 (archived at https://perma.cc/FCV6-JZ8S)
16 R Torrence. Walmart teams up with Health at Scale to help its workers find
the right doctor, Fierce Healthcare, 31 January 2022. www.fiercehealthcare.
com/retail/walmart-teams-health-scale-help-its-workers-find-right-provider
(archived at https://perma.cc/XXM5-QHEP)
17 A McBride. How AI is transforming employee wellness benefits, VentureBeat,
14 February 2023. venturebeat.com/programming-development/how-ai-is-
transforming-employee-wellness-benefits (archived at https://perma.cc/
T5R5-2YNC)
18 J S Adhyaru and C Kemp. Virtual reality as a tool to promote wellbeing in the
workplace, Digital Health, 4 March 2022. journals.sagepub.com/doi/full/
10.1177/20552076221084473 (archived at https://perma.cc/E2RZ-2TRC)
19 N Griffin. Patient data ‘10–15 times more valuable than credit card data’, Irish
Examiner, 19 May 2021. www.irishexaminer.com/news/arid-40293149.html
(archived at https://perma.cc/7N7J-SDEL)
172 Data-driven and AI-enabled HR in practice

20 V Zarya. Employers are quietly using big data to track employee pregnancies,
Fortune, 17 February 2016. fortune.com/2016/02/17/castlight-pregnancy-data
(archived at https://perma.cc/E4AY-RENR)
21 S McGee. How employers tracking your health can cross the line and become
Big Brother, Guardian, 1 May 2015. www.theguardian.com/lifeandstyle/us-
money-blog/2015/may/01/employers-tracking-health-fitbit-apple-watch-big-
brother (archived at https://perma.cc/R4NL-ENEQ)
 173

PART THREE
Making data-driven
and AI-enabled HR
happen
Having explored the various HR services and activities in Part Two – and
explored how other organizations are successfully using data and AI – you
might be wondering how to apply what you’ve learned in your own organi-
zation. This part will help you on that journey.
Essentially, these chapters answer the question, ‘So where do I go from
here?’ Together we’ll look at identifying and prioritizing use cases (key data
projects), building the necessary skills and culture for data-driven and AI-
enabled success, and creating the right infrastructure. For me, these are the
foundations of intelligent HR.
174

THIS PAGE IS INTENTIONALLY LEFT BLANK

 175

Identifying the 12
use cases for
your organization
Back in Chapter 5, we talked about creating a data and AI strategy. We
looked at the importance of linking your strategy to the organization’s and
HR function’s strategic objectives and explored the core questions that form
part of any good data/AI strategy. In this chapter we build on that strategic
process and look at how to identify specific use cases/projects for your
­organization.

Why talk about use cases now?

And why not back in Chapter 5, you ask? Well, between Chapter 5 and here,
we’ve explored many ways in which organizations can leverage data and
AI – from delivering better HR insights to improving employee safety and
wellbeing, and a whole lot in between. Your mind might be buzzing with
possibilities. (Great!) Or maybe you feel a little overwhelmed at the thought
of applying all that you’ve learned in your own organization. (Totally un-
derstandable.)
As such, now is a good time to press pause and revisit how you can use
data and AI strategically. How you can best use data and AI to deliver your
HR objectives and contribute to the business’s wider strategic objectives.
Because, yes, there are many possibilities, but they won’t all be relevant or
realistic for your organization. So how do you know which projects to focus
on – and how many to focus on? This chapter will help. Therefore, think of
this chapter as a bridge between strategy and practical implementation.
Naturally, this is intrinsically linked to the fundamental process of creat-
ing a data and AI strategy, so it’s a good idea to cast your eye back over
Chapter 5 once you’ve read this chapter (or now, if you prefer).
176 Making data-driven and AI-enabled HR happen

Identifying potential use cases

In Chapter 5, we briefly looked at the potential ways in which you can use
data and AI in HR. Broadly speaking, these fall into three main categories:
generating better HR insights to aid decision making across the organiza-
tion; delivering a better service for employees and candidates; and driving
efficiencies in HR.
When identifying potential use cases, these three categories provide a use-
ful jumping off point. I also recommended that you look at your biggest HR
and organizational challenges, as these may point to priority projects. And,
of course, you’ll want to look at the organization’s wider corporate objec-
tives to see how you can use data and AI to help deliver the company’s
­mission.
Essentially, you’re trying to identify areas where data and AI could really
make a difference in your organization – whether that’s by solving your big-
gest challenges, helping to drive the organization forward or simply deliver-
ing the best possible HR service.

Start with a brainstorming session

When working directly with clients, I usually recommend starting with a
brainstorming session. The size of the session and number of people in-
volved will depend on your organization – it could be one person or it could
be everyone in the HR team plus other key stakeholders in the organization.
The goal of the session is to identify specific use cases (or you might pre-
fer to call them ‘opportunities’) to put data and AI to work. Once you start
thinking about it, you’ll probably be surprised at how many you can come
up with. Don’t worry about having too many opportunities on your list.
Later on, we’ll talk about narrowing your list down. For now, let your im-
agination loose and try to come up with as many as possible.
That said, it’s a good idea to ensure that the opportunities you identify
serve at least one of the purposes we’ve already talked about, i.e. generating
better HR insights; delivering a better HR service; driving efficiencies in HR;
solving your biggest HR and organizational challenges; and helping the or-
ganization achieve its strategic vision. If an opportunity doesn’t serve one of
these purposes, are you sure it would deliver value for the organization?
Remember, just because something can be done with data and AI, doesn’t
mean it should be done.
 Identifying the use cases for your organization 177

Quick wins and transformational projects

Next, I recommend organizing your list of opportunities into two catego-
ries:

●● The first category is ‘quick wins’ – projects that provide immediate returns
for a relatively low investment. Quick-win projects are great because they
help you hone data and AI skills, build confidence in using these
technologies and ultimately create buy-in for bigger initiatives. Speaking
of which…
●● The second category is ‘majorly transformational’ or ‘strategic, longer-
term’ projects – those game-changing applications of data and AI that
may require a bigger upfront investment and longer development period,
but which can create a fundamental change in the way HR serves the
organization (or even a fundamental change for the organization itself).

Narrowing down your priorities

You’ve found where the opportunities lie and categorized them according to
quick wins vs. transformational projects. Now you need identify your top
priorities.
The truth is, some of the ideas you’ve come up with during brainstorming
are likely to be completely unrealistic, or too technical, expensive or difficult
to do right now, and that’s fine. Up until now, the exercise has been focused
on identifying opportunities and, by doing so, understanding how and
where data and AI could be useful. That in itself is a valuable process. But
since you have to be realistic, you’ll need to create a shortlist of your top
priorities – the use cases which are most viable to put into operation and
will deliver the most value.
In my experience, the teams and organizations that are most successful at
leveraging data and AI tend to be those that start out with a small but highly
focused number of use cases – use cases that all closely align with strategic
goals. If you pick too many, it can quickly become unmanageable due to the
number of variables and the large number of stakeholders involved.
Particularly within organizations where there is perhaps not widespread
buy-in for digital transformation, attempting too many use cases at once can
lead to overwhelm, disappointment and an ongoing lack of engagement (the
old ‘I knew it was a waste of time, and I was right’ attitude).
Another danger of attempting too many initiatives all at the same time is
that it can be difficult to determine exactly what is affecting what! You can
178 Making data-driven and AI-enabled HR happen

easily get to the point where one initiative is impacting metrics associated
with another initiative, and you might not know whether change is happen-
ing for the reasons you expect, or whether it’s happening because of some-
thing completely unrelated.
Clearly, you need a shortlist. But how many opportunities should make it
onto that list? Generally, I think it’s a good idea to prioritize between one
and three ‘major’ use cases, depending on the size of your organization and
budget, and one to three ‘quick wins’. It’s important to have a spread across
both categories – meaning you need the ‘quick wins’ to help build trust in
what you’re doing, and you need the more majorly transformational use
cases to demonstrate long-term value.

Fleshing out your use cases

For each use case that you’ve prioritized, you’ll now want to answer the fol-
lowing questions. (By the way, I have a handy ‘use case template’ with all
these questions on my website, bernardmarr.com – and it’s free to down-
load.) Apply this list of questions (or fill out the template) for every use case
you’ve prioritized.
The questions are:

1 How does this use case link to a strategic goal? Because intelligent HR
is all about using data and AI to solve problems, reach your goals and
help drive the organization forward.
2 What is the objective of this use case? Back in Chapter 5 we talked about
identifying your most pressing unanswered business questions – those
questions you need to answer if you’re to achieve your strategic goals.
You’ll probably find that answering those questions will be the objective of
your chosen use cases (or be very closely linked to the objective).
3 How will you measure the success of the use case? There’s not much
point doing anything with data and AI if you can’t tell whether you’ve
been successful or not. This means you will have to define what success
looks like, and identify the indicators that you can use to measure success.
4 Who will be the use case owner? Every use case needs someone who will
take overall responsibility for ensuring that it can be delivered. Someone
who’s responsible for making sure it is planned and implemented at a
tactical level. Of course, particularly in a smaller organization, there may
not be enough people involved for every use case to have a different
owner, so one person might take responsibility for several use cases.
 Identifying the use cases for your organization 179

5 Who will be the data customers? These are the people that will monitor
and act on the insights discovered by the use case. They could be HR
personnel, or people in other business functions (department managers,
for example). It’s important to understand who these people are to
ensure that the use case and resulting insights can be communicated in
a way that will be actionable by the people that matter – the data
customers.
6 What data do we need? At this point, you need to think about where
you will get the information needed to answer the key business questions
and achieve the use case objective. You may find that you already have
(or can get access to) all the information you need within the organization
itself. On the other hand, you might identify knowledge gaps that need
to be filled by looking outside of the organization to external data
sources.
7 What data governance issues need to be addressed? As we’ve seen
throughout this book, data and AI are hugely powerful tools for driving
business change, but they also bring important challenges around
governance, compliance and regulation (Chapter 4). A wrong step here
can have serious repercussions – and not only in a legal sense. Privacy,
consent and data bias will be key considerations here.
8 How do we analyse the data and turn it into insights? Once the data
has been collected, you need to turn it into insights. For that, you need
to choose the right analytics techniques (text analytics, for example).
9 What are the technology requirements? Here, you start to identify the
specific technology infrastructure requirements (more on this coming
up in Chapter 14). For ‘quick win’ use cases, particularly in smaller
organizations with limited budgets, a cloud-based HR analytics platform
may well deliver everything you want. For larger projects, you may need
a more specialist solution.
10 What skills and capabilities do we need? Next, you have to assess the
human skills and resources needed to get the job done. (This is the sub-
ject of Chapter 13.) You may or may not have these capabilities in
house. If not, you have three main options – upskill the existing work-
force, hire new people with the skills you need or outsource the project
to external partners and service providers.
180 Making data-driven and AI-enabled HR happen

11 What are the issues around implementation we need to be aware of?

The final point to consider at this stage is what practical considerations
might prevent you moving your use case from idea, to plan, to execu-
tion? In particular, you’ll want to make sure there are clear lines of com-
munication between all involved parties, but specifically between the
people whose job it is to pull out the insights, and the people whose job
it is to take action on them.

Linking your use cases to your wider data/

AI strategy
Some of the above questions will be familiar because they featured in
Chapter 5. In other words, Chapter 5 helps you focus in on your overarch-
ing objectives/challenges/unanswered questions, but many of the same stra-
tegic questions also apply to each individual use case.
If you haven’t yet created your overarching data and AI strategy, that’s
fine. Because you can in fact use your fleshed-out use cases to form the
framework of your strategy. After all, each priority use case aligns with an
HR goal or challenge – and your strategy is all about delivering your goals
and solving your problems. What’s more, by fleshing out your use cases,
you’ve gained an understanding of data requirements, infrastructure impli-
cations, and so on – and you’ll probably find, having been through the pro-
cess of fleshing out your use cases, that there is some crossover between the
use cases in terms of data, skills required, governance issues, etc. There will
be synergies and potential efficiencies that could feed into your strategy as a
whole. This is particularly true with governance considerations, as these will
typically apply across all of your use cases.
And if you have already created an initial data and AI strategy, now is a
great time to revisit it, with your priority use cases in mind. This is the per-
fect opportunity to look at how your use cases affect your strategy as a
whole. And how your strategy as a whole ties in with your use cases.
Now is also a good time to think about budgeting. If one of your use
cases is likely to eat through a large portion of your budget, you might strug-
gle to get everything done. For this reason, you should think about budget-
ing for your priority use cases as a group, rather than as individual projects.
 Identifying the use cases for your organization 181

Learning from other businesses that have

been on the data and AI journey
Let’s revisit a couple of examples from this book and re-examine them from
a use case perspective.

