DATA

MASTERY
How data-powered organizations
outperform their competitors
INTRODUCTION
Today’s business environment is both more competitive the right “data behaviors” – the people, processes, skills,
and more volatile. Traditional enterprises are constantly and culture. We further found that these data masters
challenged by digital natives who have perfected the art of outperformed their peers significantly on financial metrics
leveraging data-powered insights at scale, using their agility such as profitability. For FY 2019–20, when compared to the
and data-powered intelligence to outperform their more rest, data masters on an average realized:1
ponderous competitors. At the same time, disruption, and
• 70% higher Revenue per Employee
systemic shocks – including environmental, societal, and
• 245% higher Fixed Asset Turnover
global-health issues – are intensifying in our connected world.
• 22% higher profitability.
Organizations are forced to look towards data – along with
sophisticated analytics and smart technologies such as AI – to In this new research, we wanted to pinpoint what sets these
fine-tune their strategic and tactical responses. data masters apart. We spoke to more than 50 executives
from this group:
However, many are still far from achieving data mastery.
In our previous report on data-powered enterprises, we • Technical leaders in areas ranging from information
assessed around 1,000 organizations across 10 sectors on security to AI and analytics
their ability to derive value from growing volumes of data. • Functional leaders from HR to supply chain.
We found that only a small minority (16%) could be classed as
high-performers – which we call “data masters.” These were We also spoke to executives at big tech firms and digital
organizations that possessed the right “data foundations” natives recognized for extracting valuable insights from data.
– the tools and technologies to utilize data effectively – and Based on this research, we have identified four best practices:

2 Data mastery
 STRATEGIZE ENTRUST MODERNIZE ACTIVATE

Create a data-driven Bridge the data Revamp your Drive data-powered

strategy driven by the trust gap data landscape decision making and
overall business goals actioning

• Align your data • Invest in trusted data • Prioritize value streams • Create a streamlined
strategy with the availability data organization
business strategy • Decommission legacy
• Establish guidelines for in phases • Facilitate
• Have a forward-looking trusted AI data-powered
data collection • Migrate to cloud and innovation across the
strategy • Focus on data integrate multi-cloud value chain
democratization
• Customize • Foster a data-powered
data-discovery tools as culture across all levels
initiatives scale of management

• Accelerate innovation
with DataOps

Source: Capgemini Research Institute Analysis.

In this report, we go deeper into each of the four areas.

3
1. STRATEGIZE
CREATE A DATA STRATEGY
DRIVEN BY THE OVERALL
BUSINESS GOALS
A significant majority of “data masters” ensure that the wider enterprise
strategy drives their data strategy. This helps them set strategic priorities
and ensure foundational elements – such as measurement and collection
– are therefore driving those wider goals:

Align your data

strategy with Define a data strategy that meets
the business the business priorities
strategy

• Ensure a strong top-leadership Have a

support to the data organization forward-looking
• Identify the right metrics data collection
• Invest in data sharing ecosystems strategy

Source: Capgemini Research Institute analysis.

4 Data mastery
1.1 Data masters align their data
strategy with their business strategy

WHY AN ALIGNED DATA

STRATEGY IS IMPORTANT?
Many organizations have a data strategy. What is important, however, is its
alignment with overall business objectives. IT-driven data strategies tend to
focus on IT-centric outcomes, such as reducing the TCO of the data landscape
or migrating to the cloud. While these outcomes are beneficial, they could lose
sight of business objectives. Our previous research highlighted this issue with
only 38% of business-facing executives (as opposed to IT-facing) feeling that their
organization’s data strategy was aligned with the overall business strategy. 2

Data masters define a data strategy Focusing on the specifics of the problem you are solving
can ensure you do not get lost in the sheer volume of data
that meets business priorities available or become focused on the technology itself rather
Data masters start with reviewing the business’s strategic than the end goal. Sameer Amin, global director data driven
priorities, from growth (generating new revenue) to marketing & media at Reckitt Benckiser, outlines how they
operational performance (improving cost efficiency or used a customer-focused approach to ensure data drives
improving sustainability). They then delve deeper into the specific propositions: “What you can end up doing is being
strategic issue to define the data strategy and approach. For infatuated by technology, rather than understanding how
instance, when the objective is to enhance client acquisition the technology can solve the business problem. We should
by X% in a year, the data strategy could be to improve the instead look at what questions we want solved: What if we
quality of leads fed into the lead management system or to had X amount of data or information; what would we do with
make comprehensive leads data available to sales teams. it to enhance the consumer experience?”

Danone’s former global Data & Analytics Officer, Ignacio “For example, with Air Wick, we wanted to understand what
Marinas outlines how Danone balances this activity, saying: types of scents people use in different rooms or at different
“Our strategy or our mission is enabling Danone’s business parts of day. We created a connected product, and, through
to take the appropriate decisions. And, based on this, we are that, we found that people use floral or citrus scents in the
focusing on functions such as finance, sales, and marketing mornings to revitalize, and muskier or lavender-type scents in
and departments that deliver value across the value chain. the evening to help them sleep better,” he adds.
So, the focus is then more driven by business priorities rather
“If you start off with technology first,” Sameer suggests,
than by something else. From our side, the approach is also
“what you end up doing is buying a Ferrari and only driving in
about building the foundational aspects, such as master
first gear.”
data management, data governance, data quality, and data
science.” Aligning the data strategy with the business strategy also
means building data trust among the business teams. As we
see in the section 2, there is a large trust gap that affects
decision making.

5
1.2 Data masters have a forward-
looking data collection strategy

WHY A FORWARD-LOOKING
APPROACH TO DATA COLLECTION
IS IMPORTANT?
Data strategy should encompass the changing regulatory and privacy landscape as well
as focus on the data elements to be collected, including ways to translate these into
business metrics:
• Unless a data element can be tied to a business metric, organizations are
accumulating “dark data,” and incurring costs in its storage (while increasing their
CO2 footprint).
• A data item earmarked for a specific purpose will be managed correctly, presented
as appropriate and documented with the right business and technical metadata for
clarity.
Dark data may also add regulatory compliance costs. Privacy laws surrounding
consumer data require an organization to disclose to consumers how and why they are
collecting certain information. Therefore, clarity on the data collected is essential.

