The Composable Customer Data Platform:

What It Is and Why You Need One

How modern organizations supercharge
marketing efforts with a composable CDP

This is a collaborative whitepaper among Databricks, Hightouch

and Snowplow. We thank Nick King (Chief Product and Marketing
Officer at Snowplow) and Alec Haase (product evangelist at
Hightouch) for their contributions.

In the current era of a privacy-first internet, first-party customer data has never been more valuable to digital
organizations. In an effort to effectively capture, manage and activate customer data, the customer data platform,
or “CDP,” was born. Gartner defines a CDP as “software that collects and unifies customer data — from multiple
sources including first- and third-party — to build a single, coherent and complete view of each customer.”

Here’s the bad news. “CDPs are both one of the most disappointing investments brands made in 2020 and the
number one investment they want to make in 2021,” explained a blog from Databricks and Amazon Web Services
citing Gartner numbers.

Simply put, when it comes to rationalizing and putting customer data to work, there is a lot of hype around CDPs,
and an alarming rate of failure.

The customer
data problem

Marketing teams live and breathe customer data, but While most marketers are confused by their data, they are also
they don’t always trust it. “Nearly 80% of marketing baffled by the range of customer data solutions available to
leaders say data, analytics and insights are very them. Many who try to harness customer data ultimately fail,
important to winning and retaining customers,” according especially as use cases expand. “There is a high risk of failure
to a recent pulse poll of select CMO Council marketing for customer data projects. Organizations increasingly expect
leaders. Every modern marketer wants direct access to customer data platforms to support customer experience
reliable customer data, seeking to gain an advantage over (CX) use cases that fall outside of marketing’s direct control,
their competition, yet in reality, “nearly two-thirds of all such as the next best action for customer service or customer
marketers are only moderately confident (or worse) in journey insights. These use cases depend on, or overlap with,
their data, analytics and insights systems,” as reported capabilities in IT systems such as master data management or
by the CMO Council. cloud-based data warehouses,” Gartner stated.
 Those that solve the customer data problem have a massive competitive advantage over those who don’t.

• 6
 5% of business leaders say their goal • O
 f the 14% with a 360-degree view,
is to build a cohesive data ecosystem 44% say they achieved a 360-degree view
and to standardize data collection. via a customer data platform. (Gartner)
(Treasure Data)
• 7
 5% of CDP customers expect 5x or
• O
 nly 14% of enterprises have a 360-degree But higher ROI, and most see a positive ROI
view of customer data. (Gartner) in the first year of adoption. (Twilio)

• T
 wo-thirds of marketers lack confidence • T
 wo top-reported CDP benefits include
in their data. (Gartner) a unified customer view (88%) and
analytics (54%). (CDP Institute)

Embracing first-party customer data CDPs: Beyond departments to the C-suite

One of the greatest assets of the modern digital organization Many organizations have turned to CDPs to help overcome
is first-party customer data. The rapid rise of the privacy- these significant challenges. While CDPs offer a traditional
centric consumer has led to a monumental shift away solution to collecting, cleaning and activating customer data,
from third-party tracking methods. Organizations are adopting organizations have struggled with their rigid data
now scrambling to implement a data infrastructure that, models, long onboarding times and data redundancy across
leveraging first-party data, can enable the personalized analytics and marketing tools. The CDP Institute’s latest
experiences that customers expect with every interaction. survey found that only “58% of companies with a deployed
CDP say it delivers significant value,” leaving much to be
Those that want to engage customers effectively must build
desired from these packaged CDP solutions.
actionable intelligence on top of their first-party customer
data. Actionable intelligence means improving customer Yet CDPs are essential for managing customer data and
relationships, building consumer trust to capture rich first- powering meaningful customer engagement initiatives. They
party data, and using that data to build intelligence that can are too vital a solution to be bought as shadow IT or kept
optimize the customer experience. under the purview of a single department. While a CDP may
be first leveraged by marketing, customer data is so vital that
Historically, building a rich, behavioral dataset of customer
CDPs expand to more and more use cases, becoming both
interest and intent was difficult. Using that data to make
fundamental and strategic. A CDP done right is immensely
meaningful inferences and predictions about individual
valuable; when done wrong, it results in frustration and
customers at scale was even more challenging. And activating
money down the drain. To avoid this, various C-level execs
that intelligence across a myriad of customer touchpoints
from IT to finance should be invested in implementing a CDP
and marketing channels seemed almost unachievable.
as an element of their broader data management strategy.

