1.
Perform an analysis of Speech Analytics solutions across SaaS contact center (look at key
competitors, potential disruptors and newcomers), other related cloud software industries
(e.g. Voice of Customer, Customer Experience Management) and how they impact this
potential product. Or any other inspiring source who struck your attention (i.e. niche AI -
artificial intelligence - players). Keep your main focus on Speech Analytics (voice)
competitors landscape;
Speech analytics is frequently adopted by contact centers to identify the proper
interpretation of the spoken words during a conversation between an agent of a contact center
and a remote party. Several systems employed in contact centers in the form of algorithm-based
processing of signals generated by digital conversation to recognize the words and their
meanings. Since the speech analysis is done to interpret the speech. These types of speech
analytics systems are termed semantic-based speech analytic systems (Shaffer, Hayes & D'arcy,
2019).
Generally, the implementations of the current innovation provide a speech analytics
system ("SAS") with computer-program goods, processes, frameworks, tools, and computational
entities that can deliver indications for semantic and non-semantic speech features. These speech
characteristics are identified during the conversation between the agent and remote party and
certain other actions linked with call handling (Sachdev, Jayaraman & Millwala, 2018). On the
other hand, several companies in the market are calming to be the best service provider in terms
of speech analytics solutions. For instance, callMiner has declared that their Eureka 10, is one of
the best speech analytics suite, among others. Eureka 10 is claimed to be the most user-friendly
and intuitive in terms to efficiently making customers. The suite is less dependent on
professional services and more self-sufficient. Similarly, NICE Nexidia has also claimed to be
the only interaction analytics suite that utilizes neural networks of deep learning, allowing the
algorithm of analytics algorithm to offer the maximum reliability for conversational speech in
real-time. With such an ability, it produces word-level voice transcriptions, a grammatical index,
and unprecedented accuracy and speed of customer perception scorings. Genesys also provide a
wide verity of speech analytics solutions. The company does not rely on anecdote, hearsay from
customer or agent or limited sampling instead of it that deliver quantifiable responses, objectives,
and direct points those are required to achieve success in terms of sales. To gain a deeper
understanding of the customer needs, companies have incorporated speech analysis with a
combination of social media, recorded data, and external broadcast network data to develop
enhanced analytics systems (Shim, Koh, Fister & Seo, 2016).
The new entrants to the group of contact centers, with models that took advantage
of all the hardships of those who came before them, like Talkdesk, Sharpen, and CoreDial. These
new entrants came in with cloud-based API frameworks with core contact center features,
complemented by precious add-on market places. Several other new startups have been
witnessed by the market, leading to maximize investment in speech analytics solutions. For
example, Cisco's spent 10 percent equity in Uniphore Software's speech analytics firm in 2016.
Cisco claims that in the area of voice recognition, these voice analytic providers can boost their
awareness of deep technology. Cisco seems to believe that these technologies can grab a strong
market potential. Speech Analytics is becoming incredibly popular in the telecoms sector
because of the enormous volume of informational data is produced by the emergence of the
globalized society definition (Zetsche, Buckley, Arner & Barberis, 2017).
2. Breakdown Speech Analytics key features and define the inputs against it (real time,
batch, type of customer data, interaction data, etc)
Real-Time Speech Analytics
The main advantage of real-time speech analytics is the capability to guide the agents during a
live call. This feature of speech analytics assures that agents will say with product accuracy in
every call they made, either it is an inbound or outbound call. The enhanced delivery mechanism
ensures that agents are utilizing the optimal conversational approaches. So far, this feature
enables high quality and consistent massage interaction during communication with the
respondents. Representatives communicate more often with Real-Time Speech Analytics than
other technologies, therefore helping to enhance conversational efficiency rapidly. This system
must be adequately rolled back. However, with ineffective advice, no representative wants to or
is continuously being alerted (Shaffer, Hayes & D'arcy, 2019).