Data masters ensure a strong Regulation [GDPR], and the data hosting itself.
We are focusing on bigger, broader strategies.”
top-leadership support to
the data organization One of the principles of the GDPR states that personal
data should be adequate, relevant, and limited to what is
In addition to the realization that activated data is hugely necessary in relation to the purposes for which they are
beneficial for the organization, changing privacy and processed.3 This not only means limiting the amount of
regulations make the ambit of data collection strategy, data collected, stored, and processed, but also deleting
far reaching. There is growing support for data strategies it when it is no longer necessary. Similarly, in the United
at C-suite level, as Swapna Tom, from the enterprise data States, the California Consumer Privacy Act (CCPA) requires
analytics team at HP, confirms: “The data-strategy team organizations to inform users about the categories of
is tasked by the leadership executive team itself and they personal information they collect and the intended purpose.
have full support from the CEO. And our strategy is forward- Therefore, data minimization definitely has a number of
looking: 10 years from now, rather than three years from advantages:
now. We are looking at cloud strategies, the new regulations
that might emerge around the General Data Protection

6 Data mastery
• Risk reduction: Regulatory and compliance risk Data masters ensure that the necessary collection,
is minimized processing, and reporting processes are in place to enable
• Cost reduction: Reduced storage and processing costs data-powered decision making. When collecting and collating
• Sustainability: Reduced CO2 footprint data from different systems, organizations must be mindful
• Accuracy: Better quality of data for AI and analytics of the granularity required to take meaningful action.
use cases.
Data masters invest in data
Data masters identify the right metrics sharing ecosystems
Reporting metrics should be identified based on business In our connected and networked world, changes that impact
requirements. Data collection and data processing should be suppliers, distributors, or customers will also impact the
directly tied to the business’s needs, as Ram Iyer, worldwide parent organization itself. Sharing data with and using data
director for digital strategy and eCommerce at Microsoft, from partners boosts agility and resilience. As Figure 1
outlines: “In my view, there are two kinds of metrics that you shows, our research found that a significant majority of data
have to look at: performance metrics and transformational masters draw on a wide variety of external data sources, from
metrics. For example, I would say anything that’s “sales” is publicly available competitor data to data from partners and
really a performance metric – how are your sales trending suppliers. Our previous research on data sharing ecosystems
and what’s your mix? An interesting metric from the found that organizations that use external data extensively
transformation perspective would be around an issue like (i.e., using more than seven external data sources) exhibit
consumer usage or renewal behavior in subscriptions.” superior financial performance, i.e., up to 14x higher fixed
asset turnover and 2x higher market capitalization.4

In my view, there are two kinds of metrics

that you have to look at: performance
metrics and transformational metrics. For
example, I would say anything that’s “sales”
is really a performance metric – how are
your sales trending and what’s your mix? An
interesting metric from the transformation
perspective would be around an issue like
consumer usage or renewal behavior in
subscriptions.”

RAM IYER
Worldwide director for digital strategy and eCommerce
Microsoft

7
 FIGURE 1 Data masters use a variety of external data sources to enhance insights

98%
Publicly available competitor data 58%
68%
97%
Open data 55%
66%
Proprietary datasets from 92%
42%
data aggregators 54%
91%
Analyst/Industry reports 41%
53%
Data from hyperscalers 90%
47%
(like Google, Amazon, Facebook etc.) 58%
84%
Data from distributors/partners 42%
52%
77%
Social media data 39%
49%
77%
Data from blogs/product reviews etc. 42%
51%
72%
Supplier data 44%
51%
69%
Consumer usage data 21%
33%
67%
Data from platform providers 34%
42%
67%
Anonymous consumer data 24%
34%

Data masters Others Average

Source: Capgemini Research Institute, Data-powered enterprises survey, August 2020; N=87 data masters, N=500 technology executives.

US-based automotive multinational, General Motors (GM), Reinsurer, Munich Re, developed a team called “data hunters”
has put in place a location intelligence system that listens to tasked with searching for interesting data sources (both
a wide variety of information and generates alerts in case of inside and outside the company) for various use cases.6
adverse events. For instance, if a fire breaks out at a supplier
Clearly, organizations stand to gain by investing in data
plant and is covered by a local news channel, the team
sharing ecosystems. They must create a data sharing strategy
managing this system will be notified of it in near real time.
that identifies the purpose of engaging in data-sharing and
To set up this system, GM worked with its tier-1 suppliers and
identify use cases that create or add value to the organization
their suppliers to collect all the location information. It also
as well as to their partners. Other key areas of consideration
partnered with several companies to collect data from news
include what data the organizations themselves should share
sources and providers of critical weather information. This
and what the ecosystem partners would share.7
system improved event-detection speed and accuracy and
the efficiency of GM’s response.5

8 Data mastery
 Our strategy or our mission is enabling Danone’s business to
take the appropriate decisions. And, based on this, we are
focusing on functions such as finance, sales, and marketing
and departments that deliver value across the value chain. So,
the focus is then more driven by business priorities rather than
by something else...”

IGNACIO MARINAS
Former global data & analytics officer, Danone

The data-strategy team is tasked by the leadership executive

team itself and they have full support from the CEO. And our
strategy is forward-looking: 10 years from now, rather than
three years from now. We are looking at cloud strategies, the
new regulations that might emerge around the General Data
Protection Regulation [GDPR], and the data hosting itself. We
are focusing on bigger, broader strategies.”

SWAPNA TOM
Enterprise data analytics, HP

9
2. ENTRUST
BRIDGE THE DATA TRUST GAP
ACROSS AND OUTSIDE THE
ENTERPRISE
Our research showed that data trust is key to organizational agility, collaboration, and generating value.8 However, the research
also highlighted the massive trust gap between the IT-facing arm of organizations and business units (see Figure 2). Kiran
Malviya, a director at Philips, told us: “This trust gap is very evident across many organizations. As a first instinct, business wants
to challenge the source of the data. They would want to challenge the accuracy, the reliability of that data. And that is why they
always want to triangulate it with experiences.”

FIGURE 2 Business executives do not trust the data

62%
Our business executives trust and
completely rely on the data for
decision making
20%
We entirely trust the data we get

Technology executives Business executives

Source: Capgemini Research Institute, Data-powered enterprises survey, August 2020; N=500 technology executives, N=504 business
unit executives.

A managing director at a cloud provider adds: “I think the partner. IT needs to change that perception by providing
issue of trust goes back to having business at the table value. And, it is not about just building a data store based on
and driving data decisions based on business decisions, a bunch of technical requirements. It’s about getting value to
not technical ones. What tends to happen in traditional the business by helping them become more efficient, helping
organizations is a lot of the business users, especially them to sell more things, or helping them sell new things and
business leaders, see IT more as a cost center than a business create new revenue streams.”

10 Data mastery
Data masters focus on three pillars to establish trust: quality, democratization, and trusted AI:

INVEST IN ESTABLISH FOCUS

TRUSTED DATA GUIDELINES FOR ON DATA
AVAILABILITY TRUSTED AI DEMOCRATIZATION

Source: Capgemini Research Institute Analysis.