The huge multinational

I referenced Shell in Chapter 3 and talked about how the company’s
GameChanger unit partnered with video game start-up Knack to identify
the key characteristics of top idea generators within the business. Let’s pre-
tend we were in the room when Shell first identified this data opportunity.
In this case, the underlying objective was to help the company more easily
identify the best ideas among those submitted to the GameChanger unit –
thus ensuring that great ideas don’t get overlooked and that those working
in the GameChanger unit can focus their time and attention on ideas that
are likely to deliver the best value. You could say their unanswered ques-
tions here were, ‘What characteristics unite our best idea generators?’ and
‘How can we assess people for those characteristics?’ The strategic link is
crystal clear.
Just as an aside, I’ve worked with Shell on their AI strategy, and it’s im-
portant to note that, even though Shell is a large organization with impres-
sive resources, we still limited ourselves to no more than three
transformational use case priorities, plus a couple of quick wins. Because
even large companies (some would say especially large companies) run the
risk of getting caught up in the excitement of data and AI, and losing sight
of what they’re trying to achieve. So it goes to show that if a huge organiza-
tion like Shell can drill down to a handful of priority use cases, you can too.

How a tech trailblazer got it wrong

Amazon has placed data and AI at the very core of everything it does: intel-
ligent products (such as Alexa and Echo), intelligent services (for example,
Amazon’s personalized customer recommendations) and intelligent business
operations (such as warehouse operations that are highly driven by data).
Yet, in Chapter 4, we saw how Amazon had to stop using a machine
learning algorithm that was designed to assess job applicants, after discover-
ing that it was, well, sexist. No doubt this use case started with the best
182 Making data-driven and AI-enabled HR happen

strategic intentions – to save recruiters time, to provide a more efficient HR

service to candidates and perhaps even (ironically) to reduce the risk of
human bias in the recruiting process. But, clearly, where this use case fell
short is in the governance and implementation considerations. When flesh-
ing out this potential AI opportunity, did the possibility of data bias come
up for discussion? When looking at the potential implementation pitfalls,
what mechanisms were considered to ensure the algorithm was fit for pur-
pose before being rolled out? Were the measures of success perhaps flawed,
in that they focused more on efficiency as opposed to the quality and diver-
sity of candidates?
We can’t answer those questions for certain. But it goes to show that even
extremely tech-savvy businesses have had to navigate the process of becom-
ing an AI-enabled, data-driven organization – and they don’t always get it
right. We can all learn from this, by which I mean pay close attention to the
governance issues, measures of success and potential implementation road-
blocks.
This use case aside, one thing that really impresses me about Amazon is
how seamlessly different divisions have got behind data and AI – from the
ecommerce side of things, to its AI-as-a-service offerings, to the fulfilment
operations. Building a culture of data and AI is crucial to success in intelli-
gent HR. But more on that coming up in the next chapter…

Key takeaways

To summarize the key points on identifying use cases in your organization:

●● When it comes to identifying potential data and AI opportunities, start

with a simple brainstorming session. The goal of the session is to identify
specific ways (or opportunities) to put data and AI to work.
●● Next, categorize your use cases according to quick wins and major
transformational projects. You need a mixture of both, because quick
wins help to build trust and demonstrate the value of data and AI (in a
relatively quick and inexpensive way), while the transformational
projects bring huge value for the long term.
●● Now it’s time to identify your top-priority projects – as in, those that are
most viable and/or deliver the most value. It’s a good idea to prioritize
between one and three major use cases, depending on the size of your
organization and budget, and one to three quick wins.
 Identifying the use cases for your organization 183

●● If you haven’t yet created your overarching data and AI strategy, your
use cases can form the framework of that wider strategy. And if you
have already created your data and AI strategy, now is a great time to
revisit it, with your use cases in mind. After all, there will be issues and
considerations that are common across all use cases.

Whatever your use cases, a lack of data and AI skills can be a major barrier
to implementation, especially in smaller organizations. So let’s explore this
topic of skills and competencies. In the next chapter we’ll see how HR teams
can prepare themselves for the intelligence revolution, overcome the data
skills shortage and build a culture of data and AI.
184

Building skills 13
and aligning
culture
Back in Chapter 10 we talked about essential future skills for the work-
force – emphasizing soft skills just as much as (if not more than) technical
or digital skills. But since this journey of data-driven and AI-enabled HR
obviously requires certain data and AI skills, let’s dwell a little on that topic.
How can you build (or tap into) the right technical skills for intelligent HR –
especially bearing in mind the huge global shortage of digital skills? And
how can you create a culture that understands the value of data and AI?
We’ll explore those questions in this chapter.
But first, as an HR professional, what sorts of skills do you personally
need to succeed? Let’s find out.

What skills do HR professionals need?

The same 20 future skills from Chapter 10 clearly apply to HR professionals
as much as the wider workforce. But, specifically for HR roles, I believe the
following skills and characteristics are absolutely vital for success:

1 Communication – HR professionals must be good communicators, espe-

cially in this era of rapid transformation, rising automation and continu-
al education. It’s vital you can communicate well with others so that you
can both support people through change and avoid misunderstandings.
And we mustn’t forget that listening is a core communication skill that
will serve any HR professional well.
2 Empathy – The ability to put yourself in someone’s shoes is essential for
providing an excellent HR service, and helping the organization manage
change successfully.
 Building skills and aligning culture 185

3 Discretion – HR professionals routinely deal with sensitive, highly per-

sonal information, and encounter difficult situations (such as harass-
ment). This requires the ability to be discreet, to communicate carefully
and sensitively and to keep confidential information private.
4 Ethics – Similar to discretion, in that behaving ethically means keeping
sensitive information private, and so on. But, more than that, modern HR
requires you to have a good understanding of ethics and ethical chal-
lenges – especially around the changing nature of work, the role of new
technologies in the workplace and being able to harness data and AI
without alienating employees.
5 Organizational skills – Now more than ever, there’s a lot of information
and responsibility flying around the average HR role. From keeping track
of evolving regulations and compliance issues, to helping the workforce
cope with change – not to mention providing an awesome day-to-day HR
service – today’s HR professionals have to be organized and skilled multi-
taskers.
6 Problem solving – Aside from the major challenges that digital transfor-
mation brings to the workplace, HR teams face perennial challenges such
dealing with employee conflicts, or how to identify and attract the best
talent. Excellent problem-solving skills are therefore a must.
7 Decision making – Oh boy, do HR professionals have to make some
tough decisions! Decisions that can affect people’s lives in major ways.
Despite this, you must be comfortable grappling with difficult decisions
and finding the best solutions. As we’ve seen throughout this book, data
can play a huge role in improving decision making. Which brings us to…
8 Data literacy – Intelligent HR means working with data on a daily basis.
Using data to inform decisions, drive HR efficiencies and provide a better
HR service. It’s therefore really important that all HR professionals are
comfortable working with data (and keeping data safe). But let me add a
caveat here – because being data literate should not come at the expense
of the more human ‘soft’ skills like communication and empathy.

In other words, being an effective HR professional in this era of rapid trans-

formation requires a blend of digital and soft skills. And that’s true regard-
less of your specialism within the HR field. Even if your role is heavily
focused on HR analytics, for example, you still need to be an excellent com-
municator, empathetic, and so on. Or let’s say you specialize in training and
development – you still need to be data literate.
186 Making data-driven and AI-enabled HR happen

Data literacy and HR

Since data underpins everything about intelligent HR, let’s delve into this
notion of being data literate.

What does data literacy look like, and why does

it matter in an HR context?
In essence, data literacy is the ability to understand, work with and make
sense of data. It includes the ability to think critically about what the data is
highlighting, to ask questions about where the data comes from and how
reliable it is (considering potential bias, and so on). And, perhaps most im-
portantly of all, it includes the ability to apply data in your work and turn
insights from data into action – be that through your own decision making
or by communicating the data to other decision makers in the business.
A data-literate HR professional understands that data is one of the most
precious assets an organization has. And that data can add serious value for
the organization. In today’s highly digital world – where almost everything
we do can be measured – this ability to work with data is more important
than ever. Because, as we’ve seen throughout this book, data can lead to bet-
ter decisions (i.e. decisions that are based on hard evidence as opposed to
gut instinct or what’s worked in the past), deliver new efficiencies in HR
(using HR chatbots, for instance), and help to enhance HR services (such as
better recruitment, or learning and development).

How can HR teams develop data literacy?

The first step is to be open to working with data. Don’t be afraid of data,
basically. Even if you’re not a typical ‘numbers person’, embrace the idea
that data can enhance the work that you do.
It’s well worth investing in data literacy training – either training that’s
developed specifically for your organization or the HR function, or more
general data literacy training that’s offered by any of the major learning
platforms. (I give some tips on data courses later in the chapter.)
If you already have an HR analytics tool, get stuck in and experiment
with it. Get familiar with using the tool, interrogating the data, and creating
outputs (visualizations, reports, etc.). If you don’t yet have an HR analytics
tool, you can polish up your Excel skills – although an HR analytics p­ latform
 Building skills and aligning culture 187

will obviously help you work with data in more advanced, and often more
intuitive, ways. Thanks to these new analytics tools, it’s becoming much
easier to interrogate data – so you don’t need to be a statistician or data
scientist to work effectively with data.
You should also start using data as the basis of team discussions. In HR
meetings ask questions such as ‘What does the data say about this?’ or ‘What
evidence do we have to support that?’ Making data a part of everyday dis-
cussions in this way ensures the whole HR team views data as a key business
asset.

Working with internal data experts

If your organization has a central data team or analytics function in house,
connect with them and ask them for the support you need. Indeed, involving
in-house specialists, if you have them, is a vital part of creating your data/AI
strategy and fleshing out your use cases (see Chapters 5 and 12). But you can
also connect with your in-house experts in less formal ways, for example by
inviting them to talk about their work at an HR team meeting, or just asking
them questions about their work.
Wherever possible, I think it’s really important have a central data or
analytics group within the organization – people whose job it is to immerse
themselves in the data, but work closely with other teams to support their
needs. (This is often referred to as the ‘hub and spoke’ model in data terms,
where the central data team is the hub responsible for the organization’s
data platforms and tools, while the spokes are the various business func-
tions, each of whom have their own data and data needs.) In large HR teams
you may even want to recruit your own data scientist or data group that sits
within the HR function.
That said, I recognize it’s not possible for every organization to have a
central data team, or for HR functions to have their own data specialist –
and we’ll talk more about that later in the chapter. In such cases, it’s espe-
cially important that you work to improve data literacy among everyone in
the HR team and the organization as a whole. I’m not talking about becom-
ing data scientists yourselves. Rather, I’m talking about being comfortable
working with data and using data in your everyday work.

Raising data literacy across the organization

HR must also play a vital role in raising data literacy in the wider work-
force. This starts with establishing current levels of data literacy in the
188 Making data-driven and AI-enabled HR happen

­ rganization – for example, are teams and individuals routinely using data
o
to back up their decisions? Knowing where the organization is right now
gives you an idea of where people may need to boost their data literacy
skills.
It’s also really important to communicate why data literacy is essential
for your organization’s success. As with any new initiative, when people
understand the ‘why’, they’re more likely to support it.
Of course, it’s likely that employees will need some training on the impor-
tance of data and how to use the data. Again, this can be a purpose-built
training solution for your organization, or you can tap into one of the many
courses already on the market. As part of this, teams must be taught how to
think critically about data. Just as in the HR team, every employee should
know to ask questions such as ‘How was this data collected?’, ‘What can be
learned from this data?’ and ‘How reliable is this data?’ Employees must
also be trained on the safe and ethical handling of data.
And, naturally, people must also have access to the data they need to do
their jobs. It sounds obvious, I know, but too often I see teams that simply
can’t access the information they need on a daily basis. While this is beyond
the scope of the HR team, it’s an important point to raise. What you don’t
want is a situation where employees have to rely on data experts to manipu-
late data for them. Simple self-service tools (like your HR analytics tools)
are the best way forward.