C-level execs from IT to finance to data should be invested

in implementing a CDP as an element of their broader data
management strategy.
The good news is that top execs are taking customer data experiences. These platforms combine to form a best-in-
seriously. According to the CX Experience Report 2022, “65% breed alternative to traditional CDPs: the composable CDP, a
of business leaders say their goal is to build a cohesive data category-leading solution for modern organizations that want
ecosystem and to standardize data collection.” to better manage their first-party data strategies and leverage
their existing investment in the cloud data warehouse.
With the rise of the modern data stack, platforms like
Databricks, Hightouch and Snowplow have emerged to This whitepaper will discuss traditional CDPs, the rapid
offer future-proof approaches to behavioral data creation, adoption of composable CDPs, and how the latter can add
data storage and management, personalization and digital transformational value to any modern organization.

The team with the best left-to-right view of their customer data can outmaneuver competition
more effectively.

 hanges in privacy rules make an organization’s ability to understand and take action on its
C
KEY first-party data more important than ever. Embrace it for competitive and customer advantage.
TA K E AW AY S 
There are many ways the implementation of a CDP can go wrong. Often, the purchase decision
is made independently on a departmental level, usually marketing, without C-suite or cross-
organizational involvement.

 C-suite should be active in the scoping and acquisition process for a CDP, demanding a
The
customer data solution that is simple, flexible and extensible.

 n organization’s CDP should be viewed as an element of its broader data management strategy.
A

What is a customer
data platform?
The definition of a customer data platform has
evolved numerous times since its inception in
What is a CDP?
2013. Gartner currently defines CDPs as:
Data Sources Destinations
A software application that supports marketing
and customer experience use cases by unifying Clickstream
a company’s customer data from marketing Messaging
User Database
and other channels. CDPs optimize the timing (Email / SMS
and targeting of messages, offers and customer Content Metadata / Push)
Customer
engagement activities, and enable the analysis Email Logs Data Marketing
of individual-level customer behavior over time. Platform Channels
Marketing Data
Traditionally, digital marketing professionals Websites
Ad Server Data
& Apps
apply a number of tools to create, implement
CDP Logs
and measure their multichannel campaigns.
However, these tools are too often siloed and
 The components of a standard CDP offering:

Data collection Data storage and modeling Data activation

CDPs are designed to collect customer CDPs provide a proprietary repository CDPs offer the ability to
events from a number of different of data that aggregates and manages build audience segments
sources (onsite, mobile applications different sources of customer data leveraging the data
and server-side) and append these collected from most of the business’s available in the platform.
activities to the customer profile. These SaaS and internal applications. The Thanks to a wide array
events typically contain metadata to unified database is a 360-degree view of prebuilt integrations,
provide detailed context about the of each customer and a central source these audiences and
customer’s specific digital interactions. of truth for the business. Most CDPs other customer data
Event collection is typically designed to have out-of-the-box identity stitching points can then be pushed
support marketing use cases such as functionality and tools to create custom both to and from various
marketing automation. traits on user profiles. marketing channels.

do not offer a holistic approach to customer data. “CDPs Traditional CDP: Traditional CDPs, also called fully integrated
promise to solve this by centralizing data collection and CDPs, are better for some organizations than others. One
unifying customer data from disparate sources into profiles,” benefit is the “one vendor, one solution” scenario. If it can
according to Gartner. “CDPs then create and manage check off most of the organization’s boxes, a traditional CDP
segments and push those segments into priority channels. might be a good fit. However, the cons for data teams include
The CDP is not a substitute for an enterprise’s database rigid data models, long evaluation and onboarding times,
of record, but it can ensure that customer profile data, and multiple data silos. In addition, the cons for marketing/
transactional events and analytic attributes are available to operations are vendor lock-in and limited extensibility. If any of
marketing when needed for real-time interactions.” these raise concerns, a traditional CDP may not be a good fit.