2.1 Data masters invest in improving

trusted data availability

WHY TRUSTED DATA

IS IMPORTANT?
Trusted data acts as an accelerator to innovation. To ensure trusted data
availability, organizations must focus on strong data management capabilities and
invest in better data quality, given that quality is a major driver of lack of trust.
The consequences of poor data quality can be enormous, with serious implications
for revenue, operational efficiency, productivity, regulatory compliance,
scalability and, ultimately, trust:
• Revenue loss: Research from Gartner has found that organizations believe poor
data quality to be responsible for an average of $15m per year in losses.9
• Higher costs: 20–30% of operational expenses are directly related to bad data.10
• Productivity: Knowledge workers waste up to 50% of their time dealing with
mundane data-quality issues.11
Our previous research on AI-powered enterprises highlights that “improving data
quality” ranks as the number-one approach that AI-at-scale leaders use to get
more benefits from their AI systems.12

11
How are data masters improving says: “Trust is also about a good alignment on the business
data trust? concept itself. The root cause is not really the data itself or
the way we process the data, but in the definition of the
Data masters focus on strong data management: concept: what is a customer, what is your install base, from
Organizations that are early on in their data-mastery journey when do you understand that it’s really a new customer?”
may not have a complete picture of their data landscape, It is critical that the business itself understands and defines
such as: what “data of good quality” is. Many data quality projects
• Which data elements are digital and which are physical fail because the definition of data quality is monolithic and
• Classification of data for security purposes (confidential, does not consider the specificities of different business
internal, public, etc.) and relationships between teams. Take telcos as an example, where how marketing
data entities. defines good data quality will differ drastically from revenue
• Business meta-data – how data sets have been produced, assurance:
when for what purposes and what use • Marketing will be interested in analyzing overall customer
• Technical meta-data including how data is stored and use to optimize rate plans and next-best offer and will be
where – on premises or cloud. primarily motivated by high-quality customer data.
• Data flow across systems/processes. • However, revenue assurance will be more interested in the
For instance, if an organization promotes its product on a completeness of call data records (CDRs) for all customers
social-media page, the “likes” it receives are also data that can (they will want to analyze all CDRs regardless of quality
inform analytics. Such non-traditional data sources should of the customer data, as they need to detect any revenue
not be overlooked. Organizations must identify and list all leakage).
the data assets they can draw on, who owns the data, and Data quality projects must take into account these business
when it should be deleted in order to comply with regulation. specificities. Otherwise, they risk being irrelevant.
In addition, data masters have well-defined access policies,
including who can access the data and for what purpose. Include quality parameters early on

Data masters build trust by focusing on: Data engineers and data scientists spend a lot of time on
data cleansing and preparation. However, it is important
• Cataloging: Data catalogs store “metadata” and help to remember that data quality means different things to
understand the various data attributes different teams, so masters tend to set a clear definition early
• Business glossary: While data catalogs store data on on so that it’s in place for the entire lifecycle of the data.
data, business glossaries contain “business definitions” Additionally, fixing data in the source system is often the best
of the data attributes and the ‘business rules’ that can be way to ensure effective customer experiences and analysis
applied to this data on the other end of the process. Swapna Tom from HP adds:
• Improving data discovery and lineage: This includes “We make sure that the system of records is cleaned up first,
understanding where the data comes from. because we believe in fixing the source, rather than the
Data masters ensure better data quality downstream.”

Data masters realize that investment in data quality Focus on data stewardship
ultimately helps their business. Therefore, they ensure that Designated data owners must act as custodians of data
business teams take ownership of various aspects of quality quality. Data stewards combine business knowledge
and do not see it as a completely IT-driven exercise. As Ronan and data-management knowledge, and are responsible
Corre, enterprise data quality lead at Microsoft, told us, for everything related to data collection, maintenance,
“Proximity with the business is fundamental because in the organization, and compliance. Also, all users and generators
ocean of data, ‘boiling the ocean’ becomes a risk. So, focusing of data are responsible for its quality at data master
on areas most impactful for a user is key.” Data quality ranges organizations such as Microsoft. “We want to create this
across a spectrum of issues: culture that quality data is a everybody’s matter,” says
• Data-sourcing quality issues (“garbage in, garbage out”) Delphine Clement, director business program manager,
• Technical and business-quality issues (for instance, the Enterprise Data Management at Microsoft. “Everybody in
definition of a “prospect” could differ in different systems) the data supply chain needs to be accountable for that clean
• Data-veracity issues (accuracy, precision, and house because everybody benefits from it. For example,
trustworthiness of data). when we, as central data stewards, audit the quality of data,
we always differentiate adherence to standards between
From our research, we found that data masters focus on the data which has been created over the past month versus
following areas to improve data quality and reliability: historical data. This is because we want to make sure that
the recent streams of data created is better in complying to
Clearly defined nomenclature
our standards than the historical data set, which is then a
Jean-Luc Clarot, director architecture, data and program testimony that our initiatives for establishing a data quality
management at Proximus, a large telco based in Belgium, accountability culture work.”

12 Data mastery
LEARNING FROM DATA MASTERS:
USE OF AI FOR DATA HYGIENE
Fixing the data-integrity issues at point of collection (or soon after) helps to ensure data hygiene and provide quality data
for business intelligence. However, maintaining data integrity is often a challenge when dealing with scores of external
siloed data sources. Issues such as transfer error, human error, or misconfigurations can make the integrity-checking
process arduous. Organizations such as Microsoft are turning to AI to help.

“Our sales data, from our eCommerce partners worldwide, comes into data engine MS
Sales which is the single repository” says Microsoft’s Ram Iyer. “Once the data comes in,
Use AI to check it is important to do integrity checks. This is where we can use the power of AI. Now, you
data integrity have really good AI tools that can scrub data and potentially measure and improve the
quality of data with a data-hygiene index.”

2.2 Data masters establish guidelines

for trusted AI

WHY ETHICAL AI IS
IMPORTANT?
As we have seen, lack of data trust is hindering businesses in capitalizing on data
insights. While the quality of data is one factor that impacts data trust, other issues
include the “black-box” nature of AI algorithms and possible inherent bias. Unless
business executives have a clear understanding of how AI works – and are confident
in the fairness of the AI systems – they may not be able to trust the output of these
algorithms.
Furthermore, the past few years have seen growing concern around the ethical aspects
of AI. By analyzing 500,000 blogs and 60,000 English news stories, Stanford University’s
2021 AI Index report found that articles related to AI ethics stories topped the list of
most-covered AI topics in 2020.13 This showcases the growing interest in ethical AI.
For more details on how we define ethical AI please see our Code of Ethics for AI (see
also Appendix 2).
In addition to executive trust, another critical aspect of trusted AI is the growing
awareness of ethical concerns among consumers in their use of AI systems. For
instance, our research on “ethics in AI” showed that 71% of consumers want an
organization’s AI systems to be able to clearly explain the results it arrived at, and 67%
expect organizations to take ownership of their AI algorithms when they go wrong.14