Addressing the data skills shortage

We can’t ignore the fact that, around the world, demand for data and other
technology skills massively outstrips supply. So what does this mean for
your ability to tap into data skills?

Hiring in a skills shortage

Demand for work involving advanced data analytics and AI is growing
every day and there just aren’t enough people with data and AI skills to go
around. As an example, the number of job openings posted on sites like
LinkedIn and Glassdoor for people with data science skills is growing expo-
nentially and, according to data gathered by Quanthub, the number of posts
exceeds the number of people searching for these jobs by a factor of three.1
And what talent there is tends to be hoovered up by giants like Google,
 Building skills and aligning culture 189

Apple, Microsoft, IBM, and so on. For everyone else, attracting the right
talent is a real challenge.
Naturally, this may impact your ability to hire people with data and AI
skills. Which means you may need to get creative. One option is to recruit
people who have strong analytical skills, as opposed to specific data science
or AI skills – mathematicians, for example, or people with a background in
statistics. You can then train them to use the required data and AI tools.
As with any position, the ability and desire to grow is incredibly valuable.
Someone who doesn’t tick all the boxes on paper but is very keen to learn
new skills and grow with the business will always be a better fit than some-
one who is fixed in their ways and unwilling to learn – no matter how expe-
rienced and knowledgeable they are in data and AI. Besides, the world of
analytics and AI is moving so fast and new technologies and applications are
emerging all the time, which means the ability to adapt and learn is becom-
ing increasingly important.

Upskilling existing employees

If you don’t have a central data or analytics team, and hiring in-house ex-
pertise isn’t an option, how else can you tap into data skills? The first option
is to upskill your existing employees. Not just in terms of data literacy –
which is a must for all employees – but also in terms of more advanced AI
and data science skills.
The truth is, while you want everyone in the organization to become
more confident around data and future technologies, it’s likely that ‘star
players’ will start to emerge across the business – people who have a natural
affinity for working with data and AI, even if they don’t necessarily have
formal training in them. Investing in more advanced training and upskilling
for these team members is likely to bring the greatest rewards. You might
even start thinking of these people as your ‘data ambassadors’ – people who
can spread the word about data and how it benefits the business.
Upskilling a workforce can take some time but it doesn’t necessarily have
to be hugely expensive. Lots of free and inexpensive training material is
available online, from both universities and organizations. Some resources
to consider include:

●● ‘A crash course in data science’ by John Hopkins University, which is

available on the Coursera online learning platform
●● ‘Introduction to data science’, available for free on Alison.com
190 Making data-driven and AI-enabled HR happen

●● Various data science courses available on datacamp.com

●● ‘What is data science?’ by IBM – another course available through the
Coursera platform
●● ‘Google data analytics professional certificate’ – going a little deeper than
some of the above introductory courses, this course aims to set anyone up
as a professional data scientist in six months. While it’s primarily aimed
at those seeking a first job in data science, the skills taught are equally
valuable for someone looking to upskill within their own organization.
You can find this course on Coursera.

Tapping into skills outside the organization

When upskilling employees isn’t an option, or you need more immediate
support, you’ll have to start looking outside the organization for skills and
capabilities.
Your main options here are:

●● outsourcing to an external provider

●● acquihiring
●● setting up a tech incubator/innovation hub
●● crowdsourcing

Outsourcing to external providers

There is a large market of providers out there who can handle your data and
analytic needs – and the market is growing all the time. Whether you’re
looking for an all-in-one service covering everything from collecting data to
presenting key insights, or you just need some help with applying AI analyt-
ics to data that you already have, there will certainly be a provider who can
meet your needs.
Generally speaking, you can partner with a provider for the longer term –
or, if you don’t want to be locked in with a specific provider, you can out-
source on a project-by-project basis. Personally, I tend to favour more of a
partnership approach rather than straight-up outsourcing, as that way your
partner is likely to be more invested in achieving your organization’s goals
and vision. Partnering may also help you boost skills within your organiza-
tion – aiding knowledge transfer, as opposed to simply offloading your pro-
jects onto another firm’s to-do list.
 Building skills and aligning culture 191

Some of the biggest data and AI providers are household names like
Facebook, Amazon and IBM, and they all offer consultancy services and
practical tools, but you certainly aren’t limited to the big corporations.
There are plenty of smaller contractors out there and these may provide a
more personalized, tailored service or have specialist knowledge of your
industry. In fact, in my experience, industry-specific providers are becoming
the norm as opposed to big generalists.
When looking for a third-party provider, it’s a good idea to start with
recommendations from your networks and contacts. Failing that, there are
many case studies available online and in books (including my own books
Big Data in Practice: How 45 successful companies used big data analytics
to deliver extraordinary results, and Artificial Intelligence in Practice: How
50 successful companies used AI and machine learning to solve problems)
and these help to highlight providers who are doing excellent, innovative
work.
It’s vital you partner with a provider who understands what you’re trying
to achieve. The better your contractor understands your key business ques-
tions, your strategic goals and the challenges you face as you work towards
those goals, the more likely they are to get to the insights you really need.
Always ask for examples of their other client work – even if you’ve read
about their work in case studies or they came recommended to you by a
trusted contact. You will want to find out as much as possible about how
their previous projects unfolded, what the key challenges were and, cru-
cially, what concrete results the clients saw as a direct result of working with
that provider. Also consider whether specialized knowledge of your business
sector is important or not, as that will inform the selection process.

Acquihiring
Acquihiring (a mash-up of ‘acquisition’ and ‘hiring’) is proving a popular
solution for many businesses. The term refers to the emerging trend where
companies who need to boost their AI and data skills simply acquire small
start-ups. By buying up a small start-up, the acquirer gets quick access to
data engineers and scientists who have experience building and training AI
models – thereby accelerating their progress.
However, as with any acquisition, there are pitfalls to this approach. One
MIT study found that a whopping 33 per cent of acquihired talent leave in
the first year after purchase,2 which demonstrates the importance of manag-
ing the transition carefully and looking for acquisitions that are a good
cultural fit with your own organization.
192 Making data-driven and AI-enabled HR happen

Setting up a tech incubator or innovation hub

Another way to access external skills is to set up a tech ‘incubator’ or inno-
vation hub, which can be a good way for companies outside the tech indus-
try to gain access to talent and foster collaborations with AI experts. For
example, brewer AB InBev – the world’s largest brewer, and makers of
Budweiser, Stella Artois and Corona – has created a Silicon Valley innova-
tion hub called the ‘Beer Garage’ to explore how cutting-edge technologies
like AI can help drive performance. The Beer Garage is designed to help the
company research, develop and test technology-driven solutions – but also
puts the company in close proximity to the vast network of tech companies
and venture capitalists in the Silicon Valley area. This fosters collaboration
with local start-ups and helps to drive innovation. The company says it’s
learning a lot from working with these inspiring tech experts.3

Crowdsourcing
One final option that may be worth considering is crowdsourcing platforms
such as Kaggle – a competition-based platform that allows businesses to tap
into an army of armchair and citizen data scientists. As we saw in Chapter 7,
even large corporations like Walmart have turned to Kaggle to find analytics
talent.
Kaggle acts as a middleman: companies bring their data (whatever it may
be), set a problem to solve as well as a deadline and offer a prize (usually
cash, or sometimes a job). It’s a fascinating idea which has so far seen con-
testants compete to solve problems ranging from analysing medical records
to predict which patients are likely to need hospitalization, to scanning the
deep cosmos for traces of dark matter. Hal Varian, Chief Economist at
Google – another mega-company that has used Kaggle’s services – has de-
scribed Kaggle as ‘a way to organize the brainpower of the world’s most
talented data scientists and make it accessible to organizations of every size’.
When working with a crowdsourcing platform, you might want to use
synthetic data, to avoid privacy concerns or any worries about commer-
cially sensitive data falling into the hands of competitors via a public
­platform.
Crowdsourcing has great potential for identifying emerging talent and it
provides businesses with new ways of engaging with people who can poten-
tially help them solve their problems and answer key business questions.
And because the competitive element ensures those taking part will strive to
make sure their ideas stand out from others, this encourages out-of-the-box
thinking that can lead to some very innovative solutions for businesses. So,
 Building skills and aligning culture 193

if you are struggling to attract talent or, for whatever reason, you don’t want
to partner with an external provider, crowdsourcing your data analysis
could be an option. It’s a great way to supplement skills, access additional
analytical brainpower and test the waters on new data projects.
So which approach is right for your organization – crowdsourcing, acqui-
hiring, setting up a tech incubator or partnering with third-party providers?
It’s likely you’ll need a combination of these different strategies rather than
relying on one single approach.
And remember, for relatively simple data and AI projects, many off-the-
shelf HR analytics tools can meet your needs and require little or no prior
knowledge to get started. They’re designed for HR professionals, after all,
not data scientists.

Preparing for a cultural shift

in HR and the organization
It’s virtually impossible to realize this vision of intelligent, data-driven, AI-
enabled HR if the HR team – and indeed, the organization as a whole – isn’t
on board with data and AI initiatives. You therefore need a culture that
recognizes the importance of data and AI. A culture of innovation that em-
braces change and continual learning. A culture that’s open to new opportu-
nities instead of clinging stubbornly onto old ways of doing things.
Technology has always brought about new ways of working, but what’s
different about the current wave of transformation is the sheer pace of
change. AI in particular is going to fundamentally change the work that
humans do. And not in a generation’s time – I predict we’ll see enormous
changes within the next five to ten years. The HR team and the organization
as a whole must start to prepare people for these changes and build a culture
that embraces rather than fears change.

Seeing automation as a positive force

I don’t want to downplay the impact of automation – in many industries
and jobs, the impact of automation will be keenly felt, and many jobs will
change or be displaced. However, it’s really important to recognize the posi-
tive side of automation and AI. Because, just as the previous industrial revo-
lutions have ultimately created more jobs (and better jobs), AI and
automation will create more jobs than are destroyed. My hope is it’ll create
better jobs for humans.
194 Making data-driven and AI-enabled HR happen

In fact, AI has been described as ‘the greatest job engine the world has
ever seen’.4 Rather like the internet before it. Yes, the internet had a negative
impact on some jobs, but look how many more jobs it’s created and how it’s
enabled businesses to branch into new markets, reach new customers,
streamline their business processes, and so on.
Research by Capgemini backs up this idea of AI and automation aug-
menting rather than displacing human workers altogether. In a survey of
1,000 organizations that have already deployed AI-based systems, four out
of five companies had created more jobs.5 I spoke to one of the survey re-
spondents, insurance giant Prudential. The company’s global head of AI,
Michael Natusch told me, ‘Instead of looking for ways to replace humans
with AI, we are seeking the most fruitful complements.’6 He cites robotic call
centre assistants as one example. ‘Clearly, nobody wants to talk to a robot.
But if a robot answers a phone call on the second ring and provides the right
information at the right moment in time, then there is value in this. Our call
centre agents appreciate the collaboration with robots as they are now able
to focus on harder problems that require their experience, creativity, and
empathy.’
In other words, AI systems are great at automating the boring, mundane
and repetitive stuff, those tasks that are easily repeatable, rules-based and
uncreative. And this, in turn, allows humans to focus on more creative, em-
pathetic and interpersonal work.
All this means it’s important to approach automation and AI with an
open mindset – and for the organization to find ways to augment the work
of human employees, rather than making it redundant altogether. It’s about
using AI to help people do their jobs more efficiently and add greater value
to the organization.
HR’s role is to guide the organization through this process, to help find the
right balance between machines and humans, and, of course, prepare people
within the organization for change. Whatever changes take place in your
organization, it’s important that people feel change is not done to them. That
they have opportunities to learn and enhance the work that they do.