CDPs offer major benefits. “The promise is extremely Composable CDP: Also known as a modular CDP, this
valuable – to understand your customers and communicate approach stitches together best-of-breed components into
with them in a consistent way to provide good offers,” an extension of the organization’s broader data management
said Dan Morris, Senior Director of Industry Solutions - strategy. With a composable CDP, teams don’t need to start
Communications, Media & Entertainment at Databricks. from scratch. Any work done acquiring data solutions and
harmonizing data can be pulled forward.
Three routes to CDP – pluses and minuses
Composable CDPs are component-oriented, making
There are three main paths to a CDP: them more agile, flexible and future-proof. When done
well, a composable CDP can deliver greater functionality
DIY: Building a CDP may be a great solution for an
by choosing the solutions best suited for their role in the
organization if it has a fully resourced development team and
composable stack. As the organization grows and use cases
its entire data team is highly involved. Note that it’s critical
evolve, teams have the flexibility to replace the piece of the
to factor in the ongoing support and maintenance required
stack without affecting the whole system. Specific team
for custom-built CDP solutions over time. Before embarking
members (like the data team) can then work within the
on a custom solution, consider answers to questions such
context (and tools) they know and are confident using.
as: What is our staffing plan when key developers leave the
company? What is our CDP product roadmap? Is our CDP Organizational alignment with a data-first composable CDP,
extensible, able to add innovative new functionality that may in conjunction with a strong data mapping strategy, ultimately
have just come to market? helps an organization expand to more use cases faster.
 Why traditional CDPs may
not be the right fit
Most companies struggle with their CDP implementation when they think of the
CDP as a “system that’s off to the side,” effectively disconnecting it from the rest
of an organization’s data stack. This approach is inevitably not aligned with the
The data team
existing data management strategy and leads to countless issues and concerns for challenge:
IT, data engineers, analysts/domain experts and data scientists, all of whom should What you can’t see
ultimately drive the CDP’s success. can’t help you
When a monolithic CDP
Beware of shadow IT causing implementation
comes with its own data
and data group issues
warehouse, how does a team
A common challenge with traditional CDPs occurs when marketing makes these know what’s in it? How do they
purchases in isolation. Often, this results in the marketing team struggling with trust it? In many cases, data
implementation complexity, reaching out to IT months later with little to show. professionals don’t.

“Often a CDP vendor goes straight to the marketing team, convincing them they One marketing professional
can stand up the CDP solution, that they don’t need IT support, and bam, they’re from a major record label
good to go,” Databricks’ Morris said. “Turns out that six months later, it’s a little bit made this point: “I often
more than that. Then IT needs to get involved. And it becomes more challenging. IT heard this from the data and
ends up living with the sins of marketers past.” BI teams who didn’t trust
the CDP’s profile unification
Another challenge that organizations face with their CDP implementations is
model, as traditional CDPs
that traditional solutions often promote the idea of sending data directly to the
are a black box in terms of
CDP rather than the organization’s data warehouse. While this may work for the
modeling. The BI team was
marketing team and their needs, it’s not optimal for the organization and its overall
tasked with reporting on the
data strategies. Data teams end up challenged with using tools native to the CDP
total first-party audience
and managing audience segments in multiple places with multiple “sources of truth.”
using the deduplicated and
Data extracted from traditional CDPs to the data lake lacks granularity, quality,
unified dataset captured
explainability and completeness, and needs significant wrangling before it can be
within the CDP database. In
used for AI-powered data products. Case in point: According to research published
our case, both the CDP and
in 2022 by Snowplow, 91% of data practitioners face challenges in delivering
the BI team used Google
BigQuery, so data syncs
happened in real time
between the two databases.
Consider these questions in any CDP assessment process:
Unfortunately, once the BI
• Are you already building a CDP or rolling out a single-vendor CDP solution? team started working with
the data, they questioned the
• Do you already have a lakehouse feeding and supporting your customer
unification model so much
data efforts?
so that they re-ingested the
• What is the skillset of your current data team?
same data sources like ESP to
• What solutions do they use and how might these fit into your CDP strategy? build their own model.”
 Web & Mobile Events Marketing Automation

Server Events Advertising Platforms

User
Profile
Analytics

CDP

Multiple “sources of truth”

SaaS Tools (CRM, Success)