Batch-processing
The batch-processing interface is used to understand speech just as the WebSocket system as
well as other HTTP frameworks. The batch-processing structure embraces several of the
parameters that any other interfaces provide. With or without speech analytics, batch-processing
functions the same from an interface viewpoint. In the input pool, the service checks and
processes every audio recording. For each input file, it fills out the output bucket with at least
one new (. notice or. json) file. Also, the interface can be used to get speech analytics through
audio files for conversations and speakers. To produce speech analytics, the system further
evaluates the findings of its speech recognition to obtain accurate information on the subjects of
the overall interaction and the sound, mood, keywords, and individual speaker topics (Filonov et
al., 2017).
Interaction Data
Customer interaction analytics is the method of collecting unstructured interaction data
embedded in the sound from recorded conversations, messages, chat recordings, or other
customer conversations and converting it into structured interaction data that can be identified
and evaluated. The first phase of the interaction analytics system is to integrate communications
from the source (call archive, VOIP stream, email systems) and related information like which
operator conducted the interaction, what time and day it took place and who the information
about the customer (Scheidt & Chung, 2019).
Call Metrics Dashboards
Dashboards for call indicators provide a bird's eye view of customer interactions, providing
insight into caller intent. Such panels allow users to conduct a thorough analysis of critical data,
such as the volume of calls, speaking time and non-talk time, spot current and historical trends,
and discover previously undiscovered connections among call topics.
Business Intelligence
Organizations use enhanced applications, and these operations are used to collect, access, and
analyze. Business intelligence systems help businesses develop a broader understanding of their
company. Techniques for business intelligence involve content, research, keyword finding, voice
recognition, emotion detection/feeling, analysis, customer experience.
A. What are the key personas behind the product besides the contact center supervisor?
Within a contact center, a supervisor's duties are to ensure that individual advisers succeed
against targets while assessing their results and encouraging them to achieve so. The role of
supervisors is to identify the evidence of this progress, or lack thereof, by measuring the
performance of the agent over time against set criteria.
Furthermore, if only a percentage of calls are received, the information is not usually definitive
either way, maybe the supervisor has just selected a terrible communication. However, there is
indeed a lot more to the role than that because multi-tasking is quite essential to the job. Team
Members are often required to visit their Boss regularly, For both providing feedback and
receiving orders and actions. Primary roles and responsibilities include, help to advance
procedures, target setting and performance measures for teams and individuals, offer feedback
and guidance, motivate groups and individuals, ensure implementation of the company’s
compliance and policies.
B. How could Artificial Intelligence / Machine Learning (AI/ML) power/drive Speech
Analytics? Why?
Artificial Intelligence (AI) is transforming the world by educating computers or robots on how
and when to learn from information and make choices just like human beings. Companies will be
capable of automating operations with the right deployment plan, better understanding
consumers, reduce costs, and customize the user experience. An additional layer of speech
analytics software is Artificial Intelligence. AI takes speech analytics from just a way of
extracting informational data and building a database for officials of a contact center to scan,
evaluate, and automatically verify the information gathered and offer solutions. Through training
the machine to make decisions that usually interact with human, companies are separating
themselves in many ways from their competition. Deliver customized user experiences, enhance
cybersecurity, 24/7 customer service, rapidly address customer issues, tailor agent training,
replace IVR, analyze calls in greater depth, artificial intelligence is the path forward and forms
improvements in speech analytics systems (Matula & Ponting, 2016).
Most of the leading speech analytics providers are now heading in that direction. For example, in
October, Verint Systems launched automated voice analytics enabled by machine learning.
According to Verint, examining all contacts through specific customer journeys or recognizing
connections from different callers of the same language could help analysts identify the cause of
repeat communications. Then this information will cause rapid corrective action to accelerate
resolution. Calabrio has also been working with machine learning as a means of helping
companies anticipate human behavior, like whether a user who decides to abandon a facility is
bluffing or trying to churn (Korvel, Kurowski, Kostek & Czyzewski, 2019).
C. What are the killer use cases where Speech Analytics could help our Customers. Which
industries do you think would love to be in the early adoption list and why?