What this means for human skills

When parts of jobs are automated by machines, that frees up humans for
work that is generally more creative and people-oriented, requiring skills
such as problem solving, empathy, listening, communication, interpretation
 Building skills and aligning culture 195

and collaboration. These are the skills we looked at in Chapter 10 – skills

that humans are still better at than machines.
According to Deloitte, as machines take on the mundane, repetitive work,
jobs become more human, which makes the work and contribution of peo-
ple in the workplace more valuable and important. This is borne out by the
results of a Deloitte global human capital trends survey:7

●● 62 per cent of respondents were using automation to eliminate transac-

tional work and replace repetitive tasks.
●● 47 per cent were using automation to improve existing work practices
and boost productivity.
●● And 36 per cent were ‘redesigning work’ as a result of automation.

The upshot is that individuals and organizations will need to develop both
the technical (data literacy and tech awareness) and softer human skills to
succeed. Employees at most levels will be required to access data and work
out what action to take based on what the data tells them. And everyone
must be able to understand the potential impact of new technologies on
their industry, company and job. And, on top of that, most employees will
need to cultivate softer skills like communication and creative thinking.
This is why it’s so important to foster a culture that embraces continual
learning and curiosity.

Building an AI and data culture – practical

steps
To recap what we’ve covered so far in the chapter, if you were to create a
roadmap of how to build the right culture for data-driven and AI-enabled
success, it might look something like this:

Step 1: Redefine work and the role of people

What does work look like in the HR function and the wider organization in
future? Much of today’s work is geared around performing a particular
function. But this is likely to change as organizational structures become
more fluid. The focus will shift to projects and outcomes rather than par-
ticular tasks that need to be repeated over and over again (because the latter
is exactly the sort of work that machines excel at). People within the HR
196 Making data-driven and AI-enabled HR happen

team and the wider organization will need to develop the skills and mindset
to cope with this, but they can’t do it alone – HR needs to lead in this area,
to show how data and AI can add serious value for the organization.

Step 2: Improve technical skills

As I’ve said, we won’t all need to be data scientists or AI experts in the future,
but we will all need some degree of data literacy. The HR team and wider
organization will need to develop data literacy skills so that people are
equipped to take advantage of AI and data, and learn to ask questions such as
‘How can we use this to improve performance?’, ‘How can we ensure we’re
using this technology ethically?’ and ‘How accurate is the data we’re working
with?’ Most likely, you will also need to beef up the company’s more ad-
vanced technical capabilities, whether that means partnering with external
providers, acquihiring, crowdsourcing or hiring in-house data and AI experts.

Step 3: Develop human potential

As machines begin to master more tasks typically performed by humans,
humans must begin to focus on the areas in which they outperform ma-
chines – creativity, imagination, critical thinking, communication, and so on.
As such, your learning and development programmes can’t afford to over-
look these inherent human abilities.

Step 4: Redefine learning to focus on continual

learning cycles
One report by Dell Technologies and the Institute for the Future predicts
that 85 per cent of the jobs that will be available in 2030 don’t exist yet.8
People will no longer start a career path and grow with one role, and many
of us may be doing very different jobs in a decade’s time. Which makes
learning all the more important. But not just learning; continual learning. As
the world of work continues to evolve, education, learning and training
must become a continual endeavour.

How to successfully manage change

We know that change can be painful, and badly managed change can sap
morale, reduce performance and ultimately hinder the adoption of new
 Building skills and aligning culture 197

practices or technologies. It’s therefore up to the HR team to carefully man-

age change.
Communication is key here. When people don’t understand why change
is happening, they’re much more likely to resist it. Demonstrating the clear
link between AI and data initiatives and the organization’s key strategic
priorities helps to secure buy-in, as does showcasing success stories. (You
can always showcase examples from other companies that have successfully
harnessed data and AI, perhaps using some of the examples from this book.)
This is also where your ‘quick win’ use cases come into play (see
Chapter 12), since these help you demonstrate the value of data and AI, in a
relatively short space of time and for relatively little spend.
What’s more, it will be up to HR teams to listen to employees’ concerns
and provide reassurance. Many people assume AI and automation will lead
to job losses, but that’s not necessarily the case. AI will create more jobs than
it displaces. And even for those jobs that are impacted by automation, the
likelihood is that many of them will change rather than be lost altogether.
Engaging and involving everyone in these discussions is key for getting peo-
ple on board with change.

Key takeaways

To summarize the key takeaways from this chapter:

●● It’s vital that everyone in the HR teams sees data as a key asset, and that
HR professionals build their own data literacy. It’s also up to HR to lead
the way in developing data literacy across the wider organization.
●● When it comes to tapping into data and analytics skills, the ideal options
are to work with your organization’s central data/analytics team, or
maybe even hire data/AI specialists into the HR team. Recognizing that
hiring won’t be an option for many businesses – especially given the
current shortage of available tech talent – other options include
upskilling the existing workforce, partnering with third-party providers,
acquihiring, setting up a tech incubator or crowdsourcing your data
analytics needs.
●● It’s vital organizations and individuals prepare for the changes that are
coming our way. This means building data skills and capabilities, but
also developing our uniquely human soft skills.
198 Making data-driven and AI-enabled HR happen

●● You want to build a culture that embraces data, AI and automation. But
to do this, the HR function needs to listen to the workforce and ensure
people do not feel like change is being done to them. Those companies
that can build a positive data and AI culture, successfully manage
change and bring everyone along on this transformative journey, are the
ones that will succeed.

Of course, skills and culture aren’t the only ingredients for data-driven and
AI-enabled success. You also need to build the right technology infrastruc-
ture. In the next chapter, we’ll explore the different layers of technology in-
frastructure that you will need to consider.

Notes
1 Quanthub. The data scientist shortage in 2020, Qaunthub, 2020. quanthub.
com/data-scientist-shortage-2020 (archived at https://perma.cc/D7SD-4ZXL)
2 M Somers. Your acquired hires are leaving. Here’s why, MIT Sloan School, 8
January 2019. mitsloan.mit.edu/ideas-made-to-matter/your-acquired-hires-are-
leaving-heres-why (archived at https://perma.cc/KLX2-BABW)
3 B Marr. The amazing ways the brewers of Budweiser are using artificial
intelligence to transform the beer industry, Forbes, 9 September 2019. www.
forbes.com/sites/bernardmarr/2019/09/09/the-amazing-ways-the-brewers-of-
budweiser-are-using-artificial-intelligence-to-transform-the-beer-industry
(archived at https://perma.cc/UK8F-S43R)
4 B Reese. AI will create millions more jobs than it will destroy. Here’s how,
SingularityHub, 1 January 2019. singularityhub.com/2019/01/01/ai-will-create-
millions-more-jobs-than-it-will-destroy-heres-how/amp (archived at https://
perma.cc/2RJQ-FHRZ)
5 Capgemini. Artificial intelligence – where and how to invest, Capgemini, nd.
www.capgemini.com/service/digital-services/insights-data/data-science-analytics/
artificial-intelligence-where-and-how-to-invest (archived at https://perma.cc/
P6NV-XMQV)
6 B Marr. Instead of destroying jobs artificial intelligence (AI) is creating new jobs
in 4 out of 5 companies, Bernard Marr, nd. bernardmarr.com/default.asp?
contentID=1194 (archived at https://perma.cc/N5GH-ME64)
7 J Schwartz et al. From jobs to superjobs: 2019 global human capital trends,
Deloitte, 11 April 2019. www2.deloitte.com/us/en/insights/focus/human-
capital-trends/2019/impact-of-ai-turning-jobs-into-superjobs.html (archived at
https://perma.cc/TH7K-RFNS)
 Building skills and aligning culture 199

8 Dell Technologies and Institute for the Future. The Next Era of Human–
Machine Partnerships, Dell Technologies and Institute for the Future, 2017.
www.delltechnologies.com/content/dam/delltechnologies/assets/perspectives/
2030/pdf/SR1940_IFTFforDellTechnologies_Human-Machine_070517_
readerhigh-res.pdf (archived at https://perma.cc/V4NP-9K94)
200

Creating the 14
technology and
data infrastructure
By this point in the journey, you’ve created your data and AI strategy, identi-
fied your use cases and considered the skills and capabilities that you need.
Now’s the time to think about technology. You might be surprised that tech-
nology comes so late in the process, but that’s deliberate. You should never
start with the technology itself, because that’s a sure-fire route to ‘technol-
ogy for technology’s sake’. But with all your requirements fleshed out, you
can finally begin to identify the technology – i.e. the software and hard-
ware – that will help you deliver your goals.
But first you need to ensure that the HR tech/IT strategy fits with the
organization’s wider technology strategy and infrastructure. So that’s where
we’ll start this chapter. After that, we’ll move onto the infrastructure itself,
and explore the three layers that you need to consider before investing in
any technology.

Your HR technology strategy – and how

it fits with the wider business strategy
It’s really important to align any HR technology infrastructure with the
overall technology infrastructure in place in the organization. What I mean
by that is, the vendors that you use and analytics products that you choose
might be informed by vendors and products that the organization already
deploys elsewhere in the business. (Of course, they might not, if there really
is no relevant infrastructure already in place in the organization. My point
is, you should look at existing frameworks before bringing new technology
or vendors into the organization.)
 Creating the technology and data infrastructure 201

Obviously, this is something you’ll want to discuss with your IT team

(and central data team, if there is one). No doubt your business will have a
wider technology/IT strategy in place that covers the business as a whole, so
that’s a good place to start your discussions.
You may also have your own HR technology strategy (that covers the
wider HR tech landscape, not just data and AI). If that’s the case, you should
review that strategy in light of your data and AI requirements. And, in the
process, ensure your HR tech strategy is aligned with the business’s over-
arching tech strategy.
If you don’t have an HR technology strategy, we’ll look at that next. But
again, you’ll want your HR technology strategy to be an extension of the
entire organization’s technology strategy.
The key point, then, is to ensure there’s a clear line of sight between your
HR technology strategy, and the wider business technology strategy – just as
you would ensure a clear line of sight between the HR function’s strategic
goals and the organization’s strategic goals.

What is the HR technology strategy? (And is it

different from your data/AI strategy?)
The HR technology/IT strategy is like a roadmap that sets out how the HR
function can get the most out of technology and transform HR workflows.
In other words, it looks at the broader digitization of the HR function, not
just the role of data and AI (although, naturally, the data and AI component
will form a major part of your technology strategy). So, it might cover HR
social media usage, for example, or job application systems, as well as peo-
ple analytics systems.
Creating a technology strategy generally involves:

●● Confirming the HR function’s strategic goals. Because your technology

strategy should cascade down from HR goals, which in turn cascade
down from the wider business goals.
●● Looking at the HR technology already in place, and assessing whether it’s
fit for your current and future needs. Basically, are your technology
systems supporting your goals?
●● Assessing existing technology tools and providers – again, to ensure they
meet current and future requirements.
202 Making data-driven and AI-enabled HR happen

●● Identifying technology infrastructure gaps and skills gaps that need to be

filled in order to deliver the HR function’s strategic goals.
●● And, of course, identifying your strategic data and AI requirements, as
we’ve discussed elsewhere in this book.

Ordinarily, this would all be developed in partnership with your IT function.

Your IT colleagues will be able to advise on technology solutions and pro-
viders, and work with you on implementation – but it’s up to you, the HR
function, to maintain a laser-like focus on your strategic goals. To put it
another way, your IT colleagues are the technology experts, but you are the
expert on exactly what it is that HR is trying to achieve.
As we saw in Chapter 13, it’s not just about having the right strategy in
place, or inspiring use cases in mind – you also need to create the right cul-
ture and foster the skills that will help you execute your strategy success-
fully. So do circle back to Chapter 13 for a recap on skills and culture.