Advertising Data ETL

Internal Databases BI Tools

actionable insight from customer data. This limits the ability Every enterprise has different customer data needs. This
of data science teams to derive data for adjacent use cases, is one reason why rigid, all-in-one CDPs aren’t always the
such as content optimization and contribution to machine answer, and why the rise of a new class of CDP, a composable
learning models. Compounding these issues, the CDP can best-of-breed solution, could be a better option. The next
become a siloed copy of an organization’s critical asset: section will discuss the extensive benefits of such a solution,
customer data. including not having to start from scratch with existing
infrastructure while extending the capabilities of a lakehouse
It’s critical to engage all the major stakeholders, including the
to support additional use cases. Most importantly, marketers
data and IT teams, and make sure the direction that’s chosen
and data teams will work from the same data warehouse
meets their needs as well, not just the needs of marketing.
without the loss of CDP capabilities.
This will help ensure the CDP implementation aligns with the
organization’s broader data management strategy.

Traditional CDPs offer many features, but they present critical data choices an organ­iz­ation
must be willing to accept or are prepared to overcome for it to be a good ROI.

Be aware: A traditional CDP can become a siloed copy of an organiz­ation’s customer data.
KEY
There are many CDPs to choose from. Get input from IT and data stakeholders to make the
TA K E AW AY S right choice.
 Solution: The Composable “Rather than make a large upfront
Customer Data Platform (and risky) commitment to an
all-in-one vendor, the unbundled
A composable CDP consists of the same components as
its traditional counterparts: data creation, collection and
(composable) CDP allows you to solve
integration. By implementing a best-in-class solution at each the most important problem in front
layer of the composable CDP, organizations can achieve a of you, incrementally. This allows
far more extensible CDP solution that can solve problems you to choose the best solution and
well beyond the common use cases of traditional CDPs.
components for your business. You’re
Understanding each of these components allows teams
to make the most informed architecture decisions when able to educate yourself throughout
implementing their own composable CDP. IT can carefully the process, and also future-proof and
and properly implement each selected component, and swap out certain components down
move faster for less cost without the vendor lock-in of an
the line when your needs change, or
all-in-one CDP.
when a specific tool just isn’t cutting it.”
There are key advantages to a composable CDP, sometimes
Tejas Manohar, Founder, Hightouch
called an unbundled CDP. “You don’t need to buy a bunch
of tools at once. When you buy an all-in-one CDP, you’re
forced to evaluate everything upfront because you’re locking
yourself into a single platform. As the name indicates, the With composable, you can build
unbundled (or composable) CDP is the exact opposite,” on what you have
argues Tejas Manohar, founder of Hightouch. “Rather than
Most organizations likely already have a data management
make a large upfront (and risky) commitment to an all-in-
strategy and a data platform. With a composable CDP, an
one vendor, the unbundled CDP allows you to solve the
organization can align its teams to its data management
most important problem in front of you, incrementally. This
strategy and build upon its data platform. IT and data teams
allows you to choose the best solution and components for
can also move more quickly; instead of starting from scratch,
your business. You’re able to educate yourself throughout
they simply extend what they already have.
the process, and also future-proof and swap out certain
components down the line when your needs change, or when Data professionals like to work with solutions they know and
a specific tool just isn’t cutting it.” trust. When a new CDP is implemented off to the side of an