Amongst the most exciting developments for contact centers in recent years has been the
growing advancement of analytics software, enabling up to 100 percent of consumer calls to be
analyzed. The following are some use cases in which speech cases are used to resolve the issue.
Eliminate incidences of agent non-compliance and customer fraud
Around 1% to 2% of communications in a typical contact center is received by either a QA team
member or a supervisor. Until something unforeseen happens, such as concerns about incidents
of non-compliance, lost sale, or violations of confidentiality, these call recordings are just sitting
on the shelf. (Filonov et al., 2017). Although listening to them can help to identify what gone
wrong once the problems occurred and the damage has been done. Modern voice analytics tools
can quickly review call transcripts to detect compliance and either alert the agent or turn to a
manager for corrective action. Many approaches may even do that on-the-fly when calls are
ongoing. Understanding that every interaction gets responded to and reviewed for compliance
with regulations, marketing best practices, and privacy regulations helps the company to escape
the reputational and financial harm that may result from violations in any of those fields (Konig
& Ristock, 2018).
Improve agent performance and reduce churn
Many supervisors, as well as QA team members, only get to react to a couple of calls per agent
per week in a contact center. Nevertheless, it remains a significant resource to ensure agents are
on the right path, and continually improve by pursuing their professional development and
training road map. Analytics of the speech can listen to any request. By reacting to phrases and
keywords, ensuring that the agents are following the protocols, and evaluating the language and
attitude of the voice of both clients and agents, each call can be instantly scored against several
criteria. What's more, those regulations will be consistently applied well. As a consequence,
extra training to plug the established knowledge and skill gaps will improve the overall
efficiency. This also aids in enhancing the speed of skills for new employees and to minimize
churn during those first few days that are all significant (Scheidt & Chung, 2019).
Gain insights into customers and competitors
Unless you log all conversations and listen to a small proportion for enforcement and training
purposes, this still ensures that many of the valuable data in those communications are not taken
into consideration. The customers hand over lots of incredibly important details when
communicating to the agents in off-the-cuff remarks about issues like how they use the goods,
how they compare the prices, how open and active after-sale service is, and most significantly
how you match up with the direct competition. These are gold dust information that a company
would usually need to do thousands of surveys to find, and it's right there in the unheard-of call
archives on a server (Matula & Ponting, 2016).
D. How would you fit such a solution in the broader Talkdesk products portfolio?
In the first half of 2019, Talkdesk launches many new products, all motivated by
Talkdesk artificial intelligence. The company Spring 2019 release introduces and extends
product integrations, offers updates on features, and enhances health. The company also raises
customer base and continues a global presence with a focus on R&D. Talkdesk is already in the
business of cloud computing, and contact canter business designing, technologies in speech
analytics are imperative for Talkdesk and need to be in their product range. Talkdesk has
demonstrated its recently launched Talkdesk IQ. Infused across the Talkdesk framework, it
allows customers to use artificial intelligence to boost CX and EX. For example, the company
stated AI-infused labor force monitoring, which maintains the worker at the contact center with
functionality such as time-off requests ad self-service swaps, approachable from the agent
computer or their mobile phone, and predictive analysis for short-term and long-term staffing
decisions (Coimbra, 2016).
3. Define 2-3 metrics to measure the success of your product adoption by our customers.
The apparent fact that many software individuals don't have sufficient adoption understanding
and knowledge. Besides the visitor-to-user classic client, the user to customer rates, there is a
whole diverse area for calculating new characteristics adoption. There are three metrics for
measuring product adoption rates to help assess a new product's performance (Chung & Lee,
2019).
Adoption Rate:
It's the ratio of a range of new clients to total customer numbers. There is a simple math equation
to address the question as to how the rate of adoption should be measured. Number of new users
/ Maximum Users 100. the company has 22 unique users, for instance, and the maximum number
of users is 200: The adoption rate is 22/200x 100= 11 percent. It is measurable on a regular,
weekly, monthly, or annual basis.
Time-to-first key:
The average time a new customer needs to use an existing feature in a contact center, or a current
user to use a unique characteristic for the very first time.