Ensuring your technology strategy stays agile

As with any strategy, it’s really important that your technology roadmap can
flex and adapt to changing needs. Because, as we’ve seen throughout the
book, things are changing fast. You need your technology systems – data-
and AI-related systems and other systems – to be able to respond to change
without causing major upheaval.
As such, I recommend reviewing your HR technology strategy on an an-
nual basis to assess how you’re doing against your plan and where things
have changed (both in terms of HR requirements and the wider business
direction).
When it comes to the technology itself, the wide array of cloud-based ‘as-
a-service’ solutions typically provide plenty of ‘plug and play’ options –
meaning you can quickly reconfigure your subscription, turn on new op-
tions and turn off services that you no longer need. Staying agile is therefore
easier than it used to be, back when investing in new software and hardware
meant a big financial outlay and long-term commitments. That said, do keep
the need for agility in mind as you assess technology options. Always try to
choose technology solutions that are flexible, and can scale up or down ac-
cording to your needs.
 Creating the technology and data infrastructure 203

Putting in place the right data

and AI infrastructure
Now, finally, we come to the process of identifying the right data- and AI-
related technology for you. In essence, this is where you decide on the soft-
ware or hardware that will take your data and turn it into insights. It’s likely
you will need to invest in some tools and services to make this happen, even
if your company already has some existing data/AI infrastructure in place.
And the good news is, you have more choice than ever.
Until recently, it was difficult for the average business function to work
with a wide variety and volume of data without making heavy infrastruc-
ture investments – expensive software and hardware, storage facilities, a
team of data analysts, and so on. Thankfully, that’s no longer the case.
Developments like data- and AI-as-a-service and the ever-expanding market
of third-party providers now allow even the smallest company to harness
data and AI relatively easily.
When thinking about the technology infrastructure you need, there are
three layers to consider:

●● infrastructure to collect data

●● infrastructure to store data
●● infrastructure to access and analyse data

We’ll discuss each layer below. But, given that most companies have some
existing infrastructure in place, it makes sense to consider what you already
have in relation to each of these layers. Therefore, as you work through
these sections, make a note of what technologies your organization already
has in place (work with your IT or data team on this). You will probably
have to make changes and additions to existing infrastructure, but just keep
in mind that some of your existing systems may have a role to play in meet-
ing your data and AI needs. For example, you might already be collecting
useful data through your customer service centre, even though you don’t yet
have the ability to analyse it. Depending on your use cases, you might want
to consider whether your existing infrastructure could be updated to give
you those abilities.
Ultimately, there’s no one-size-fits-all solution. Believe me, it would be
great if I could just say ‘any HR team should buy this infrastructure-as-a-
service solution from company X’ but in reality, you are going to have to
work through each of the following layers and consider what will be the
204 Making data-driven and AI-enabled HR happen

best fit for each of your data and AI use cases. Again, you should work with
your IT/central data team as you consider each layer. And if you don’t have
in-house expertise to help you assess infrastructure needs, remember that
there are plenty of data and AI experts out there – myself included – who
can consult with you.

Layer 1: Infrastructure to collect data

The data collection layer is where the data arrives at your company, whether
it is internal or external data, structured or unstructured. This may include
data from your employee feedback, recruitment channels, social media
channels, sensors (either wearable sensors or on machines) or any other
source. Match the data that you have available to the data requirements of
your chosen use cases and see if there are any holes that need filling.
You might need to source some or all of the data required, and sourcing
new data might mean it’s necessary to make new infrastructure investments.
If your requirements are for external data, then it’s just a case of finding
the right provider, and you might not need to make any infrastructure
changes at all. Often, ‘off the shelf’ cloud services can simply be plugged into
freely available public datasets and set to work. But if you’re looking for
more specialized data, such as data on your specific industry, then it’s likely
you will need to find an appropriate data broker. Luckily, as businesses have
reaped the benefits of moving to data-driven decisioning, a market has
sprung up where data on just about any industry or activity is up for sale,
for those who need it.
If your requirements are for internal data, you will have to find the infra-
structure tools you need to collect it. Circle back to Chapter 3 for a re-
minder of the different types of data available these days (such as sensor
data, activity data and conversation data) – and how to collect it. Remember
that synthetic data may be a good option when the data you want doesn’t
exist.
Exactly what tools or systems you need for capturing data will depend on
the type of data you need, but some options include:
●● sensors
●● fitted to devices, machines, buildings, vehicles, packaging, employee
badges or anything else you want to capture data from
●● CCTV and video
●● website cookies
 Creating the technology and data infrastructure 205

●● employee feedback systems

●● employee communication systems
●● social media feeds

Be aware that collecting new forms of data introduces new vulnerabilities to

your organization, and you’ll need to take the necessary steps to protect that
data. But we’ll talk more about that in the next layer…

Layer 2: Infrastructure to store data

Having identified your data capture needs, next you have to think about
where you will keep your data. The main choices we have here are between
traditional, on-premises data warehouses and cloud-based systems. Often,
though, this won’t be an either/or decision, as hybrid options are becoming
increasingly common.

On-site storage options

As far as on-premise storage solutions go, regular hard disks of very high
capacity are available very cheaply these days, by enterprise IT standards. If
you’re a small business and you don’t need to store a huge amount of infor-
mation for frequent, high-volume analytics, this might be all you need.
Other options include solid state storage solutions, and even old-fash-
ioned magnetic tape. Solid storage is most frequently used for smaller vol-
umes of data that you need to access very frequently, as they offer very high
access speeds but are relatively costly. The price of this storage is falling all
the time though, and in the future it’s likely to become an increasingly viable
solution for large-volume, long-term storage. Compared to mechanical,
magnetic hard disks, solid state drives offer very high reliability, low failure
rates and low latency, which makes them ideal for tasks where speed and
precision are key.
Magnetic tape solutions may seem old-fashioned and often people that I
speak to are surprised to find that they are still a viable commercial option
in this day and age, but in fact a 2019 survey found that 90 per cent of the
organizations surveyed still rely on tape data for some of their storage re-
quirements – often a considerable amount.1 Tape storage is mainly used
because it is very cheap in comparison to other storage media, so for data
that needs to be archived long-term and accessed very infrequently, it’s often
the ideal choice.
206 Making data-driven and AI-enabled HR happen

In the cloud
These days, however, there are many reasons that you might choose to
forego on-premise data storage infrastructure, or at least complement it
with off-premise solutions, in the form of cloud storage solutions.
Just as is the case with cloud-based analytics solutions, cloud storage so-
lutions offer the advantage of letting you get up and running right away.
What’s more, cloud storage is incredibly flexible, you can create additional
storage whenever it’s needed and when you take up-front set-up costs into
account, it’s generally the more affordable option – particularly if you’re
planning on scaling up to work with bigger and bigger datasets.
Cloud storage simply means that your data is stored on servers that are
owned and operated by a cloud service provider, usually remotely, but con-
nected to the internet so you can access the data from anywhere at any time.
There is a high amount of built-in redundancy, meaning that your data is
distributed as multiple copies in numerous locations, so if one cloud data
centre experiences problems, you will still be able to access your data. The
price will change according to how much data you store or the volume of
data you stream in and out of the cloud.
With cloud, security is certainly an important consideration. It might
seem logical to assume that storing data outside of private company servers
inevitably creates risk, and it does. However often this is offset by the over-
all robustness of the security provided by global enterprises that have built
their entire business model on data. You can (hopefully) be relatively confi-
dent, for example, that a large cloud provider will keep all of its software
patched and up to date. Firewalls will be in place to prevent DDOS attacks
and the physical premises where the data is actually stored will be patrolled
by security guards. Additionally, as mentioned above, your data is likely to
be spread and duplicated across multiple locations, so even a catastrophic
event like a fire or earthquake hopefully won’t result in the loss of all of
your data. Of course, it’s possible to make sure all of these measures are in
place when you are storing all of your data on-premise, but it’s a big re-
sponsibility.

The difference between public, private and hybrid cloud

Public cloud generally refers to cloud services provided by a third-party or-
ganization that specializes in doing so. The data itself is not usually public –
rather, ‘public’ refers to the fact that the third-party organization provides
 Creating the technology and data infrastructure 207

cloud computing services to customers (either individuals or businesses) and

anyone can use them. The biggest public cloud providers at the moment are:

●● Amazon Web Services

●● Microsoft Azure
●● Google Cloud Services
●● Alibaba Cloud
●● IBM

Private cloud, on the other hand, is simply a term for what companies do
today when they maintain everything in-house, but following similar mod-
els of deployment, access management and infrastructure maintenance as
the public cloud providers. A company that outsources all of its data and
computing requirements to another organization – one that does not also
offer the same services publicly – could also be considered to be operating a
private cloud infrastructure. The primary reason for doing this is often secu-
rity – in some cases, businesses will be working with data that is so sensitive
that it isn’t permissioned for storage outside of the organization’s immediate
jurisdiction. Additionally, your own private cloud infrastructure can be con-
figured exactly as you require.
Of course, as with everything related to analytics and even computing in
general, with cloud there’s no one-size-fits all approach. This has led to the
emergence of what are termed ‘hybrid-cloud’ and ‘multi-cloud’ models.
Hybrid cloud usually refers to a solution comprised of elements of public
and private clouds. This can be useful when, for example, a subset of the
data you are using is too sensitive to let out of your hands, but other data is
safe to host publicly, where you can take advantage of the infrastructure
that public providers make available. This creates a very agile environment
where the best elements of each ecosystem are on-hand, as and when they
are needed. A challenge is that software needs to be able to communicate
across public and private servers, even if it can’t access the same data on
each one, but this is generally catered for by the big providers of hybrid
cloud services.
Another option that you might come across is known as virtual private
cloud. This is usually a service offered by a public cloud provider that will
deploy its infrastructure within your own premises or data centre, again
with the aim of creating a ‘best of both worlds’ situation, where you have
the total control of everything being under your roof, while also benefitting
from the tools, services and interfaces provided by the public provider.
208 Making data-driven and AI-enabled HR happen

Finally, multi-cloud refers to picking and choosing different solutions

from different cloud providers. Sometimes this is to ensure the highest level
of availability, particularly if you are serving data to users or customers
spread throughout the world.

The importance of avoiding data silos

An essential factor when deciding where to store your data is accessibility.
This doesn’t just mean making sure you can get hold of it when you need
it – it means making sure it’s as readily available as possible to those who
need it throughout the organization. These decision makers may not be part
of the HR function, but they may still need access to HR data in order to
make better decisions and drive performance. This thinking goes as far back
as 2002 when then Amazon CEO Jeff Bezos issued a ‘mandate’ that all data
was to be made as widely available throughout the company as possible.
Siloing is something that occurs in organizations when data is collected
by disparate teams and simply stored without any consideration for whether
it might be useful for another team. Often, one team doesn’t even know
what data the other team has, and over the years this can lead to countless
wasted hours and money as data capture and storage is replicated. Basically,
data silos happen because, in the average business, data lives in multiple
systems and locations across the company, with each team only having ac-
cess to ‘their’ data, and having no idea what other data might be available
elsewhere in the business. As you can imagine, this reduces transparency and
collaboration – the very opposite of what you want to achieve with intelli-
gent HR.
So how can you spot a data silo? Dead giveaways are when data is incon-
sistent or conflicting across the business, or when data is hard to access.
Another is when department heads or managers complain that they simply
don’t have the data they need to do their job.
Data silos are a major problem for the organization as a whole to
­address – ideally by investing in infrastructure that allows data to be stored
in one easily accessible place. Even when data comes from multiple sources,
it can still be stored in one place. The organization should also have
­company-wide procedures in place for managing and maintaining data.
This isn’t to say you shouldn’t have your own HR data systems. Naturally
you will. But it’s really important to consider the need to share data with oth-
ers in the business, and ensure data is easily accessible to those who need it.
This is also why it’s so important to align your strategy with the wider busi-
ness technology strategy. By aligning the HR strategy with the ­organization’s
 Creating the technology and data infrastructure 209

overarching technology strategy – and by considering who else might need

access to your data – you should be able to avoid the data silo trap.