A composable CDP provides many benefits

CTO / CIO Data scientists & analysts Marketing

• Reduce vendor lock-in • Support for use cases • Consistent segment

adjacent to CDP definitions
• Right tool for the right job
• Contribute model output • Organizational support
• Additive and future-proof
and alignment
• Avoid duplication of
• Data ownership
business rules • Better data quality and
• Reduce TCO
completeness
organization’s existing data platform, issues arise. New and redundant workloads support for new capabilities as they are
are added to already taxed resources. Data engineers have to work within tools needed, instead of buying everything
native to the CDP, not the ones they know and love. Analysts and other domain up front. In terms of ROI, the boost
experts must manage business rules and audience segments with the CDP, when comes from being able to move faster
they are accustomed to using web analytics tools like Adobe Analytics or Google because they don’t have to wait for a big
Analytics. Finally, data scientists love having a single place to get all the customer implementation to enjoy CDP benefits.
data. How exactly do they contribute their models to the CDP?
Creating a customer data platform
When it comes to a CDP, don’t reinvent the data wheel. “Using an internal data that works for a specific business is
warehouse benefits marketers by tapping into the growing data and people one of the smartest and most strategic
resources within other parts of the organization. In the last few years, companies things a leadership organization can
have moved their legacy systems onto a modern data stack. In the process, they do. In the next three to five years, this
have already integrated the same systems as a CDP. In addition, the BI team has is something organizations absolutely
staffed up with analysts and data scientists to mine the data for better business must figure out – and do right.
insights,” argued Lee Hammond, owner of KLH Consulting, on CMSWire. “These
internal resources can deliver far more than reports. They can enable marketing
teams with richer, directly actionable data about their customers directly to all
activation channels—email, advertising, ecommerce and B2B CRM systems.”

The CDP should be part of an aligned marketing, data and IT ecosystem.

When it comes to TCO and ROI, composable wins big

A traditional, monolithic CDP, when done right, indeed offers a positive ROI.
Composable CDPs, when it comes to the total cost of ownership and ROI, stand to
KEY
do even better. Some 75% of CDP customers expect 5x or higher ROI and most see a
TA K E AW AY S
positive ROI in the first year of adoption. And, according to the 2021 CDP Institute User
Survey, top CDP benefits include a unified customer view (88%) and analytics (54%).
Composable means the CDP
doesn’t just start as a blank
A traditional or monolithic CDP can deliver serious payback. Composable offers slate but remains malleable,
ready to respond to new
even more. A composable CDP is cheaper to operate since data teams can get
technologies and changing
up to speed more or less immediately. By using best-of-breed components,
business needs.
organizations get a better return by going composable. A composable CDP is an
effective way to align marketing, data and IT. While marketing is the
primary CDP consumer (at
Consider a marketing team that bought a monolithic CDP, but did not get what was least initially), a composable
promised. Now think about the effort required by the engineering team to make all CDP brings value to other
parts of the organization
the required connections to make it work. A CDP should be about making teams as
through improved
efficient as possible; otherwise, an opportunity cost is incurred if they’re not working
processes, supporting
on their assignments. And if team members don’t support the CDP, that creates a adjacent workloads and
significant cost, especially when their organization is trying to move fast and innovate. avoiding opportunity costs.

And, indeed, the marketing team needs to move fast if critical KPIs are to be met. Think of the CDP as
The truth is, organizations move faster when they don’t start from scratch, and are an extension of the
organization’s broader data
more efficient when they have support and buy-in from all their different teams.
management strategy, not
Composable CDPs can also offer lower total cost of ownership (TCO) and higher simply something that lives
off to the side.
ROI. With composable CDPs, organizations do not have to start from scratch, but
can instead leverage the data solutions they already have. They can also bring in
 A deep dive into the composable
CDP: The art of the possible
In order to realize the full potential of the composable CDP, it’s time to revisit the three main components of the CDP and the
composable, best-of-breed solutions that enable possibilities beyond traditional options.

Data Collection Data Storage & Modeling Data Activation

1st Party Sources 1st Party Destinations

Websites Internal Application

Customer Golden
Mobile Apps Record Internal Databases

Servers

Source Tables
3rd Party Sources 3rd Party Destinations

Advertising Data

SaaS Tools

Behavioral data creation (Snowplow)