13
How do data masters ensure could mean creating hotlines or channels for employees to
speak about various ethical issues they may face during the
their AI is trusted? design, development and implementation of AI systems.
Data masters set up robust processes for governing AI
Set up processes around AI design and development
Ensuring AI solutions are trusted involves a series of actions.
Having a set of guidelines and processes around AI design
Some of the best practices that data masters follow include:
and development will be beneficial for the organization in the
Define an AI charter long term. Such practices include:

Organizations must determine what AI ethics means • Laying out the purpose of an AI system and communicating
to them, and how they can define and enforce it. While it transparently to all the stakeholders
regulation provides guidance, it is up to organizations to • Identifying and preventing accidental or malicious misuse:
define their AI codes of conduct. Nicolas Economou, co-chair, In 2014, Amazon started working on AI-based program for
Law Committee of the IEEE Global Initiative on Ethics recruitment to review the resumes of applicants and select
of Autonomous and Intelligent Systems, and CEO of H5, the top profiles. However, the AI results were gender-
explains: “Beyond what you must comply with, you need to biased and after failing to guarantee that the program
determine what you stand for as an organization. What are would remain non-discriminatory, the AI system was
your brand values? You can then define your ethical AI code disbanded altogether.18
on the basis of what you stand for; what that implies for how • Ensuring accountability of AI systems – human oversight
you think of the impact of your decisions on your company, through mechanisms such as human in the loop (HITL),
employees, customers, and society at large.”15 human on the loop (HOTL) or human in command (HIC)
helps in maintaining control and accountability.
Mohsen Rezayat, chief solutions architect at Siemens Digital • Building diverse and inclusive teams – Lack of diversity in AI
Industries Software, says: “Ethical considerations need to be teams often causes any inherent bias to remain hidden. By
thought about at every stage of product development. We bringing in diversity into the design and the development
are investing in research to enhance robustness and security, teams, both in terms of people as well as in discipline,
explainability, safety, privacy, and accountability (including organizations can make fair AI possible.
guarding against harmful bias and unintended consequences) • Auditing AI systems for adherence to the AI charter or the
to make AI solutions ethical and trustworthy in a B2B code of conduct.
context.”16 • Training programs for sensitizing both developers and
Establish leadership and governance management on various ethical aspects of AI. Google, for
instance, trained 5,000 customer-facing cloud teams in
Data masters establish a position that oversees the data asking critical questions to spot potential ethical issues.19
and AI ethics. For example, H&M, the Swedish clothing-
retail company, appointed a head of AI policy and created a Use tools and frameworks to develop ethical AI systems
checklist that is used for all ongoing and new AI projects. This While an AI charter helps guide the teams, equipping the
checklist is centered around nine areas: focused, beneficial, development teams with tools and frameworks helps
fair, transparent, governed, collaborative, reliable, respecting them implement the concepts of ethical AI. For instance,
privacy, and secure.17 Such a position helps define the AI our research shows that data masters rely on tools and
charter and drive the ethical use of AI across the organization. frameworks that help them in auditability of AI. As Figure 3
Data masters also establish mechanisms to raise and address shows, four in five data masters rely on such tools, compared
the ethical concerns around the development of AI. These to just 33% of others.

FIGURE 3 A majority of data masters rely on tools to help the auditability of AI algorithms

78%
Our organization uses tools that help in the
auditability of the algorithms, instead of being 33%
treated as a "black box"
41%

Data masters Others Average

Source: Capgemini Research Institute, Data-powered enterprises survey, August 2020; N=87 data masters, N=500 technology executives.

14 Data mastery
Explainable AI-as-a-Service is offered by Amazon, Google, • A score explaining how much each factor contributes to
and Microsoft among others. Google’s Explainable AI, a set the model predictions
of tools and frameworks to help analysts understand and • A “what-if” tool to help investigate model performances
interpret predictions made by their machine-learning (ML) for a range of features in the dataset, optimization
mode, includes features such as:20 strategies, and even manipulation of individual datapoint
values.

2.3 Data masters focus on data

democratization
“Democratization” of data is the ability to provide the required data at speed and in the right form and granularity, thereby
allowing the business users to explore data and derive actionable insights without waiting for the technical teams. This not only
empowers the business but also makes them more accountable for driving data-powered decision making.

WHY DEMOCRATIZATION OF
DATA IS IMPORTANT?
• The ability to access and understand data translates into faster decision making
and greater agility. German pharmaceutical company, Boehringer Ingelheim, has
democratized data by sharing metadata across clinical-trial phases to provide real-
time data to researchers. It has since achieved a better and faster data flow in its
clinical-trial processes and accelerated its drug development pipeline. 21
• Arming customer-facing departments, such as customer services and sales, with
accurate customer data allows them to deliver a seamless customer experience.
• Democratization creates more empowered, engaged, and accountable employees.
Bernard Marr, author of Big Data in Practice (2016), says: “When you allow data
access to any tier of your company, it empowers individuals at all levels of ownership
and responsibility to use the data in their decision making.”22
• And finally, it is key to establish data trust. Himanshu Shekhar, vice president,
enterprise data, analytics and insights at PVH, a US apparel company that owns
brands such as Van Heusen, Tommy Hilfiger, and Calvin Klein, comments: “Trust has a
lot to do with data transparency – with data democratization. The lack of trust from
senior leaders comes from the fact that they are reliant on individuals to provide the
information to them, and then different individuals start providing different pieces
of the puzzle, with different filters or in a different interpretation. This leads to
executives lacking trust in the data or a desire to extrapolate results themselves.”

15
How are data masters democratizing data? wind farms. Then, we also have a subsidiary in Hartford in the
US that specializes in IoT and in monitoring wind farms for
Data masters empower businesses through self-service: other customers, and we have a weather team in Houston
Through the availability of trusted data, data masters strive that is expert on wind and climate. Those three groups
to make the business teams agile. Andreas Kohlmaier, head of came together and combined their knowledge about their
data engineering at Munich Re, the multinational insurance experience in running wind farms, IoT, the sensor data, and
company, outlines an example of how their group-wide data the weather. They combined it in a new service product for
platform promotes collaboration and innovation, saying: anyone who plans to invest, run, and build a wind farm.”23
“One of the things that happened recently is a use case As figure 4 shows, data masters invest in training business
about offering cover for wind farms. On one side, we have users in self-service analytics.
our asset-management team that actually owns and runs

FIGURE 4 Data masters train their business users in self-service analytics

68%

Business users are trained on self-service analytics 24%

30%

Data masters Others Average

Source: Capgemini Research Institute, Data-powered enterprises survey, August 2020; N=72 data masters, N=504 business executives.