Percentage of users who performed the core action for the first time:
The metric's name exposes its definition. It is the proportion of clients the first performance of
core functionality in a given period. The use of a method as a reference can be useful in tracking
and evaluating SaaS adoption rates. The company can adopt some software that can study the
onboarding process of the new user to identify the phases with which users have trouble.
Analytics tools like Mixpanel, Amplitude, Woopra, etc. are excellent tools for calculating
product acceptance with adjustable funnels and plenty of useful resources.
4. Describe how you would deploy such a solution in an iterative and customer-centric way,
outlining on a high level what you would ship in the short term and long term.
It takes tremendous coordination to make every consumer segment the focus of a custom
interaction effort. The team needs to find a way to have a birds-eye sight of every aspect of the
experience of every single customer. The company needs usage rates, health ratings, and real-
time readable customer statuses so they can respond quickly to every opportunity to develop a
safer and more efficient connection with the customers. The digital business transition has made
SaaS success as of today. However, the ease of access has provided clients the opportunity to
move quickly from one channel with minimum downtime or impact. Maximizing customer
satisfaction and halting the subscription-based business churn rate has become critical to
sustainable growth. To SaaS companies, customer-centered marketing tactics need to become a
way of thinking and not just to work. This means engaging all clients in the customer satisfaction
cycle at the enterprise. It seems that these approaches are so unimportant that ready access to
customer information can't improve it (Scheidt & Chung, 2019).
For SaaS firms, the best customer-centric communication strategies that produce ROI are based
on a company-wide approach. Each member of the customer service team must have exposure to
every piece of customer information. The company needs to be able to visually track the path of
each client through the system and be prepared to respond to the next opportunity of the
interaction (Saltan & Smolander, 2019).
Reference
Chung, J., & Lee, L. (2019). To Buy or to Resist: When Upward Social Comparison Discourages
New Product Adoption. Journal of the Association for Consumer Research, 4(3), 280-
292.
Coimbra, M. (2016). Why an entrepreneur should be a marketeer (Doctoral dissertation).
Correia, P. (2019). Road To IPO (Doctoral dissertation).
Filonov, V., Lum, H., Culbert, J., Blander, D., Subramaniam, S., Cicchitto, A., & Gvildys, P.
(2017). U.S. Patent No. 9,565,296. Washington, DC: U.S. Patent and Trademark Office.
Konig, Y., & Ristock, H. W. A. (2018). U.S. Patent No. 9,912,815. Washington, DC: U.S. Patent
and Trademark Office.
Korvel, G., Kurowski, A., Kostek, B., & Czyzewski, A. (2019). Speech analytics based on
machine learning. In Machine Learning Paradigms (pp. 129-157). Springer, Cham.
Matula, V. C., & Ponting, K. (2016). U.S. Patent No. 9,258,421. Washington, DC: U.S. Patent
and Trademark Office.
Sachdev, U., Jayaraman, R., & Millwala, Z. (2018). auMina™-Enterprise Speech Analytics. In
Interspeech (pp. 1053-1054).
Saltan, A., & Smolander, K. (2019). Towards a SaaS Pricing Cookbook: A Multi-vocal
Literature Review. In International Conference on Software Business (pp. 114-129).
Springer, Cham.
Scheidt, S., & Chung, Q. B. (2019). Making a case for speech analytics to improve customer
service quality: Vision, implementation, and evaluation. International Journal of
Information Management, 45, 223-232.
Shaffer, S., Hayes, S., & D'arcy, P. (2019). U.S. Patent No. 10,270,914. Washington, DC: U.S.
Patent and Trademark Office.
Shim, J. P., Koh, J., Fister, S., & Seo, H. Y. (2016). Phonetic analytics technology and big data:
real-world cases. Communications of the ACM, 59(2), 84-90.
Zetsche, D. A., Buckley, R. P., Arner, D. W., & Barberis, J. N. (2017). From FinTech to
TechFin: The regulatory challenges of data-driven finance. NYUJL & Bus., 14, 393.