(PII) pseudonymization, organizations have a complete and
Data creation is the foundation of a composable CDP.
compliant view of every customer interaction.
With Snowplow’s Behavioral Data Platform, data teams
can manage the end-to-end creation of behavioral data, Once data has been created, Snowplow’s private
collecting first-party behavioral data from every customer deployment model and native connector to Databricks
touchpoint (mobile app, website, server-side, etc.). means an organization’s unified event stream lands in the
Delta Lake in real-time, ensuring it maintains complete
Snowplow offers the ability to define an unlimited number
ownership of its data. Data is modeled out-of-the-box at
of events that can be enriched with any number of entities
an interaction, session and user level, creating a single view
and properties, creating data bespoke to the business
of customer behavior without data wrangling. Data is ready
and providing unlimited opportunities for activation and
to stitch with other data sources or can be used to develop
personalization. The approach to events and entities helps
future BI or AI data products right away.
organizations better understand customer behavior, activate
data and support other evolving use cases within the business. Yali Sassoon, co-founder and chief strategy officer of Snowplow,
summarized the immense benefit of such a solution:
Accurate and compliant identification of users is the
cornerstone of any CDP. Snowplow captures out-of-the-box Our ability to engage with customers and prospects
and custom user identifiers with every event. With in-stream effectively depends on our ability to understand those
privacy tooling, including personally identifiable information individuals: who they are, what they are interested in and
what they are trying to accomplish
right now. Behavioral data is the
fuel that powers that insight — it
reveals not just what decisions those
individuals make, second-by-second
and minute-by-minute, but how they
make those decisions, understanding
customer intent and improving
predictions. There is no type of data
that is more predictive or explanatory,
which is why using this data to
segment customers, directly in your
lakehouse, is so effective.

Storage and modeling

(Databricks)
Creating and maintaining a single view
of the customer delivers a tremendous
amount of value for organizations
big and small. Whether it’s used by a
marketing team to facilitate cross-sell/
upsell opportunities or a product team
to personalize the user experience,
the value of this asset can and should
be realized across all organizational
boundaries. This can be achieved by
using the Databricks Lakehouse Platform
as the storage and modeling layer of an
organization’s composable CDP.
Add an unlimited number of entities and properties to enrich event data.
The benefit of this approach is three-
fold. First, the Databricks Lakehouse
Platform natively supports any
type of data, whether it’s batch or
streaming, structured or unstructured.
This means having a common way
to work with all of an organization’s
data, whether it’s clickstream
data streaming in from Snowplow,
marketing data that is updated in
batches from Fivetran, unstructured
text data from a customer service tool
such as Zendesk, and more.

Second, all this data being managed

in an organization’s lakehouse can
be used directly for BI and ML/AI. For Define model evaluation for hyperopt using MLflow.
example, using Databricks SQL, the return on ad spend for marketing campaigns
can be easily analyzed using an organization’s BI tool of choice. Likewise, with MLflow
and AutoML natively integrated into Databricks, data scientists can easily train and
productionize models, such as propensity to churn, from behavioral data leveraged
from Snowplow, and then use the output of those models for activation via Hightouch.

Lastly, because the Databricks Lakehouse Platform is simple, open and

With every
collaborative, the single view of the customer can be managed and governed in a element in a
consistent and scalable way, for present and future use cases alike.
composable CDP,
Data activation (Hightouch)
organizations
Data Activation is the final piece of the composable CDP. All business teams, from sales
can choose the
and marketing to support and customer success, need relevant, accurate and near best-in-class
real-time customer data to add critical context to the software they already use.
data collection,
Leveraging a technology coined “Reverse ETL,” Data Activation platforms stream storage, modeling
data out from the data lakehouse to any business application. Whether an
organization is enhancing communications with customers via CRM, optimizing ad
and activation
spend with audience targeting, or personalizing email/SMS campaigns, Hightouch tools that fit the
makes data actionable, with no scripts or APIs required.
requirements
Thanks to Hightouch’s no-code audience builder, marketing teams are able to centrally of each of
manage all their cross-channel personalization and targeting efforts within the
platform. Teams can connect Databricks as a Hightouch source and, in minutes, take
their teams.
action on their lakehouse data in any of their various downstream marketing tools.

According to Hightouch’s Manohar, “Data Activation is the method of unlocking the

knowledge stored within your lakehouse, and making it actionable by your business
users in the end tools that they use every day. In doing so, Data Activation helps
bring data people toward the center of the business, directly tying their work to
business outcomes.”

Define the data to sync with SQL or a no-code audience builder.