Keeping your data safe

Data security is another key consideration that falls under data storage. It’s
vital that you take steps to keep your data safe from accidental loss or mali-
cious data breach. Both of these can have legal ramifications under legisla-
tion (such as GDPR), but are also essential considerations of any data
governance policy.
The heaviest governance burdens lie with personal data. However, one of
the most useful ways to reduce your liabilities is to follow a strategy of data
minimization. You can also take steps to anonymize personal data, severing
the link between the data itself and the person related to that data.
Data breaches are an increasingly prevalent threat to business. But, again,
there are some steps you can take to mitigate these risks. One is a system of
permissioning that defines who is allowed access to any particular set of
data. Another is the use of encryption – because if data is encrypted it’s far
less useful to anyone who might want to steal it. You might want to consider
homomorphic encryption as an option. Here, data is encrypted in such a
way that it can be analysed while remaining in its encrypted form – even the
analytical algorithms don’t ‘see’ the unencrypted data. The data can even be
edited in the cloud by people with the right permissions, without the unen-
crypted data being exposed to the cloud servers.
Two other techniques that can be used to de-identify data are masking
and tokenization. Masking involves obfuscating sensitive elements of the
unencrypted data with other data of the same type, while leaving other ele-
ments intact. This might involve swapping, say, all of the information in
each ‘city’ field for a different city. Only those with the correct permissioning
will see the correct data, but the data still remains useful for many applica-
tions. Tokenization is similar but replaces key or sensitive parts of the data-
set with anonymized, randomized tokens. Unlike with encryption, there’s no
mathematical way to reverse-engineer the original data from the hidden
data (which is possible with huge amounts of computing power with many
forms of encryption). This is because the tokens are randomly assigned
rather than being mathematically derived from the original data. Also, while
encryption is typically applied to a whole record, tokenization (and mask-
ing) are generally applied to specific fields within the record.
Circle back to Chapter 4 for a recap on the major data and AI risks, and
how to overcome them. But, overall, it’s important to remember that data
210 Making data-driven and AI-enabled HR happen

security is a highly specialized field and it’s something you will probably
want to discuss with experts.

Don’t overlook threats from the IoT

The ever-increasing number of smart, connected devices has led to a rise in
the angles of attack that hackers have at their fingertips. According to re-
search, millions of IoT devices have security vulnerabilities that can be ex-
ploited to allow unauthorized access to data.2 From machine sensors to
smart employee badges, think of every connected device you have in your
network as a ‘door’ into your company – a door that has to be kept locked
and secure from intruders, just like any other door.
A very important practical first step is to ensure you always change any
default passwords or login information – this is often the way that many IoT
devices are compromised. This is also another area where a policy of ‘mini-
mization’ can pay dividends. Consider just how connected you need your
equipment to be – of course, most devices need to interface with a smart-
phone or computer app, but do they really need to be able to connect and
interface with any other devices? Be sure to understand exactly ‘what is
talking to what’.

Layer 3: Infrastructure to access and analyse data

Finally, you need to consider how you will process and analyse the data
you’ve collected and stored, in order to extract the insights you need. In
Chapter 3 we looked at the main options currently used for data analytics,
many of them falling under the heading of AI. This layer is about selecting
the tools (i.e. analytics software) that you need to do this.
The process of extracting insights from data can be distilled into three
steps:

1 preparing the data – identifying, cleaning and formatting it so it can be

analysed efficiently
2 building the analytics model
3 drawing a conclusion from the result of the analytics

As with storage functions (layer 2), much of this data processing layer is
now done in the cloud (although some of the analytics providers can deploy
cloud on-premise – depending on the decisions made during the previous
layer). Options like BigQuery (Google), Amazon Web Services and Microsoft
 Creating the technology and data infrastructure 211

Azure HDInsights all provide tools that carry out analytics on whatever
data you throw into them. In addition, Amazon QuickSight, Infobright, IBM
Cognos Analytics, Hortonworks Data Platform, Cloudera Data Warehouse,
Pivotal Analytics, Sisense, Alteryx, Splunk and SAP Analytics Cloud are all
tools that have proven their capabilities and are used by businesses globally.
As well as these established companies, a large number of start-ups have
emerged offering solutions tailored to specific workloads or industries.
This layer is really where AI comes into its own, because many of the
advanced analytics options on the market are built on machine learning. But
before you invest in infrastructure or an ‘off the shelf’ AI analytics service,
do remember to look at the systems already in place in the organization. In
other words, while there are plenty of HR-specific analytics tools on offer,
you want to make sure your chosen option fits with the business’s wider
strategy. For example, if your organization is already using a certain analyt-
ics platform, first consider whether that platform has an HR offering.
You will also want to consider how you’ll communicate insights from the
data to the people who need them – within both the HR team and other
decision makers in the business. Ultimately, this is about putting systems or
processes in place to make sure insights are easily accessible and under-
standable.

Key takeaways

Here’s what we’ve learned about infrastructure requirements:

●● It’s vital you ensure there’s a clear line of sight between your HR
technology strategy and the wider business technology strategy – just
as you would align HR’s strategic goals with the organization’s
overarching strategic goals.
●● The HR technology/IT strategy is like a roadmap that sets out how the HR
function can get the most out of technology. In other words, it looks at
the broader digitization of the HR function, not just the role of data and AI
(although, naturally, data and AI will form a major part of your technology
strategy).
●● While you will work with IT on developing (and implementing) any HR
technology roadmap, during that process your role is to remain laser-
focused on HR’s strategic goals. (So, HR takes the lead on what it wants,
and IT helps to achieve that.)
212 Making data-driven and AI-enabled HR happen

●● When it comes to putting in place the right infrastructure, there are three
layers to consider: infrastructure to collect data; infrastructure to store
(and protect) data; and infrastructure to access and analyse data.
●● Given that most companies have some existing tech infrastructure in
place, it makes sense to consider what you already have in relation to
each of these layers. So, for each layer, do consider what technologies
your organization already has in place. You might have to make changes
and additions to existing infrastructure, but just keep in mind that some
of your existing systems may have a role to play in meeting your data and
AI needs.

We’ve come to the end of our data-driven and AI-enabled journey. So let’s
conclude with a look ahead to the future, and see what’s in store for intelli-
gent HR over the next few years.

Notes
1 K D Schwartz. Tape storage is ‘still here’, IT Pro Today, 7 February 2019. www.
itprotoday.com/backup/tape-storage-still-here (archived at https://perma.cc/
UR52-BF4H)
2 D Palmer. These new vulnerabilities put millions of IoT devices at risk, so patch
now, ZDNet, 12 April 2021. www.zdnet.com/article/these-new-vulnerabilities-
millions-of-iot-devives-at-risk-so-patch-now (archived at https://perma.cc/
J3N6-RE67)
 213

The future of HR 15
The work of HR has been utterly transformed over the last 10 years by data
and, more recently, artificial intelligence. I hope this book has inspired you
to embrace this wave of transformation. To use technology to deliver better
people-related insights, provide a more thoughtful, personalized HR service
and to harness emerging technologies to make HR even more efficient.

Future trends that will shape HR

If you think HR (indeed, the wider workplace) has changed a lot in the last
10 years, you ain’t seen nothing yet. The next 10 years will see yet more
transformation, both for HR and for business as a whole – and this trans-
formation will be driven by AI, digitization and automation. As such, the
HR functions of the future are likely to be radically different as we see new
technologies emerge. And as the technology evolves, so too will society. As a
result, what we regard as ‘workplace norms’ will also evolve. I’m writing
this in 2023, and I can already see that work will look very different in 2033
than it does today.
So I thought it would be fun to close out the book with a few predictions
for the future of work (and, in turn, intelligent HR). I believe some of the
biggest trends to prepare for are:

●● radically altered job roles

●● decentralized and total-remote work structures
●● fully integrated AI systems
●● demand for holistic employee experiences

Let’s explore each one in turn.

Radically altered job roles

In the future, my school-age children are likely to do jobs that don’t yet
exist. Jobs I can’t even imagine. If that sounds far-fetched, cast your mind
214 Making data-driven and AI-enabled HR happen

back to when you were at school – perhaps jobs like blockchain developer,
podcaster, machine learning engineer and social media influencer didn’t
exist then. (They certainly didn’t exist when I was at school.)
Advances in technology, especially automation, will alter job roles drasti-
cally. New jobs will emerge, some jobs will disappear (but more will be cre-
ated in their place) and many jobs will be altered to some degree.
Make no mistake, this is just as true for HR professionals as it is for the
wider organization. The work that you do within the HR function may look
quite different in 10 years’ time. Naturally, the underlying goals and values
will be similar (to serve employees, to help your organization succeed, and
so on). But your day-to-day activities? They will likely evolve over the next
decade.
Because even those jobs that aren’t displaced by automation will use data
and AI tools to get the job done more efficiently. Marketers, for example,
will be able to generate rich (written and visual) content at the touch of a
button. Security guards will be able to analyse masses of security footage for
suspicious activity, in real time. Designers will be able to use generative de-
sign software to create hundreds of potential designs based on their specifi-
cations, within the space of a few minutes.
This is already happening, of course. But the transformation will acceler-
ate over the next decade. HR will need to manage this transition – both in
terms of the HR team itself, and preparing the organization for the future of
work. Reskilling and upskilling will be a critical part of this.

Decentralized and fully remote work

structures
As technology continues to evolve, there will be increased opportunities to
work remotely in a more seamless way – collaborating and sharing with
colleagues as if you’re in the same room. Consider advances such as Meta’s
Horizon Workrooms – a metaverse co-working space where anyone can
come and work. More employers will have virtual offices where employees
can gather (perhaps using VR) for co-working or immersive meetings.
We may even see more companies become fully remote, with teams dis-
tributed across the globe and no one central ‘head office’. Why not? With a
global talent pool and the technology to make remote collaboration smoother,
why should employers choose one spot to set up home? Which means HR
will need to be prepared to manage teams that might never meet in person.
 The future of HR 215

Plus, as organizations become more decentralized, the role of HR might

shift from managing ‘regular’ employees in a traditional corporate hierar-
chy, to managing networks and ecosystems comprising traditional employ-
ees, freelancers, contractors – and, of course, bots and AI systems. The word
‘team’ may mean something rather different in the future. Something alto-
gether more fluid, porous and dynamic than the static, traditional teams that
we’ve grown used to.
These are interesting and exciting prospects. But there will certainly be
challenges in terms of team building, maintaining the right culture and keep-
ing employees happy and engaged. The HR teams of the future will need to
come up with novel approaches to solving these challenges – and, as usual,
technology will have a role to play in that.

Fully integrated AI systems

Increasingly, AI will be ubiquitous across the organization. Including the
HR function. In other words, we can expect AI to be fully integrated into the
very fabric of HR. From recruiting and onboarding talent, to managing per-
formance, training people and delivering an awesome employee experience,
AI systems will be involved at every step of the employee lifecycle.
I’m particularly excited about AI’s potential to deliver a more personal-
ized, predictive HR service. For example, algorithms could provide highly
tailored career development paths for each employee – something that
would be extremely time-consuming and ambitious for human HR profes-
sionals to complete (in a large organization, at least).
Bottom line, in the future, you probably won’t even have to think about
AI – it’ll be there, underpinning everything you do, in one way or another. It
will be utterly enmeshed with the work of HR.

Demand for holistic employee experiences

No doubt you’ve noticed employee expectations evolve over the last decade.
People – especially (but not exclusively) younger workers – increasingly ex-
pect their employer to be diverse and inclusive, to provide flexible working
opportunities, and to treat them as individuals rather than just cogs in the
machine. They expect to feel connected to the organization’s wider purpose
and values. They want to feel enriched by their work. And I’m not talking
about cool gimmicks like nap pods or grow-your-own-veg beds in the car
216 Making data-driven and AI-enabled HR happen

park. Employees want to feel like the work they do matters. That they’re
contributing to something genuinely valuable.
As employee expectations continue to evolve over the next decade, em-
ployees may demand a more holistic work experience that balances their
personal and professional lives. HR will therefore need to build frameworks
that support this balance. As such, the importance of employee wellness pro-
grammes will likely continue to grow. And with advancements in healthcare
technology and wearable devices, HR could provide highly personalized
wellness programmes that track both physical and mental health – poten-
tially even integrating genomic data to provide personalized health plans.
Ultimately, technology has a big role to play in providing a holistic em-
ployee experience. But we can’t ignore the fact that technology also has the
potential to take the shine off an otherwise positive employee experience. If
people don’t have the right technology tools to do their job properly, for
instance, that will sap morale. Or if they aren’t trained properly in new sys-
tems. Or if they worry that an algorithm will take their job (concerns that
may be unfounded), and that they won’t have a chance to upskill and transi-
tion to new work.
So while technology can help HR teams provide a more holistic employee
experience that helps people balance their personal and professional lives,
it’s also important to get the basics right. To ensure people can get the best
out of technology in their job role so they can truly enjoy the awesome em-
ployee experience that you’re creating.

