B R I E F I N G PA P E R

Generative AI Success in the

Contact Center Starts with
Knowledge Management

Sponsored by
SPONSOR PERSPECTIVE

Generative AI (gen AI) is reshaping the contact center, driving new

efficiencies and transforming the customer experience. But for its
potential to be fully unlocked, gen AI requires a strong foundation of
relevant, well-organized knowledge. For many organizations, this requires
reevaluating their knowledge management practices.

It’s understandable that foundational work like this can feel like a hurdle.
Consider some of the typical challenges in contact centers today: keeping
essential knowledge base articles up to date often involves tedious
manual work, and insights from seasoned customer service reps can
remain trapped in chat logs, call recordings, and case notes. Self-service
tools frequently depend on rigid, predefined scripts that don’t always
address evolving customer issues. Gen AI offers us a reason to finally
address this issue, and the tools to do so. By intelligently connecting
information sources and continuously refining what’s available to reps and
self-service tools, AI helps build a knowledge management system that
supports everyone.

At Microsoft, we’re focused on delivering solutions like Dynamics 365

Customer Service, Dynamics 365 Contact Center, and Microsoft 365
Copilot for Service that ensure gen AI drives real impact, from allowing
customers to quickly resolve issues on their channel of choice to
empowering service reps with the right knowledge and insights at the right
moment. We’ve experienced the transformational outcomes from AI in our
own service and support organization—from increased collaboration to
lower handle times to faster resolution rates.

And now, with autonomous agents, we’re advancing gen AI to deliver

even greater business value. For example, the Customer Knowledge
Management Agent perpetually refreshes knowledge articles by identifying
update opportunities or creating new content from insights in human-
assisted cases and then drafting and publishing with administrator
oversight, while the Customer Intent Agent autonomously keeps self-
service tools up to date by analyzing customer conversations, case notes,
and transcripts across all channels. It maps emerging issues and solutions
into an intent library, powering chatbots and IVRs with relevant responses
and surfacing proactive prompts for service reps via Copilot. These
agents can work together or independently to ensure accurate, current
information across self-service and assisted channels, transforming
customer experiences.

We’re excited to sponsor this report that highlights how a solid knowledge
management foundation enables contact centers to make the most of gen
AI, with strategies to help organizations build and optimize their knowledge
frameworks. Whether you’re early or further along in your AI journey, we
hope the insights here help drive your own innovation and transformation.
And know that Microsoft is here to support your success.

Learn how Microsoft can help you transform service experiences

with AI here.
Generative AI Success in the
Contact Center Starts with
Knowledge Management

When a customer reaches out to a contact center, HIGHLIGHTS

it’s clear that the interaction represents a unique
opportunity for the company to deepen the relationship.
To support these enhanced contact
When customers can quickly and easily resolve their center functions, companies
questions using self-service, or when customer service will need to apply knowledge
management practices to ensure
representatives can provide rapid resolution and a their data is relevant, organized,
comprehensive and personalized experience and make and accessible.
the process as easy as possible for the customer, it
can solidify the customer relationship and ensure Having access to vast amounts of
enduring loyalty. curated data and detailed customer
profiles will help both service reps
and virtual assistants provide a
Now, companies are increasingly looking to improve contact center operations
more personalized experience.
and the customer experience by automating and augmenting customer service
functions with generative AI (gen AI). And to do so, they are ramping up existing
knowledge management practices to ensure the quality and relevance of the
Helping service reps work more
data driving those gen AI systems. The reason is that it is critical to ensure the
efficiently has a direct impact
knowledge sources are constrained to the information required to feed into
on the customer experience.
the system and that the knowledge is up to date, accurate, and fully accessible
to help achieve the greatest value from the gen AI-driven systems supporting
customer contact. In fact, according to a survey conducted by Customer Contact
Week Digital,1 83% of respondents report knowledge management represents
the primary agent-facing AI initiative for 2024.
And all indications are that increased productivity won’t be the only benefit
of such interaction. “With further personalization, we are able to steer the
customer to the right channel for providing a unique experience. We’ve been
trying to affect that shift, and it is much more pronounced right now,” says
Oana Cheta, a Chicago-based partner at global management consulting firm
McKinsey & Co., where her focus is on gen AI in service operations. “We can

Briefing Paper | Generative AI Success in the Contact Center Starts with Knowledge Management