 The broad benefits
of a composable CDP
By harnessing the power of best-in-class tooling to create a composable CDP, there are four key benefits over a traditional CDP:

Better data governance Although behavioral data can be exported from a traditional

In today’s privacy-conscious world and with ever-evolving CDP, in reality, their data models were never intended to

data legislation, taking ownership and having full control be used outside their platform. CDP data exports from

of customer data is paramount. Rather than a traditional irregular table structures and requires complex joins and

CDP managing all of an organization’s customer data, a transformations before it can be activated.

composable CDP provides full transparency, assurance and

With a composable CDP, data science teams can directly
auditability at each step of a customer’s data architecture.
leverage the behavioral data in their organization’s lakehouse,

The composable CDP is also the central repository where along with Databricks’ enormous data processing capability,

customer data is managed via a single profile that changes to build AI models specific to the organization’s data,

as the customer requirements shift. Controlling what PII is product or business goal instead of relying on the black box

collected, how data is stored and modeled, and what data is models offered by traditional CDPs. With greater model

shared with marketing partners ensures that an organization accuracy, businesses can create additional opportunities

can comply with GDPR, CCPA and future legislation. With a and revenue from their campaigns.

single source of truth confirming who the customer is, their

As mentioned at the beginning of this whitepaper, marketers
profile and their privacy and consent restrictions, controls
have a problem with data quality. “Nearly two-thirds of all
can be more easily managed. Once customer data has
marketers are only moderately confident (or worse) in their data,
been created, whether through a mobile app, website or
analytics and insights systems,” explained the CMO Council.
other means, the customer consent is stored and applied
whenever that customer is engaged. A composable CDP A composable CDP is not only the place where accurate, up-
enables this kind of data governance at scale. to-date customer data resides. It is also where this data is
cared for — where best practices for data quality are applied.

Better results with better data quality

Future-proof and modular by design
Advanced personalization and segmentation of an
organization’s campaigns rely on a consistent source of Composable CDPs are future-proof by design, allowing
well-structured, reliable, accurate, explainable and compliant organizations to avoid the vendor lock-in and one-size-
behavioral data describing what customers are doing fits-all approach associated with any all-in-one platforms.
minute-by-minute. With a composable CDP, an organization With every element in a composable CDP, organizations can
can determine the events and entities that match its choose the best-in-class data collection, storage, modeling
business and decide how its data is modeled for activation. and activation tools that fit the requirements of each of their

Rather than a traditional CDP managing all of an organization’s

customer data, a composable CDP provides full transparency,
assurance and auditability at each step of a customer’s
data architecture.
teams. As the requirements of the business evolve, organizations can continue to
invest on top of their composable CDP, rather than implement a new stack from
scratch and rebuild the business logic, which would entail high risk and cost. The
modularity also opens up opportunities to engage with either internal data science
teams or with third-party data science partners to develop custom AI/ML solutions.
KEY
And last, but not least, organizations have the flexibility to determine their approach TA K E AW AY S
to identity resolution to ensure their team is able to deliver accurate and compliant
The composable CDP
marketing campaigns. The business has complete control over how and when to
delivers a single source
stitch together user identities, leveraging every customer data point available. of truth and better data
governance at scale.
A single source of truth across marketing and other teams Data science teams don’t
Instead of adding another data silo to the tech stack, teams can do more with the need to rely on the black
box models offered by
single source of truth they already have — their lakehouse. With the lakehouse as
traditional CDPs.
the single source of truth for the composable CDP, all teams have access to the
most comprehensive customer profiles and insights from across the business and Composable means no
can activate it through Hightouch with an easy-to-use UI and workflow. vendor lock-in — the
organization’s stakeholders
The single source of truth also has applications outside marketing use cases. It can call the shots on which
also power other use cases ranging from internal reporting to product analytics. component works for
them, not a single software
Darren Haken, head of engineering (Data & Platform) at Auto Trader, gives this provider.
example: “I’ve never seen engineers and marketing people talk to each other
as much as they are right now. All our objectives are now aligned on shared
outcomes,” he said. “Access to first-party data has made our data engineers
have more empathy for our marketers, and our marketers are now more willing to
provide insight into the campaigns they are running.”

Spotlight: Composable CDPs deliver optimal identity stitching

When it comes to a person’s identity, there are many different identifiers. There are known identifiers like PII, email
addresses and so forth, and anonymous identifiers such as device IDs that come off a laptop or smartphone.

Identity stitching is the process of tying together these identifiers to build a more complete view of user behavior. The
aggregation of these identifiers builds a profile that enables organizations to understand their customers, enhance
customer experiences, and engage them with relevant and accurate messaging.