Building the workplaces we want

These future trends are obviously driven by technology, but one thing I re-
ally want you to take away from this book is that we need to keep humans
at the centre of everything the HR team – and the organization as a whole –
does. It’s vital you maintain a tight focus on the people in your organization,
and how you can use data and AI to better serve them.
You have an incredible opportunity before you. An opportunity to use
data and AI to add value and create the workplace that you want. Dare I say,
it’s not just an opportunity, it’s an obligation. I write a lot about future tech-
nologies and I’m always keen to press the point that we should be using
technology to tackle the world’s biggest challenges and build a future that
we want to live in. Work plays a massive part in that. And so, too, does HR.
It’s up to HR functions to harness data and AI to make their organization a
better place to work.
 The future of HR 217

It’s no small task, I know. And I understand that widespread transforma-

tion – which is, after all, what we’re facing – can be incredibly daunting. To
counter any sense of overwhelm, I ask you to cast your mind back to the last
massive wave of transformation, when computerization began to transform
workplaces in the late 20th century. Computers made work and life easier,
created value-adding new jobs, and ultimately made the world a smaller,
more connected place. (If your memory doesn’t stretch back that far, you
can take my word for it.)
Has it been a perfect journey? Of course not. Was that wave of transfor-
mation wholeheartedly embraced by everyone at the time? Absolutely not.
But did the rise of computerization ultimately lead to better, easier lives for
most people? I believe the answer is yes.
I therefore believe this coming wave of transformation will make the
world better. Despite the many challenges that come with technology – es-
pecially AI and automation – I do believe it will continue to make work and
life better for most people. In particular, it has the potential to free up the
human workforce to focus our time and talents where they matter most.
And that goes for HR, too.

Tell me what you think

Now we’ve come to the end of our journey together, I’d love to hear what
you think about the future of intelligent HR. Much as I love writing books
like this, it’s even more important for me to establish a dialogue beyond
these pages. So feel free to ask questions or share your thoughts. Tell me,
how is your workplace being transformed by new technologies? Where do
you see the biggest challenges and opportunities for HR going forward?
How do you envision the workplaces of the future taking shape? How do
you see your own job role evolving?
And, of course, you can always get in touch if you need help planning for
and implementing data-driven, AI-enhanced HR in your organization. I
consult with businesses of all shapes and sizes.
You can connect with me on the following platforms:

LinkedIn: Bernard Marr

Twitter: @bernardmarr
218 Making data-driven and AI-enabled HR happen

YouTube: Bernard Marr

Instagram: @bernardmarr

Facebook: facebook.com/BernardWMarr
Or head to my website at bernardmarr.com for more content (including my
podcast), and to join my weekly newsletter, in which I share the very latest
information.
 219

INDEX

AB InBev 192 anonymous personal data 53, 55, 63, 152,

absenteeism 81, 85, 163 164, 209
see also sick days Anytime Feedback Tool 126
Accenture 112, 119, 121–22 Apple 189
acquihiring 191 Apple Watch 156
action plans 72–73 applicant tracking systems 99
activity data 40–41, 44, 53, 165, 204 appraisals (performance reviews) 81, 115,
adaptability 140, 141 120–28, 132
adaptive learning 136, 142, 143 apps 120, 147, 148, 151, 164, 165, 166
administrative (transactional) work 4, 12, artificial neural networks 25
32, 35, 195 Asana 120
aggregated data 168 assessment (recruitment) sessions 92,
agility 3, 18, 202, 207 103–04
AI (artificial intelligence) 2–3, 13–14, 16–17, attrition (turnover) 4, 14, 20, 82, 83, 84,
19–21, 23–37, 49, 58, 193–98, 215 85, 126
and bias 60–62 see also employee retention (tenure)
in education sector 137 audience identification 71, 86
environmental impact of 58–59 augmented reality (AR) 148, 160, 162
ethical considerations 56–58 automation 4–5, 14, 18–19, 32, 35, 57,
generalized 24 67–68, 193–95, 197
generative 123–24, 150–51 onboarding 107–10
and IP rights 59 recruitment 98–103
onboarding tools 107–14 training 137
performance reviews 122–25 see also robotic process automation;
recruitment tools 98–103 robots
strategy for 65–76, 180 avatars 16, 103, 112, 150
training tools 141–43
wellness tools 165–66 back pain 164–65
see also algorithms; GPT; machine badge technology 30, 44, 118, 130, 204
learning; predictive analytics BambooHR 87
AI-as-a-service 30–33, 203 Bank of America 119, 145, 165
Airbus 119 Basecamp 120
Alexa 17, 181 Beer Garage 192
algorithms 4, 24–25, 45–46, 82 behaviour analysis 20, 138, 143, 157
Alibaba Cloud 207 Bell Canada 118
Alteryx 211 benefits administration 33, 35, 50, 131
Amazon 56, 61, 101–02, 125–26, 181–82, BetterWorks 124
191, 208 Bezos, Jeff 126, 208
Alexa 17, 181 bias 19, 60–62, 67–68, 102, 103, 122,
Amazon QuickSight 211 123, 124
Amazon Web Services 207, 210 BigQuery 210
analytics 23–24, 26–34, 67, 71, 79–89, ‘black box’ problem 58
186–87, 210–11 blended learning 143
see also central data groups; predictive blockchain 15, 16, 34–35
analytics; sentiment analysis; text Bob 87
analytics Boeing 162
anonymous peer feedback (secret peer BP 147, 165
feedback) 124–25, 125–26 brainstorming 176, 177, 182
220 Index

brand image 90–92, 141 consent 20, 46, 51, 52, 56, 131
budgets 180 consistency 59
business cases 43, 53, 73–74 construction sector 160
buy-in 20, 53, 58, 73–74, 128, 157, 197 continuous (continual) learning 141, 196
contrast bias 123
C-suite recruitment 96–97 conversation data 39, 41–42, 44
California Consumer Privacy Act core objectives 69
(CCPA) 51, 52 corporate culture 29–30, 33–34, 91–92,
call centres 4, 29, 30, 119, 194 191, 193–98
call monitoring 42, 44, 52 corporate objectives 68–69, 70, 178
camera technology 38, 43, 44, 144, 148, Coursera 145, 189, 190
150, 158 Covid-19 17, 43
CCTV 14, 27, 42, 43, 44, 204 creativity 104, 140
Campus X 142 credentials verification 35, 99
capability analytics 84 criminal records 50
capacity analytics 84 critical thinking skills 139
Capgemini 103 crowdsourcing 94, 192–93
career (recruitment) fairs 16, 105 cultural intelligence 140
Carrefour 92, 103 culture 29–30, 33–34, 91–92, 191, 193–98
Catapult Sports 116–17 curiosity 141
categorization text analytics 27
Caterpillar 158–59 Daily Telegraph 130
CCPA 51, 52 Dancel Multimedia 119–20
CCTV 14, 27, 42, 43, 44, 204 Danone 142
census data 39, 45 dashboards 72, 87, 159, 165
central data groups 187, 201, 204 data 14–15, 23–24, 37–47, 65, 66, 179
change management 5, 141, 196–97 anonymizing 53, 55, 63, 152, 164, 209
chatbots 4, 14, 17, 32–33, 62, 100, 108, data accuracy 60
109, 151 data ambassadors 189
ChatGPT 31–33, 54, 56, 123–24 data analytics 23–24, 26–34, 67, 71, 79–89,
Cisco 87 186–87, 210–11
Clarks 85 see also central data groups; predictive
Classcraft 150 analytics; sentiment analysis; text
Cleveland Clinical, The 168 analytics
click-through rates 101 data breaches 53–54, 151–52, 209
cloud-based ‘as-a-service’ 202 data collection 43–45, 128, 131, 152, 203,
cloud storage 156, 161, 179, 204, 206–08 204–05
Cloud Talent Solution 101 data customers 179
Cloudera Data Warehouse 211 data insights 80, 179
cognitive testing 99 data literacy 139, 185–88, 196
collaboration 16, 20, 36, 70, 140, 145, data minimization 55
192, 208, 214 data privacy 46, 50, 51, 56, 151–52, 168
collaborative robots (cobots) 17, 194 data protection 54–56
communication skills 140, 184, 197 data quality 59–60
see also dialogue data scientists 70, 94, 187, 188–90
communications data 52 data security 206, 209–10
see also email data data silos 208–09
compensation (salaries) 35, 47, 50, 91–92 data storage 72, 205–10
see also wages data strategy 65–76, 180
completeness 60 data strategy reviews 74
complex decision making skills 140 data visualization 31
compliance 33, 51–52, 79, 108, 179 see also dashboards
computer usage monitoring 119–20 decentralization 214–16
Connected Worker 43 decision making 140, 185
connectivity 119, 123, 161 deep learning 25
see also Internet of Things (IoT) deep neural networks 25
 Index 221

deepfake photographs 26, 46 employment history 50

DEI 21, 46 encryption 34, 55, 209
see also diversity; equity; inclusion environmental impact 58–59
Deloitte 120, 121–22, 195 EnWorks 162
demographic data 45, 81, 108 E.ON 81
descriptive analytics 81–82 equity 21, 47
diagnostic analytics 82 ergonomic office chairs 164
Diallo, Ibrahim 57 ethical awareness 140, 185
dialogue 131 ethics 19–20, 52–53, 56–58, 70, 90–91,
see also communication skills 101–02, 116, 140, 185
digital literacy 139 ethics councils 58, 70
digital threat awareness 139 exam cheating 138
see also data privacy; phishing attacks Excel 23, 186
digital twins 111–12, 149 executive recruitment 96–97
digital wind farms (GE) 149 exit interviews 42, 81
discretion 185 ‘Explain My Answer’ 151
discrimination 5, 12, 19, 61, 103, 130, 168 external data 37, 39–40, 45, 71, 204
discriminators 26
distributed computing 34 Facebook (Meta) 16, 36, 93, 95, 191, 214
diversity 46, 102, 140 facial recognition 27, 46, 61, 138
driver fatigue 158–59 feedback 25, 124–26, 136, 142
due diligence checks 100 feedback loops 25, 124–25, 136
Duolingo 151 financial information 50
duplicate data 60, 206 firewalls 206
first-round interviews 100
Echo 181 Fitbit 163, 165
EcoSpot mortar board 160 fitness trackers 117, 130, 163, 165
education data 50 flat organizational structures 18
education sector 135, 136–38, 151–52 flexibility 5, 140
electronic badges 30, 44, 118, 130, 204 flexible working 18, 20, 36, 215
‘electronic nose’ 158 Fortnite 16
email data 14, 28, 38, 42, 52, 53 Four Dimensional Executive Assessment 96
emotional intelligence 19, 96, 140 fourth industrial revolution 1–2
empathy 96, 140, 184, 185, 194 freelance working 18, 97, 104
employee assistance programmes 20 Fundamental Surgery 149–50
employee engagement 28, 33, 35, 42, 85,
118, 146 GameChanger unit 41, 181
see also net promoter score gender bias 103, 123, 124
employee retention (tenure) 20, 67, 83–84, General Data Protection Regulation
84–85, 126 (GDPR) 51, 52, 55
see also attrition (turnover) General Electric 149
employee satisfaction 3–4, 29, 84–85 generalized AI 24
employee surveillance 116, 119, 128–31 generative adversarial networks (GANs)
see also call monitoring 25–26, 46
employees 17–18, 20, 67, 215–16 generative AI 17, 123–24, 150–51
development of 12, 91–92 see also GPT
and motivation 116, 118, 129, 132 generative networks 25–26
staffing levels 84 geographic data 45
wellness (wellbeing) 162–72, 216 gig economy 18, 97, 140
see also buy-in; connectivity; Glassdoor 91, 188
employee engagement; employee global positioning system tags 27, 125,
satisfaction; employee surveillance; 127, 158
onboarding; ranking systems; safety goal setting 201
monitoring; surveys Golden Gophers 92
employer brand (brand image) 90–92 Google 3–4, 188, 192, 210
222 Index