1
 Harvard Business Review Analytic Services

actually take into account preferences and current behaviors

to predict the best way to resolve [any question].”
Cheta sees three primary business benefits from gen AI
related to greatly improved customer experience, cost savings,
and operational efficiency. The first is reducing the cost base by “Humans can be freed up to answer
lowering call volume through virtual agents that provide real-
time assistance and suggestions. This level of customer self-
the harder questions or talk more
service both reduces service rep workload and improves the to customers about what they’re
customer experience. The second is automation of repetitive really trying to accomplish. [Human
and time-consuming tasks, allowing reps to focus on more
complex work for higher value add.
agents] get much faster access to
The third is professional development for service information they need to solve the
representatives. “You can improve coaching effectiveness customer’s problem. Then they can
with auto-generated, personalized plans for team members,”
she says. “You can provide better training, giving information
use their critical faculties to decide
that is precisely relevant to them [and] making suggestions on if this is reliable information or not,”
the next best action, or the next best product-transforming says Thomas Davenport, professor at
interactions, from simple transactions to value-driven
Babson College, a senior advisor to
relationships.”
To support these enhanced contact center functions, the chief data and analytics officer
companies will need to apply knowledge management program at Deloitte Touche Tohmatsu
practices to ensure their data is relevant, organized, and Ltd., and a digital fellow at the MIT
accessible. Then gen AI can bring together these multiple,
disparate systems and data sources to present those service
Initiative on the Digital Economy.
reps with a unified view of the customer interaction history.
If the data and knowledge sources are not well managed,
the systems will struggle to operate efficiently. When a gen
AI system can assemble and present reps with this unified by as much as $80 billion by 2026.2 And labor expenses often
view, or profile, it can better inform automated systems and account for more than 90% of contact center operational
is a significant help to them in their efforts to provide rapid costs. And there’s more impact for contact centers than just
and accurate service, and it saves the customer from having cost savings. Gen AI systems can expedite onboarding for
to repeat themself or wait for assistance. These profiles can newly hired customer service reps and help them get up to
also help customer service reps anticipate questions and let speed more quickly. Gen AI can also help existing service reps
gen AI provide more relevant responses. provide an enhanced customer experience, which naturally
Through interviews with subject-matter experts, this Harvard improves business value.
Business Review Analytic Services report will help guide Helping service reps work more efficiently has a direct
contact center executives as they augment their operations impact on the customer experience. “Customers can get the
with gen AI systems and apply knowledge management best of what the system can offer and the best of what the
processes and practices to ensure those gen AI systems are [human] agent can offer,” says Thomas Davenport, professor
fueled by the most relevant, organized, and accessible data. of IT and management at Babson College in Wellesley, Mass., a
The report will describe the benefits, requirements, risks, and senior advisor to the chief data and analytics officer program at
best practices that those executives should take into account to London-based global consultancy Deloitte Touche Tohmatsu
let gen AI provide self-service and assisted service to improve Ltd., and a digital fellow at the MIT Initiative on the Digital
contact center efficiency and provide an improved experience Economy. “Humans can be freed up to answer the harder
not only for customers but also for customer service agents. questions or talk more to customers about what they’re really
trying to accomplish. [Human agents] get much faster access
to information they need to solve the customer’s problem.
Cost Savings, Better Service Then they can use their critical faculties to decide if this is
Gen AI is already affecting a significant shift in contact center reliable information or not. Ultimately, we’ll be at the stage
operations and the knowledge management practices that where you can get reliable information from systems without
support them. Stamford, Conn.-based research firm Gartner needing a human intermediary. And that means it’s available
Inc. predicts gen AI will reduce costs for [human] agent labor all the time.”

Briefing Paper | Generative AI Success in the Contact Center Starts with Knowledge Management

2
 Harvard Business Review Analytic Services

“We will see increasingly enhanced experiences and use cases—

hyper-personalization where you’re making that magic moment
more like the norm. That is at our fingertips,” says Oana Cheta, a
Chicago-based partner at global management consulting firm
McKinsey & Co.