One of the great benefits of a composable CDP is the added flexibility and transparency to identify data and how that
data is stitched together. Rather than the rigid parameters of a traditional CDP, the composable CDP opens up the
possibilities with out-of-the-box stitching (four or more IDs in a single atomic table) and the ability to define customer
IDs (by enhancing the tracking JS) while the stitching logic can be defined at a later date.

The result: Organizations have the freedom to define their identity logic based on their needs and, therefore, more
accurately identify users of their website or application across multiple devices, before and after they log in.
 The composable future:
The expanding role of CDP and CCDPs
Traditional CDPs have made their impact in marketing, and experiences is beyond the sole purview of marketing teams
are aimed at attracting and keeping customers. Currently, and their technology solutions. As CIOs develop new digital
the primary use cases for CDPs include acquisition (57%), products (not just digital storefronts for physical goods), CDPs
customer value (65%) and retention (61%), according to the face a range of new demands from digital commerce, sales,
CDP Institute. But that’s only the beginning. The composable operations and customer service to unify and disseminate
CDP stands to open up a range of use cases beyond the data to power relevant customer engagements. Where the
marketing department. CDP may initially have served marketing teams seeking to
make use of data from elsewhere in the organization, the
“Interest in CDPs has not been confined to marketing: CDP
demand is now bidirectional. Expansion of use cases beyond
search activity on gartner.com among CIOs increased by
those orchestrated by marketing will challenge the CDP in
91% between 2020 and 2021. Executive leaders expect
terms of data governance, replication and workflows.”
that customer profile data will support the rapid digital
transformation pursued by IT, marketing and cross-functional Composable CDPs are potentially the enabler for
CX teams,” Gartner found. “Data to power customer organizations to deliver on these expectations.

Conclusion

Implementing a traditional CDP can be challenging, contact us or deploy Snowplow Open Source. And with
especially at an enterprise level. Because traditional CDPs the launch of Hightouch on Databricks Partner Connect,
revolve around their native customer database, product Databricks customers can establish a secure Hightouch
engineering teams must implement data collection by integration in just a few clicks.
tracking user traits and events across various websites,
backend services and apps via CDP APIs and SDKs.
Implementation can often take three to six months before
marketing efforts can even begin. Further instructions on how to get started with
each of the market-leading composable CDP
The composable CDP allows organizations to solve the most
important problems incrementally, enabling them to choose tools can be found below:
the best solution and components for their business. Teams
can educate themselves throughout the process and future- • S
 nowplow Getting Started
proof and swap out specific components down the line when
snowplow.io/get-started/
their needs change or when a particular tool isn’t cutting it.

Thanks to the partnerships of Snowplow, Databricks and • D

 atabricks Getting Started
Hightouch, getting started with a composable CDP has databricks.com/try-databricks
never been easier. Snowplow’s new Databricks Loader
allows businesses to load directly into both Delta Lake and • H
 ightouch Getting Started
Databricks. To get started with Snowplow and Databricks, app.hightouch.com/signup
 CDPs are giving way to composable
While traditional CDPs can be great, composable is
changing the playing field – and the very definition of
CDP.
KEY
TA K E AW AY S
Why start from scratch?
Don’t replace a data lake that’s trusted (and the data
Vet the solution folks already love) with one that isn’t.
When looking for a CDP, make sure the solution can
successfully harness customer data.
Overcome CDP limits with composable
Composable CDPs are overcoming the limitations of
The C-suite traditional all-in-one approaches.
The C-suite should be highly active in the CDP
acquisition process and choose a customer data
Adapt to change, drive into the future
Composable CDPs are extensible, and easily respond
solution that is agile, future-proof, simple, extensible
to new technologies and changing business needs.
and part of the overall data management strategy.

Harnessing customer data = competitive advantage Escape single vendor handcuffs

With composable, there is no vendor lock-in.
If your competition has superior customer data, even
The business calls all the shots.
your most loyal customers are easy pickings.

Optimize first-party data Extend, don’t replace

A composable CDP is a logical extension of an
Composable CDPs help enterprises collect and
organization’s broader data management strategy, not
optimize first-party data, thus building competitive
a new solution living off to the side.
and customer advantage.

Get Started Today

Discover more about the

Databricks Lakehouse for Media & Entertainment

© Databricks 2022. All rights reserved.