Some best practices emerged from our discussions: • Focus on education as a means to enable self-service:
Data masters who embark on a journey to provide self-
• Do not consider self-service analytics as a
service data to business users do not consider intuitive
one-time project:
tools a replacement for training and education around
GE-Aviation, a provider of jet and turboprop engines, uses
the initiative. The self-service data (SSD) program in
two permanent teams that support their robust self-
GE-Aviation focuses on education to obtain high adoption
service initiative:
among business users, as well as sustained use of the tools
– The Self-Service Data Team, responsible for user
and program over time:
enablement, tool administration, and identifying
– Onboarding: Regardless of role and background, all GE
opportunities for process automation
SSD users go through the Digital Data Analyst (DDA)
– The Database Admin Team, responsible for data
training as a part of onboarding
governance and user support.24
– Ongoing: In addition to DDA, the SSD team also spends a
• Clearly define a glossary of terms for business users: lot of time on improving materials on ongoing education
Kurt Muehmel, chief customer officer for Dataiku – an and training
artificial intelligence and machine-learning company – Gamification: The team launched a “Data Duel”
providing platform for democratizing access to data – told challenge – where users were given points for activities
us: “In many organizations, generally among business such as tagging a dataset, creation of new dataset
users, there is a problem of understanding the context – or documentation, etc. More points unlocked the
where the data is coming from, why the data is structured possibility to pass levels and get exclusive laptop
the way it is, and what would be appropriate uses of that stickers. Leaderboards were also used to keep the
data. So, I think the solution here is ensuring that alongside interest and engagement high.25
the data itself, organizations have embedded the context
and the understanding necessary for how to use it.”

16 Data mastery
Data masters focus on data visualization and visualization tools have been key to data masters’ fostering
storytelling: data citizenships, so that business users can create a narrative
based on the data. Data masters emphasize the importance
Data-visualization solutions are the tools used to convert raw
of widespread data literacy to build this “data-first” culture –
data into visuals that are easy to interpret by the business
as 87% of data masters expand their data-literacy programs
– consequently, they are critical to data-democratization
beyond the subject-matter experts. As Aleksejs Plonitkovs,
efforts. They can be called the “face” of self-service analytics.
enterprise data director at Microsoft, explains: “We put a
Data masters employ these visualization tools to make
significant effort into educating our business communities
data-powered insights understandable and actionable to
and our business users about the importance of data,
everyone.
governance principles, and the importance of having clean
Data masters improve data citizenship through reskilling datasets. We call this as ‘efforts towards data literacy.’ We
and training: want to be sure that every Microsoft employee is aware of
the principles of data governance and data management and
We define “data citizenship” as understanding the inherent how to sustain high quality of the data.”
potential of data, interpreting it accurately, and driving
decision-making and business users with this agency as “data
citizens.” As mentioned above, self-service analytics and

We put a significant effort into educating our business

communities and our business users about the importance of
data, governance principles, and the importance of having clean
datasets. We call this as ‘efforts towards data literacy.’.....”

ALEKSEJS PLONITKOVS
Enterprise data director,
Microsoft

In many organizations, generally among business users, there

is a problem of understanding the context – where the data is
coming from, why the data is structured the way it is, and what
would be appropriate uses of that data. So, I think the solution
here is ensuring that alongside the data itself, organizations
have embedded the context and the understanding necessary
for how to use it.”

KURT MUEHMEL
Chief customer officer,
Dataiku

17
3. MODERNIZE
REVAMP YOUR DATA
LANDSCAPE
3.1 Data masters modernize their
data estate

WHY REVIEWING THE DATA

ESTATE IS IMPORTANT?
In an omnichannel world, customers expect to move seamlessly across offline
and online touchpoints. A modern technology platform boosts the organization’s
ability to respond to shifting customer behavior and generate actionable, timely
insights. It includes the ability to combine data from multiple internal and external
sources and deliver insights at speed and scale.
CarMax, an automotive retailer, has adopted many leading practices in software,
including continuous integration, continuous delivery, application programming
interfaces (APIs), and microservices-based architecture. It enabled the
organization to improve code-release frequency from once every two months
to multiple times a day. This constant improvement allows the company to roll
out products, updates, enhancements, and fixes to market much more quickly,
improving customer experience.26

How are data masters modernizing their data estate?

Most companies have already accepted the need for digitally enabled IT systems. However, the required multi-year
transformation often costs tens of millions of dollars and leads to loss of competitive position. Also, data-platform tools and
technologies evolve rapidly, and companies need to make crucial build-versus-buy decisions to keep their investments aligned
with the technologies they need.

18 Data mastery
Data masters focus on the following areas to deliver business priorities:

MIGRATE TO ACCELERATE
PRIORITIZE CLOUD AND INNOVATION WITH
VALUE STREAMS INTEGRATE DATAOPS
MULTI-CLOUD

CUSTOMIZE
DECOMMISSION DATA-DISCOVERY
LEGACY IN TOOLS AS DATA
PHASES INITIATIVES
SCALE

Source: Capgemini Research Institute Analysis.

Data masters prioritize value streams (products, services, or a portfolio of these)

As we saw in section 1, data masters ensure alignment between the business strategy and the data strategy. The business
objectives can be aligned with data priorities by mapping value streams over the modernization efforts. To achieve this, they
focus on the following elements:

• In data masters, these value streams emerge from larger business priorities,
such as expanding market share, reducing platform costs, improving customer Identify and prioritize
engagement, and automating processes. They also help in early value value streams that are
realization. most impactful for
• Data masters logically group applications and infrastructure required to launch the business
applications, and provision datasets required for prioritized value streams.

• At an early stage, data masters remove legacy dependencies and

Use value streams to
abstract applications using APIs for delivering benefits.
pinpoint system
• They also identify and fill gaps in capabilities necessary to deliver dependencies
those value streams.

19
Prioritization of these value streams helps bring the business transforming data for use in the new systems. After this, they
and technical teams onto the same page, and governance divide the monolith into containers of larger components,
structures should therefore include representatives from which can be used independently.
both sides. The overall estate modernization roadmap should
Canadian eCommerce retailer, Shopify, has adopted a two-
be linked to the delivery of such value streams as milestones
pronged approach to decoupling its legacy systems. It started
to ensure strong alignment with business objectives. Further,
out by drawing boundaries through the monolith, consisting
data masters ensure that they have a backlog of these value
of related components. Gradually, it strengthened those
streams with defined priorities.
lines to create full-fledged boundaries with a stewardship
Data masters decommission legacy systems in phases team assigned for each component. Secondly, it carved out
isolated and clean components from its legacy codebase,
Companies that have worked with monolithic systems for
where relevant.27
years cannot practically uproot these and replace them with
modern platforms. Companies start out by mapping relevant
transactional and historical data with the new systems and

LEARNING FROM DATA MASTERS:

DECOUPLING WORKSTREAMS FROM
LEGACY IT
DoorDash, a US-based online food ordering and delivery platform, shifted its pricing algorithms out of the monolith to
quickly test new pricing components.28

DoorDash’s pricing algorithm was implemented in a legacy monolithic codebase, which

caused a number of issues:

1. When customers checked out with multiple items in their cart, each line item had
duplicated implementations dispersed throughout the codebase.