Having access to vast amounts of curated data and detailed chatbots for instant messaging, 61% use gen AI to optimize
customer profiles will help both service reps and virtual emails, and 55% deploy gen AI to personalize services for
assistants provide a more personalized experience. “We will customers, such as making specific product recommendations.
see increasingly enhanced experiences and use cases—hyper- Companies also report using gen AI for creating optimized
personalization where you’re making that magic moment more text messages (49%), personalized advertising (46%), and
like the norm. That is at our fingertips,” says McKinsey’s Cheta. website copy (42%). According to that same survey, 64%
A more comprehensive and personalized experience of respondents believe gen AI will improve their customer
is certainly the most significant benefit customers will relationships. FIGURE 1
experience. “Further personalization is really able to steer
the customer to the right channel for providing a unique
experience. We’ve been trying to effect that shift, and it’s Training Data and Human Agents
much more pronounced right now,” she says. Gen AI is fueled by the data fed into the system, so ensuring
Gen AI is also having an impact on the emerging category the accuracy, timeliness, and relevance of that data is critical.
of the contact center as a provided service. Gartner defines Contact center executives must make every effort to ensure
contact center as a service (CCaaS) as a software-as-a-service-
based application that helps customer service reps manage
multichannel customer interactions from both the customer FIGURE 1
and agent perspectives. “CCaaS has had a lot of development
in recent years,” says Cheta. “It has been reaching a level of Gen AI Helps Companies Reach Customers
prominence and widespread applicability. The big thing Companies are deploying gen AI for customer contact
with CCaaS is having add-ons targeting personalization, across all channels
targeting automation, targeting efficiency, [and] targeting
advanced routing, and predictive analytics and gen AI are 73%
helping with these.” Instant messaging (such as chatbots)
This new era of more capable AI technology changes the
61
dynamic in this environment, says Ian Barkin, founding
Emails
partner of 2B Ventures, a Tampa, Fla.-based technology
investment firm. “It’s better at working across multiple
55
enterprise systems and data types. It can assist the [human] Personalized services (e.g., recommendations for related products and services)
agents, it can supercharge the agents, it can evolve the work
the agents are doing. It’s not just better at understanding the 49
words you’re saying. It’s better at interpreting and assigning Text messages
meaning, gauging sentiment and intent, and recommending
next best actions.” 46
Despite gen AI’s advantages and benefits, many Personalized advertising
organizations are still in the early stages of their journey into
the technology. “There are lots of pilots and prototypes, but not 42
Written content (including web content)
much in production. And you don’t get much business value if
you don’t enter something into production,” says Davenport.
36
Systems in use for customer experience cross virtually all
Phone calls
modes of possible contact. According to a Forbes Advisor
survey, 73% of businesses use or plan to use gen AI-powered Source: Forbes Advisor survey, April 2023

Briefing Paper | Generative AI Success in the Contact Center Starts with Knowledge Management

3
 Harvard Business Review Analytic Services

“If you have ten technical notes related to a particular product, you
have to carefully determine which one is most current, accurate, and
relevant before putting it into the system. If you end up loading all
ten, which happens in many poorly managed knowledge management
environments, the gen AI system is not going to be able to figure out
the right answer for the customer or the [human] agent,” says Babson
College’s Davenport.