Google Cardboard 148 Igloo Vision 147

Google Cloud Services 207 image analytics 24–25, 27, 28
Google Cloud Talent Solution 101 implementation plans 180, 182
Google data analytics professional in-house data experts 187
certification 190 incentive schemes 28, 121–22, 123, 169
Google Glass 148 inclusion 21, 143, 150, 215
governance 49, 179 Industrial Revolution 11–12
see also compliance industry-specific providers 191, 204
GPT 17, 31, 59, 150, 151 Infobright 211
see also ChatGPT Information Commissioner’s Office 51
GPT-3 59 innovation hubs 192
GPT-4 151 iNostix 120
graphs 31 ‘intelligence revolution’ 1–2
group break policies 30, 119 intelligent HR 2–4, 65–76
internal data 37, 39, 44–45, 71, 102, 204
handheld computers 127 internal promotions 97–98
Harvard Business School 144 internal recruitment workflows 100
HBX Live 144 International Football Association
head-up displays 119 Board 117
health 141 International Scholarship and Tuition
see also mental health; physical health; Services 85
wellness (wellbeing) internet 15–16, 194
Health at Scale 166 Internet of Things (IoT) 17, 54, 118–19,
health data 50, 167–68 157–58, 164–65, 210
Health Insurance Portability and see also wearable technology
Accountability Act (HIPPA) 51, 52 Internet of Things (IoT) minimization 210
health sector 156, 165–66, 167–68 interpersonal communication 140
see also Fundamental Surgery interviews
heat sensor technology 161 exit 42, 81
Hilton 100 one-way video 100
HireVue 100 shortlisting (scheduling) 99, 100
hiring see recruitment (hiring) IP rights 59
HoloLens 148 IT teams 70, 202
homomorphic encryption 55, 209
Honeywell 43, 148, 159 JetBlue Airlines 95, 147
Horizon Workrooms 214 job application data 50
Hortonworks Data Platform 211 job postings (adverts) 46, 67–68, 99, 101
HR analytics (people analytics) 13–15, 67, job redesign 195–96, 213–14
79–89, 186–87 Johnson & Johnson 67–68, 84, 101
HR analytics reviews 86 judgement 140
HR function Juniper Networks 67
evolution of 12–15
future of 213–18 Kaggle 94, 192
goal setting 201 key fob technology 127
strategy 65–76, 200–02, 208–09 KF4D 96
HR information systems 13 Khan Academy 145
hub and spoke model 187 Kintsugi 166
Humanyze (Sociometric Solutions) 30 Knack 41, 181
hybrid cloud storage 207 Knewton 136
hybrid reality see augmented reality (AR) knowledge leakage 148
hybrid working 20, 21, 140 Korn Ferry 96
see also remote working
‘leadership mountain’ (Accenture) 112
IBM 14, 62, 83, 99, 189, 190, 191, 207 leadership skills 140
IBM Cognos Analytics 211 leadership team 70, 73
 Index 223

recruitment to 96–97 musculoskeletal problems 163

learning 21, 146, 196 see also back pain
adaptive 136, 142, 143 Musk, Elon 27
blended 143 MyMarriottHotel 93
continuous (continual) 141, 196
deep 25 narratives 31
micro 143, 145 natural language processing 31–33, 56
mobile 143 see also ChatGPT
passive 148 net promoter score 84, 91
personalized 136–37, 142–43 networking 109, 141
remote 144 neural networks 25–26
self-directed 142 new hire check-ins 83
learning comprehension 146 new hire turnover 83, 84
learning engagement 146 Nielsen 83
learning management systems 142–43, 145 no-code solutions 30–33
LinkedIn 15, 39, 45, 67, 93, 188 Nsflow 162
location data 53, 87 ‘Nth Floor’ (Accenture) 112
NVIDIA 26
machine learning 4, 16, 24–26, 27, 61, 80,
98, 101, 143 objective setting 68–69, 70, 178
machine vision 56 Oculus VR headset 16
magnetic tape storage 205 office chairs 164–65
majorly transformational projects 177, 178 O’Keefe, Celeste 119–20
manufacturing sector 159–60 on-site data storage 205
Marriott Hotels 93 onboarding 32, 107–14
masking 209 onboarding checklists 108
massive open online courses (MOOCs) 138, onboarding data 108
144–45 onboarding surveys 110
Matrix 35 One Accenture Park 112
medical training 149–50 one-way video interviewing 100
meetings 36, 111, 167 OpenAI 17, 33, 59, 151
Menomonee Falls School District 137–38 OptimEye 116–17
mental health 20, 141, 163–64, 166 opting in (GDPR) 51
Meta (Facebook) 16, 36, 93, 95, 191, 214 Organization for Economic Co-operation
metaverse 16, 35–37, 92, 103–04, 109, and Development 58
111–12, 149–50, 167, 214 organizational skills 185
see also augmented reality (AR); digital organizational strategy 68–69, 178
twins; virtual reality (VR) organizational structure 18, 195
metrics (measurement) 59–60, 83–85, organizational values 91
86, 178 outcomes performance management 121
click-through rates 101 outsourcing 150, 190–91
performance management 50, 116–34
recruitment 93 partnering 190–91
training 142, 145–46 passive learning 148
MGM Resorts 147 password management 54, 210
micro learning 143, 145 payroll 12, 33, 35, 55
Microsoft 16, 62, 82, 145, 189 peer feedback 124–25, 125–26
Microsoft Azure 207, 210–11 people (HR) analytics 13–15, 67, 79–89,
Microsoft Excel 23, 186 186–87
mixed reality see augmented reality (AR) performance management 28, 29, 30, 32,
mobile learning 143 35, 41, 109, 115–34
MOOCs 138, 144–45 metrics 50, 84
motion sensors 130 performance reviews (appraisals) 81, 115,
motivation 116, 118, 129, 132 120–28, 132
multi-cloud storage 207, 208 permissioning 34, 50, 207, 209
224 Index

personal attributes analytics 95–96 recruitment workflow automation 100

personal brand 141 redundancies 57
personal data 39, 49–50, 51–56, 209 remote learning 144
anonymizing 63, 152, 164 remote working 20, 97, 109, 128–29, 167,
personality fit analytics 95–96 214–16
personality questionnaires 99 reporting 71–72, 85–87, 100, 109
personalization 110, 165–66 resume screening 46, 99
learning 136–37, 142–43 reward systems 127–28, 131
Personio 87 see also incentive schemes
personnel management 11–12 robotic process automation 98, 108
phishing attacks 54, 56 robots 17, 155, 156, 158, 194
‘phishing pier’ (Accenture) 112 Royal Dutch Shell 41, 181
phone call monitoring 42, 44, 52
photo data 42–43, 44 safety monitoring 27, 43, 155–62
physical health 164–65 salaries (compensation) 35, 47, 50, 91–92
Pivotal Analytics 211 see also wages
‘plan on a page’ objectives 69 Salesforce 120
Pokémon GO 148 Samsung 16, 103
pre-screening tools 99 SAP Analytics Cloud 211
predictive analytics 82, 95–97, 100, 102, Second Life 36
120, 160, 166 secret (anonymous) peer feedback 124–25,
attrition programs 14, 20 125–26
prescriptive analytics 82 Seeing Machines 158–59
private cloud storage 207 self-directed learning 142
problem solving skills 185 self-driving cars 27
Proctortrack 138 semi-structured data 38–39
productivity monitoring 120, 128–29, 161 sensor technology 27, 43–44, 127–28,
‘Progressive Era’ 12 130, 149, 155, 157–59, 161–62,
project management software 120 164–65, 204
promotions 97–98 sentiment analysis 27, 28–30, 38, 42, 45, 91,
proprietary data 39 109, 163
Prudential 194 Shell 41, 181
public cloud storage 206–07 sick days 164
pulse surveys 91, 124, 163, 164 silos 208–09
push-button key fobs 127 Siri 17
PwC 16 Sisense 211
skills development 19, 86, 139–41, 179,
Qlik 31 184–93, 194–95
QuestionPro 87 skills gap analysis 100, 138–39
quick wins 177, 178, 179, 197 sleep studies 117
QuickSight 211 smart contracts 35
smart hardhats 160, 161
ranking systems 102, 122, 125, 126 smart home thermostats 59
ransomware attacks 54, 62 smart office chairs 164–65
Ready Player One 36 smart watches 165
recency bias 123 smartphones 42, 43, 146, 148, 210
recruitment (hiring) 32, 36, 42, 61, 87, social media 28, 29, 39, 45, 99
90–106, 181–82, 188–89 see also Facebook (Meta); LinkedIn
acquihiring 191 solid storage solutions 205
assessment sessions 92, 103–04 South Korean government 35
job application data 50 specialized AI 24
job postings (adverts) 46, 67–68 speech analytics 29
metrics 83–84 Splunk 211
recruitment centres 36, 92, 103–04 sports sector 116–17, 147
recruitment fairs (career fairs) 16, 105 Sprouts 147
 Index 225

SQL 23 see also employee retention (tenure)

staffing level optimization 84 Twitter (tweets) 38, 217
Stanford Children’s Hospital 147
start-ups 191 unions 130, 131
STEM job advertising 102 uniqueness 60
strategic longer-term (majorly University of Minnesota 92
transformational) projects 177, 178 unstructured data 14, 37, 38–39, 71
strategic partner status 86 UPS 127–28
strategy 65–76, 180, 200–02, 208–09 upskilling 189–90
organizational 178 use case owners 178
stress 163 use case shortlisting 177–78
stress hormones monitoring 17 use cases 175–83
Strivr 147–48
structured data 14, 37–38 vacation policies 81
structured query language 23 validity 60
summarization 27 value add 3–4, 53, 55
surveys 28, 29, 37, 42, 81, 91, 110, 122, values 91
124, 163–64 Varian, Hal 192
Synthesia 150 Vection Technologies 162
synthetic data 45–47, 102, 192, 204 vehicle sensor technology 127–28
synthetic resumes 46 Veriato 119
system profiles 108 Verizon 147
video analytics 27, 42–43, 44, 117, 157
Tableau 31 video content 150
talent management systems 13 video interviews 100
tape storage 205 Villarreal, Blanca Lorena 158
Target 165 Virbela 149
team working 140 virtual classrooms 144
Teamflow 150 virtual help-desks 4
tech incubators 192 virtual meetings 36, 111, 167
technical skills 139, 196 virtual onboarding 111–12
technology infrastructure 72, 179, 200–12 Virtual Park 16
technology strategy 200–02, 208–09 virtual private cloud storage 207
Tenaris 145 virtual reality (VR) 16, 92, 112, 146–48,
Tennessee school district 152 160, 161–62, 166–67
text analytics 27–28, 30, 38, 42 see also avatars; metaverse
text clustering 27 virtual recruitment centres 36, 92, 103–04
Textio 67–68 virtual training 36, 149–50
third-party data providers 39, 55 virtual workspaces 36, 214
time management 141 VirtualSpeech 148
timeliness 60 vocal biomarkers 166
timestamps 27, 34 voice analytics 28, 29, 30, 42
tokenization 209 voice interfaces 17
trade unions 130, 131 voluntary wellness programmes 168–69
training 32, 36, 46, 55, 111–12, 135–54,
159–60, 162, 186, 188 wages 127–28
data science 189–90 see also salaries (compensation)
machine learning 143 Walmart 67, 94, 147–48, 165–66
metrics 85 Walmart Academy 147–48
workflows 109 Watson Recruitment 83
transactional (administrative) work 4, 12, wearable technology 43, 116–17, 118–19,
32, 35, 195 156, 157, 158, 159, 163
transparency 20, 52–53, 58, 103, 128, badges 30, 44, 130, 204
131, 152 see also fitness trackers; smart hardhats;
turnover (attrition) 4, 14, 20, 82, 83, 84, smart watches
85, 126 weather data 45
226 Index

Web 3.0 15 workspaces 36, 214

weighting 25
wellness (wellbeing) 162–72, 216 younger employees 17, 91
wind farms 149
Wisconsin school district 137–38 Zoomi 143
workplace injuries 157 Zuckerberg, Mark 16, 36
 227

THIS PAGE IS INTENTIONALLY LEFT BLANK

228

THIS PAGE IS INTENTIONALLY LEFT BLANK