2. Legacy code became unwieldy and difficult to understand over time, owing to technical
Challenge debt (rising complexity of algorithms over time) and multiple engineers working on it.

3. The legacy systems were not sufficiently scalable to accommodate the increase in
traffic.

4. Owing to tightly integrated modules, platform-wide consistency was missing when

calculating prices. Introducing changes in pricing algorithms was complex, as it involved
altering tightly coupled components across a large codebase.

To address these issues, DoorDash separated the pricing logic out of the monolith using
microservices architecture. Creating a pricing service as a platform led to a centralized
Solution and standardized framework that allows engineers to implement and test their new
pricing components and use cases clearly and rapidly.

20 Data mastery
Data masters manage data as a strategic asset across • Large organizations can also have multiple data stores
multi-cloud environments that cater for the unique needs of different product
groups or functions. These data stores also connect to the
Cloud migration is already a priority at many large
enterprise-wide data store. As Aleksejs from Microsoft told
organizations. Appetite for cloud-based systems is driven by
us: “Our enterprise data lake comprises a lot of different
a range of advantages:
sources, albeit not everything. There are multiple data
• They are not only cost-efficient but help drive innovation. lakes because of the size and of specialty. For example, we
• They enhance analytics capabilities at scale; increase have dedicated financial data lake which is more focused
development velocity; and harmonize datasets across on the financial data, as well as own repository of data
multiple environments to provide a consistent view rules, governance principles, metadata and taxonomy
of business. definitions for the company.” Ensuring that the definitions
are standard across these data stores is critical so as to
Seeing these benefits, US-based CapitalOne Financial, for avoid misinterpretation of the data.
example, recently moved 100% of its data to the public
cloud.29 Plan for interoperability within multi-cloud
environments: Companies usually select a keystone cloud
Similarly, US-based insurance group, Nationwide, moved vendor to migrate their systems, code, and applications
over 850 business and customer-facing applications – such to public or private cloud environments. They also keep
as claims, personal, and commercial insurance policy systems peripheral vendors to get the best of capabilities from each
– to cloud. With this, it could deliver deeper insights to its vendor, reduce concentration risk, and meet regulatory
independent agents, financial professionals, and customer- requirements.
service associates to provide tailored recommendations to
small-business customers within minutes.30 Data masters ensure that data residing with multiple
cloud vendors behaves as if belonging to one source.
Some leading practices emerge from our research with data Interoperability can be improved by working with open-
masters: source or package solutions using the same codebase and
Create an integrated data management across the APIs to enhance integration among cloud vendors. Open
organization: Data masters realize that to achieve scalability APIs, for instance, use a consistent platform to deploy and
and agility, data ownership should be federated, with each manage cloud-native and legacy applications across multiple
business domain handling their own data sets. However, they clouds with a single interface. Furthermore, data teams can
also allow for interoperability across these different domains use established and verified codes from libraries, instead
by applying common data standards. Data mesh architecture, of building from ground zero, which quickens development
for instance, allows federated data ownership while allowing velocity and reduces errors.
communication between distributed data sets.31 Use “federated learning” to work with distributed
• Cloud migration helps in improving agility of data access. datasets: Transferring high volumes of internal and external
For example, Ricoh, an imaging and electronics company, data is costly, difficult, and often risky from a security, privacy
implemented cloud to connect all data from its factory or compliance perspective. Federated learning overcomes
to deliver process improvement, enhance product this; it is an approach that makes it possible to train ML
performance and lower the failure rate of machines. models on distributed data while preserving the privacy of
Telemetry data on the performance of machines is data subjects.
ingested in real time in cloud and live adjustments are For example, data from different hospitals cannot be
made to ensure optimum performance and overall product easily combined owing to a plethora of privacy rules and
quality.32 As Ignacio, formerly from Danone, points out: regulations, but it is critical to improving patient outcomes.
“We are trying to get away from on-premises storage tools Using federated learning, hospitals are now able to train
and bring all the data lakes to the cloud. With that, we have their algorithms on data from multiple institutions. At
access from everywhere and can ensure security of data. the same time, they can keep their data and proprietary
Through our mapping tool, everyone with permissions is algorithms confidential. Intel launched a federated-learning
able to access data from anywhere.” model using patient data from 23 hospitals, which was
combined to improve accuracy of anomaly detection in brain
scans by 11%.33

21
LEARNING FROM DATA MASTERS:
DESIGNING FOR INDUSTRIALIZATION
THROUGH EXPERIMENTATION
GUARDRAILS
Data masters understand that scaling a data culture To tackle this situation, in 2019 Airbnb launched an
is about creating policies and processes for fast Experiment Guardrails system:
experimentation. They achieve this by establishing
• The system has three sub-guardrails, and each
“guardrails” – policies and standards for rolling out
experiment has to pass through all the filters. One of
new features/products/services to market. Data teams
these is the Impact Guardrail, which requires that the
spread across the organization can design and run
change anticipated by a particular experiment does not
experiments within the parameters of these guardrails.
affect other key metrics above a specific percentage.
This ensures velocity is maintained but risk management
• If a team wishes to launch an experiment that has
is not forgotten. Airbnb shows the power of this sort of
“triggered” a guardrail (where the guardrails system
approach in action:
finds that it negatively impacts a key metric), they will
Teams across Airbnb run thousands of experiments on its initiate an escalation process, where stakeholders can
website every week. Each experiment, aimed at improving discuss the results transparently.
a particular metric, can affect key metrics of other • The system flags roughly 25 experiments per month
teams. For example, the trust team may prioritize fraud for escalation or review. And, of these, 80% eventually
identification, while the experiences team may prioritize launch after discussion between stakeholders and
discovery of the online experience on the company’s additional analysis, and 20% (five per month) are
homepage. stopped before launch.
This configurable system allows Airbnb to balance two
priorities: safeguarding key metrics while maintaining
speed of product development.

Sources: InfoQ, “Airbnb: Using Guardrails to Identify Changes with Negative Impact across Teams,” January 2021. News articles.