the data sets driving their AI systems are as up to date, company doing forecasting for field service technicians,”
accurate, and relevant as possible using current knowledge says Jorge Amar, a senior partner in McKinsey’s Miami
management tooling and tagging technologies. office. “We incorporated weather data, as weather is a big
“Companies [employing] knowledge management are using driver when thinking of telecom maintenance and repair.
techniques developed a while ago. You have to make sure your We also used traffic data. Getting from point A to point B at
content is accurate and relatively unique,” says Davenport. “If 9 a.m. is very different than doing it at 3 p.m.”
you have ten technical notes related to a particular product, There is now an approach for storing, managing, and
you have to carefully determine which one is most current, accessing massive quantities of documents called retrieval-
accurate, and relevant before putting it into the system. If augmented generation (RAG), which uses what are known
you end up loading all ten, which happens in many poorly as a vector database and semantic similarity models to call
managed knowledge management environments, the gen AI up needed content. “You simply can’t fit 100,000 documents
system is not going to be able to figure out the right answer into a prompt window or context window, so you have to
for the customer or the [human] agent.” work out a way to get to those documents as needed, and
Contact centers face challenges not only in the volume of that process is called RAG,” says Davenport.
data but also in the variety. “The beauty of contact centers Another approach is autonomous AI agents that are
is that they are massive, deep oceans of data. It’s probably being developed that can go beyond delivering automated
the biggest source of data in any enterprise anywhere on the responses to perform a range of complex customer service
planet, but it is also often the most unstructured,” says 2B tasks, without any human intervention. Gartner describes
Ventures’ Barkin. He advises companies to make “an honest autonomous agents as “combined systems that achieve
assessment of the data you have, the source, and then perform defined goals without repeated human intervention, using
rigorous data curation that informs the algorithms you deploy a variety of AI techniques to make decisions and generate
and the decisions you make.” outputs.” These autonomous agents, according to Gartner,
Knowledge management practices are critical to ensuring “have the potential to learn from their environment and
that accurate and relevant data is fed into any gen AI system. improve over time, enabling them to handle complex tasks.”
Only then can gen AI aggregate and analyze vast amounts As autonomous agents handle increasing volumes of
of data from numerous disparate sources such as knowledge tasks, human agents may be able to focus exclusively on
bases, existing customer data, CRM systems, and others. Yet the value-driven tasks that require critical thinking and
once data accuracy and relevance are ensured, gen AI itself empathy, with autonomous agents also taking on the role of
can help ensure that all data presented to customer service keeping knowledge bases perpetually up to date by analyzing
representatives is accurate, relevant, and comprehensive. case notes, transcripts, summaries, and other artifacts from
“Chatbots and virtual agents are tapping in [to] knowledge human-assisted cases to uncover insights—both updating
management tools to provide accurate answers,” says Cheta. self-service tools and refreshing existing content or drafting
“As you’re leveraging gen AI, you can retrieve information new articles.
faster. You can analyze the content, customer interactions, While the data driving gen AI systems in contact centers will
and feedback and update knowledge bases in real time. You’re come from a variety of sources, much of that will be customer
pulling information out but also contributing back to those data—customer interaction records, purchasing patterns, and
knowledge bases.” address and contact information—so naturally, protecting that
Bringing in external data also helps contribute to a more information is critical. “Step one is to make sure you have a
comprehensive data landscape. “I worked with a telecom private instance of a language model that is not going to use

Briefing Paper | Generative AI Success in the Contact Center Starts with Knowledge Management

4
“The beauty of contact
centers is that they are
massive, deep oceans of
data. It’s probably the
biggest source of data in
any enterprise anywhere
on the planet, but it
is also often the most
unstructured.”
Ian Barkin, founding partner,
2B Ventures
 Harvard Business Review Analytic Services