Data masters customize data-discovery tools as data Given the critical nature of this operation to understanding
initiatives scale data and deriving insights, digital-native companies such as
Facebook, Airbnb, Lyft, Netflix, LinkedIn, Twitter, Uber, and
Data discovery implies collecting, evaluating, and recognizing
Shopify have either customized or built their own solutions
patterns in data from various sources. Organizations use it
for data discovery.35
as a framework through which to understand their data. The
data-discovery process includes connecting multiple data Before building data-discovery platforms, data teams
sources, cleansing and preparing the data, sharing the data can struggle to understand whether there is an existing
throughout the organization, and performing analysis to gain data asset that can be utilized for the current problem,
insights into business processes. or which teams will be affected by changes as a result of
their implementation. At Shopify, this involved asking team
Life sciences major Bayer has automated its data discovery
members and sifting through internal code logs, resulting in
process as Holger Buchner – Pharma IT enterprise architect
multiple sources of truth, lack of full context, and duplication
at Bayer – has stated: “The data transport is now automated,
of effort. When Shopify created its data-discovery platform,
which reduces the amount of discovery work and research
it increased productivity of data teams, reduced business
for the interface development to a minimum and lets people
owners’ dependence on data teams, and improved
focus on the core of the business: understanding how
accessibility of data.36
business data fits to each other between applications.”34

22 Data mastery
Data masters accelerate innovation with DataOps Some best practices emerge from data masters for a
successful enterprise-wide DataOps strategy:
DataOps requires a fusion of technology, processes, and
people. We define it as: “A set of practices to democratize • Build a culture of collaboration with
the use of data and improve its accessibility to business, by cross-functional teams:
setting up an agile cooperative process with data analysts, DataOps must drive collaboration with the business-
data engineers, business users, and IT operations. It improves unit stakeholders. One large financial institution has
the quality, agility, speed of ingestion, and preparation, as built a data lab that creates products and solutions that
well as provisions data for use in AI and analytics use cases.” harness data and advanced analytics to allow a deeper
understanding of its 60m+ customer base. The data lab has
As data masters seek to become ultra-agile, ultra-adaptive,
taken a holistic approach to achieving this goal, starting
and ultra-responsive, they look towards tightly integrated,
with mingling disciplines such as human-centered design,
multi-disciplinary teams that rapidly bring solutions to
full-stack engineering, and data science, and then working
operations. They are employing DataOps to connect data
continuously to build an interdisciplinary team. The team
consumers with data creators to enable collaboration and
includes a project manager who oversees the entire end-
accelerate innovation based.
to-end pipeline, as well as several DevOps team members
DataOps are helping data masters to achieve a number and data scientists.37
of aims:
• Create a DataOps unit:
• Provide faster access to actionable intelligence: 85% of Data masters create a DataOps department headed by
data masters are deploying DataOps practices to improve a CDO. MoneySuperMarket, a British price-comparison
the quality and speed of end-to-end data pipelines and website, has created a data team outside of IT to
90% are using it to deliver analytical solutions quickly accelerate the company’s data-science initiatives. The
• Ensure continuous data flows by automatically identifying reorganization pulled in data specialists from IT and other
and handling data drift. ‘Data drift’ refers to the change areas to create a team dedicated to data science.38
in model input data that leads to model performance
• Focus on metadata management for a mature
issues. Ericsson’s director of data science, Dr. Sunil Kumar
DataOps capability:
Vuppala, says: “A couple of years ago, concept drift
A senior data executive for a large Asia-based bank told
and data drift were not our priorities. But now, we are
us: “You need to define data. First the metadata should be
advocating that the data is going to change over time, and
defined and, ideally, there should be a platform within the
COVID is the best example of patterns changing and of
data platform doing metadata management.”
concept/data drift. Here, MLOps and DataOps are really
helping us – as we talk about real model deployments and • Automate provisioning of data, analytics,
democratization.” and AI infrastructure:
To support the integrated-care model, Humana, a
US-based healthcare company, created a hyper-connected
ecosystem and applied DataOps principles and tools. It
automated and accelerated the provisioning of data from
multiple sources, in a variety of formats, to deliver it into a
hybrid-cloud environment.39

A couple of years ago, concept drift and data drift were not our
priorities. But now, we are advocating that the data is going to
change over time, and COVID is the best example of patterns
changing and of concept/data drift. Here, MLOps and DataOps
are really helping us – as we talk about real model deployments
and democratization.”

DR. SUNIL KUMAR VUPPALA

Director of data science,
Ericsson

23
4. ACTIVATE
DRIVE DATA-POWERED
DECISION MAKING
AND ACTION
As organizations incorporate data from third parties and
work to build their own repositories, they gain knowledge –
about competitors, customers, and the world in which they
operate. At the same time, simply accumulating data is not
enough. Organizations need to “activate” this data.
“Activated data” is about transforming data and insights
into actions within an organization’s processes and ways
of working that helps them to predict market dynamics,
anticipate trends, identify customer behaviors, manage risk,
find operational efficiencies, grow, and innovate.
Activating data for “optimization” – for improving process
efficiencies, business performance, and profitability – is not
a new territory for data masters. Our research highlights
that, with activated data, data masters on average see a 16%
increase in operational efficiency. For instance, by activating
data gathered from internet of things (IoT) sensors in parcel-
processing facilities and delivery vehicles, UPS has been able Proximity with the business is
to reduce average delivery time by a day in 63% of the zip
codes it serves.40 While activation of data for decision making fundamental because in the ocean
and monitoring performance will always be key, data masters of data, ‘boiling the ocean’ becomes
are now looking at transforming the data into new revenue a risk. So, focusing on areas most
streams and new business models (see Figure 5).
impactful for a user is key.”

RONAN CORRE
Enterprise data quality lead,
Microsoft

24 Data mastery
 FIGURE 5 Data masters are generating value with activated data and analytics

WITH ACTIVATED DATA,

DATA MASTERS ARE …

Launching new Identifying new Enabling business Using data for

products and business models elasticity and social innovation
services at speed resilience

Hindustan Unilever To develop new During COVID-19, To drive financial

Ltd.'s (HUL) people data business models for the Johnson & Johnson inclusion, US-based
center – which tracks modern customer, Ford (J&J) leveraged Atlantic Union Bank
social-media trends – Motor Company’s IT risk-simulation tools and was able to activate
was able to identify team developed a mathematical models to data, AI, and cloud to
matcha tea as a product massive data lake that predict worst-case create a new digital
that was gaining pulls together data from scenarios using live data loans portal with the
traction among over 4,600 internal and on staffing levels and help of Temenos, a
health-conscious external data sources production rates. This in FinTech. The bank was
consumers. This trend such as dealerships, turn allowed the able to deliver over
would otherwise have repairs, warranty company to plan $1.4bn in US
been hard to pick up by services, consumer interventions in advance Government Paycheck
traditional research. The buying patterns, credit such as shifting Protection Program
company launched a trends as well as driver production to loans for 6,500
matcha tea product in and vehicle health alternative locations or businesses in less than
less than nine months.41 data.43a Ford also changing shipping two weeks.45
teamed up with methods. The system
insurance carriers across also allowed supply
the U.S to provide chain executives to plan
usage-based auto for raw material
US-based ridesharing insurance policies.43b requirements better Spotlight, a tool
company Lyft analyzed and to avoid over- or developed by a
the GPS coordinates of under-ordering, and to non-profit organization,
its rides, it found that save costs as a result.44 Thorn, helps US-law
90% of rides overlapped enforcement agencies
with other rides from to find child-trafficking
nearby locations. This victims faster. The tool
insight let to the condenses a lot of the
creation of “Lyft Line” – information appearing
a service that allows in online escort ads.
passengers to share a Using ML and deep
car and receive learning-based image
discounts of up to and video analysis, the
50%.42 tool has helped officers
in North America
identify 31,197 victims
of human trafficking —
9,380 of them
children.46

Source: Capgemini Research Institute Analysis; Company websites and news articles.