your prompts you put into it to train future models, which

could well give away proprietary customer information,” says
Davenport. “And you still have to train [human] agents to not
reveal things about customers that might pop up in the system.”
Training human agents is just as important as curating the
data and training language models. “You have to train [service
“One way to create immediate value
reps] in how to create effective prompts and how to identify for contact centers is to create
potential hallucination problems so they just don’t pass along a way for agents to find the right
the problems to customer[s],” he says. “Given the quality of gen
AI systems today, it’s useful to have a human intermediary
information at the right time. I often
looking at the output to determine whether or not it makes see gen AI used to help create concise
sense. That requires a critical perspective that not [a] lot of summaries or statements to make the
agents might have right out of the box.”
Davenport stresses that there’s always the need for
answers found in a semantic search
customized training in contact center applications. “That easier to read and use during real-
means training on knowledge of your products and your time question answering,” Noelle
processes, and that really requires high-quality data,” he says.
“And most companies don’t have high-quality data, particularly
Russell, chief AI officer at the AI
high-quality unstructured data.” Leadership Institute.
Even as customer service reps are learning to get the most
value out of gen AI systems, those systems are also helping
expedite training. “The time to proficiency for a [human]
agent is becoming much faster,” says McKinsey's Amar. “It gen AI used in direct-to-customer communication when they
typically takes anywhere from 10 to 12 weeks from a training haven’t built the safety systems to ensure that the models
perspective [for new agents]. Gen AI is definitely accelerating won’t lie or hurt anyone,” Russell says. “I recommend that
time to proficiency and helping agents reach their highest they first enable gen AI features in their agent dashboard
potential much faster. Companies are doing gen AI-powered and amplify the effectiveness of their agents while they build
training, simulating calls for agents to come up to speed faster.” the practices they need to put in place to ensure safe and
responsible customer interactions.”
Davenport cautions that the potential for complexity or
Managing the Risks lack of clarity in terms of strategic direction can present
While gen AI can streamline the customer contact experience, challenges. “There’s a lot of discipline necessary to really
companies should be mindful of balancing the gen AI and get value from gen AI in contact centers or anyplace else.”
human connection. “The biggest risk is the tendency to over- He advises contact center executives to conduct controlled
rely on gen AI in direct-to-customer engagements,” says Noelle experimentation to accurately determine the best ways
Russell, chief AI officer at the Miami-based AI Leadership to deploy gen AI both to save time and to improve the
Institute. “There are plenty of ways for a company to leverage quality of the customer experience. “Most companies are
generative AI to help their [human] agents, from active call not that good at experimentation, particularly controlled
listening to note taking to sentiment analysis and post-call experimentation,” he says.
summarization.” Organizations certainly show awareness of such risks.
And contact centers can help mitigate those risks by getting According to a McKinsey survey, “The State of AI in 2023:
comprehensive customer profiles in front of customer service Generative AI’s Breakout Year,” of companies using or planning
representatives. Those profiles will be more comprehensive, to use gen AI, the three most relevant risks are inaccuracy
accurate, and effective when they are drawing from data that (56%), cybersecurity (53%), and intellectual property
is well curated and managed. “One way to create immediate management (46%).
value for contact centers is to create a way for agents to find Despite the common perception that gen AI may result
the right information at the right time,” she says. “I often see in widespread job loss, workforce and labor displacement
gen AI used to help create concise summaries or statements only ranked as the seventh highest relevant concern, at 34%,
to make the answers found in a semantic search easier to read with only 13% of respondents actively working to mitigate
and use during real-time question answering.” that potential risk. “AI in the call center space has long been
Putting gen AI in direct customer contact does require viewed as both a savior and a threat. In 2012, the narrative
certain safeguards. “Many times, companies will want to have was effectively that technology is going to put us out of work,”

Briefing Paper | Generative AI Success in the Contact Center Starts with Knowledge Management

6
 Harvard Business Review Analytic Services

says Barkin. “That was 12 years ago, and call centers continue FIGURE 2

to thrive and serve a valuable role in customer service. The

work they do may have changed thanks to better technology,
Not without Risks
but their importance has not. The narrative was one of fear, Businesses ranked what they consider the most relevant
but it didn’t manifest.” FIGURE 2 risks of gen AI

■ Considered relevant ■ Working to mitigate

First Steps 56%
Aligning the strategy and business drivers for moving gen AI 32%
Inaccuracy
from experimentation to production is the foundation for a
successful deployment. “Avoid doing technology for the sake 53
of the technology and really ensure a connection with the 38
business,” says Cheta. “Make sure you’re solving a business Cybersecurity

problem or addressing a business opportunity. When thinking 46

about the gen AI journey, start with [a] strong strategy that is 25
fully aligned with the business. You need to be methodical Intellectual property management
about how [gen AI] is connected to business objectives.” 45
And of course, since data is what fuels gen AI systems, 28
ensuring high-quality data with modern knowledge Regulatory compliance
management practices is critical. “Gen AI is helping move 39
to more automated knowledge bases. You can retrieve 18
information faster ... analyze the content interactions and Explainability
feedback and update the knowledge bases in real time,” she 39
says. “You’re pulling out information but also contributing 20
back to that by bringing knowledge management into the Personal/individual privacy
picture. So you’re not just pulling out knowledge and data; 34
[you’re] also ensuring you’re on a critical path to efficiently 13
and effectively resolving issues.” Workforce/labor displacement
Starting off with a clear focus on precisely how gen AI will 31
fit into contact center operations is also essential. “The first 16
thing I encourage organizations to do when looking to build Equity and fairness
AI solutions for contact centers is to actually spend time in 29
the contact centers,” says Russell. “You will quickly find the 16
areas of friction and pain experienced by the very people you Organizational reputation
are trying to help. Instead of focusing on increasing efficiency
Source: McKinsey & Co. survey, August 2023
in a single call, focus on how we might use gen AI to impact
total time to resolution. This is where gen AI really shines.”
Russell agrees that alignment with the business goals and
values is important for success. “The first question to drive can now be made available through a conversational interface
clarity for gen AI projects is to understand your organization’s that is more accurate and reliable than chatbots of the past,”
core values. This is critical because your systems will amplify says Russell. “Now we can ground the chatbot in information
your core values,” she says. “The second is to evaluate risk to hosted on the existing website or other defined sources of
select the right use case based on alignment with your core information, allowing us to reduce the friction for customers
values and risk profile. The third is to evaluate complexity. It is who don’t want to call in to ask their question.”
important to remember that AI is research; it’s an experiment.
Begin with small achievable goals and test those hypotheses.”
Having more rapid access to a variety of data from a variety Looking to the Future
of sources, as facilitated by gen AI and driven by properly While organizations progressing on their journey into using
curated and managed knowledge bases, holds considerable gen AI to augment contact center operations may already
potential. “Gen AI allows us to more confidently answer issues be seeing benefits, experts contend the future holds much
and questions that come from customers through digital greater potential for the technology. “Over time, systems
channels. For example, any information available on a website will get much better at having [a] dialogue with the agent

Briefing Paper | Generative AI Success in the Contact Center Starts with Knowledge Management

7
 Harvard Business Review Analytic Services

“What if companies decide that part of the value proposition of

the company is actually helping you talk to a human? That could
become [a] strong selling point. These possible consequences could
be realities within the next three to five years and change the way
companies think of customer service,” says Jorge Amar, a senior
partner at McKinsey & Co.

or customer directly,” says Davenport. “I hope companies The business case is going to change,” he says. “Now you
do enough experimentation and process redesign to figure might want customers to have human interaction to help the
out the right mix of human and machine in the customer service rep cross-sell or upsell. It’s an opening to transform
service context.” And of course, these systems will continue to the fundamental economics of customer service. There’s still
improve and evolve. “After you develop these systems, there’s cost reduction, but now also revenue generation.”
a constant maintenance activity in terms of how to input new This shift could even change the value proposition of
knowledge into the system and learn from the service agents customer service. “What if companies decide that part of
and customers,” he says. the value proposition of the company is actually helping
And autonomous agents are set to increasingly help manage you talk to a human? That could become [a] strong selling
knowledge systems and customer interactions, freeing human point,” Amar says. “These possible consequences could be
agents to focus on strategic and high-touch customer service realities within the next three to five years and change the
functions. way companies think of customer service.”
“Under the right conditions, gen AI can create a lot more As companies begin or continue their gen AI journey,
knowledge and accessibility and make it easier for customers one major factor will be how well they apply knowledge
to find out what they need to know without needing a human management practices to ensure the data feeding the systems
to intervene,” says Davenport. is as high quality as possible. “One element is data readiness,”
“It’s going to be interesting to see how adoption levels evolve,” says Amar, “and whether a company has their data ready to
says Amar. He believes there will be what he calls second- and go into the gen AI journey. They need a roadmap for creating
third-level business consequences at organizations making data products and continuing to refine the model, clean up
this shift in contact center operations. “What if the cost per the data, and capture new data.”
call is X? Then you put in solutions to make it 50% [cheaper].

Endnotes

1 Cantor, Brian, “State of Generative AI,” Customer Contact Week Digital, March 29, 2024.
https://www.customercontactweekdigital.com/tools-technologies/whitepapers/2024-march-market-study-state-of-generative-ai.
2 G
 artner, Inc., “Gartner Predicts Conversational AI Will Reduce Customer Contact Center Agent Labor Costs by $80 Billion in 2026,” August 31, 2022.
https://www.gartner.com/en/newsroom/press-releases/2022-08-31-gartner-predicts-conversational-ai-will-reduce-contac.

Briefing Paper | Generative AI Success in the Contact Center Starts with Knowledge Management

8
ABOUT US

Harvard Business Review Analytic Services is an independent commercial research

unit within Harvard Business Review Group, conducting research and comparative
analysis on important management challenges and emerging business opportunities.
Seeking to provide business intelligence and peer-group insight, each report is
published based on the findings of original quantitative and/or qualitative research
and analysis. Quantitative surveys are conducted with the HBR Advisory Council,
HBR’s global research panel, and qualitative research is conducted with senior
business executives and subject-matter experts from within and beyond the Harvard
Business Review author community. Email us at [email protected].

hbr.org/hbr-analytic-services

Copyright © 2024 Harvard Business School Publishing. MCCRE52351124