25
To activate data, analytics, and AI successfully, the following leading practices emerge among data masters:

Create a streamlined data organization

• Use a hub-and-spoke model for data organization and create dedicated data roles
• Create a knowledge repository of multiple data initiatives

Facilitate data-powered innovation across the value chain

• Actively engage with data-sharing ecosystems to feed innovation
• Incentivize innovation and promote a learning culture

Foster a data-powered culture across all levels

• Lead by example
• Identify champions to promote the cause

Source: Capgemini Research Institute Analysis.

4.1 Data masters create a streamlined

data organization

WHY THE RIGHT DATA-OPERATING

MODEL IS IMPORTANT?
A modern data platform will only bring value if the organization has processes in place
to enable access and use of data at the right time to the right person. Accelerating
speed of development will not suffice if data teams are not aligned on business
objectives, or collaboration between IT and business teams or within data teams is
missing.
The right operating model:
• Helps remove data silos between business units
• Ensures strong collaboration between data teams spread across the organization
• Creates strong linkages between data initiatives and overall corporate objectives.

26 Data mastery
How are data masters creating the The operating model for data governance should evolve from
a centralized model to a hub-and-spoke model where a chief
right data-operating model? data officer (CDO)-led central unit is responsible for policy
Data masters use a hub-and-spoke model for data making and governance, MDM and data quality, while the
organization and create dedicated data roles at operationalization of initiatives is managed by the business
leadership level units themselves. The data masters we surveyed earlier
have one thing in common: 95% of them have a CDO, either
At many data laggards, data initiatives are handled in silos as a standalone role (87%) or as an additional responsibility
by different business teams. There is also an absence of (8%), as Figure 6 shows. Essentially, members of these data
enterprise data governance and policies on data privacy, organizations should be a mix of the business
ethics, and security. Such a siloed approach can result in little
data standardization, and no enterprise-level view for C-level
executives.

FIGURE 6 At data masters, the CDO plays a critical role in realizing the data vision

95%

We have a CDO - either as a standalone role or

as an additional role taken up by one of the 55%
senior executives
62%

77%

The CDO has been instrumental in realizing the data 52%

vision of the organization

59%

Data masters Others Average

Source: Capgemini Research Institute, Data-powered enterprises survey, August 2020; N=87 data masters, N=500 technology executives.

Data masters also tend to have a chief analytics officer (CAO) Effective data governance depends on several factors,
who is responsible for business intelligence, data science, and including existing levels of centralization of decisions,
the activation of data. While a CDO is primarily responsible organizational complexity, and maturity of data initiatives.
for managing and governing data, the CAO is responsible Data masters primarily deploy a hub-and-spoke model to
for using data. A CAO can report to the CDO or to the operationalize data initiatives.
CEO/CIO even based on the scope of their responsibilities.
A CDO or a CAO’s role here is to provide guidance and
direction to the organization while allowing the spokes to
operate independently.

27
LEARNING FROM DATA MASTERS:
THE HUB-AND-SPOKE APPROACH
AT FACEBOOK
Facebook has data analysts within all of its product teams who support the relevant
department’s use cases and issues related to specific products, as well as answering the
Spoke: team’s data-related queries. Data analysts also build knowledge about the product they
are assigned to and they can easily convert departmental requirements into required
technical specifications.

Facebook also created a central team to coordinate the efforts of various data teams and
share learning and best practices. The central team:
• Maintains the supporting infrastructure and provides access to the datasets required by
data users
• Acts as the primary publisher of curated datasets, which are used by teams across the
organization
Hub: • Ensures collaboration and data sharing among various data teams residing in product
teams. The central team created a forum that allows all data professionals, data
scientists, data engineers, and data analysts to discuss their data usage and objectives
• Enables Facebook data users to speak a common language and provides common
definitions of data.
• At the corporate level, the CDO is given responsibility for nurturing a data-driven
culture.47

Data masters create a knowledge repository of multiple Once a research project is completed, the platform
data initiatives automatically delivers the results to Colgate-Palmolive
commercial managers and marketers all around the world,
Translating data into action is a critical aspect of becoming
using targeted delivery channels. The platform also provides
data-powered. Data masters create a centralized platform for
targeted insights based on business lines and regions, and AI
reviewing and sharing insights across the business.
recommendations based on the content they open, like, and
Colgate-Palmolive built an insights platform to centralize share.49
the efforts of over 120 insights associates. The platform,
powered by AI algorithms, checks new project briefs,
prevents duplication of past research, and ensures that
teams focus on new insights. “This allows us to shift our
insights budget from tactical work to experimentation and
foundational research that deepens our understanding of
people and the world they live in,” says Taylor Gordy, VP and
GM Northern Europe at Colgate-Palmolive.48

28 Data mastery
4.2 Data masters facilitate
data-powered innovation across
the value chain
WHY DATA-POWERED
INNOVATION IS IMPORTANT?
The impact and value of data-powered innovation is only fully realized by achieving
scale and widespread adoption, as opposed to ideation or experimentation.
For example, Colgate Palmolive has used AI and its database of over 80,000 oral-care
formulas, as well as recent market trends, to reduce drastically the time to develop
and market a new formulation. This has helped Colgate-Palmolive achieve major cost
savings and gain competitive advantage in a crowded market.50

How are data masters accelerating and scaling data-powered innovation?

Data masters realize the value of innovation and take active steps to enable and scale it. They define clear priorities and arrive at
a prioritized use case roadmap. They also focus on building a supportive ecosystem, including startups and universities.
“We are building complex tools in order to mix key business variables to make life easier for innovation,” says Danone’s Ignacio
Marinas. “So, from a consumer perspective, the tech team is trying to detect trends much more clearly and detect winners
early on. Otherwise, it becomes too late for us to react to changes. And, for the different variables we have, we are trying to
identify alarm signals and create a cockpit of innovation. This is where we mix competitor trends and consumer trends, and track
innovation of what’s happening in the market already.”
Data masters actively engage with data-sharing ecosystems to feed innovation:

WHY EXPLOITING EXTERNAL

DATA IS IMPORTANT?
Data masters not only make best use of their internal data, but they also use external
data to be more responsive, competitive, and innovative.
For instance, owing to pandemic-induced panic buying, US-based CVS Health could not
extrapolate demand for some items from historical buying patterns. Taras Gorishnyy,
vice president of enterprise retail analysis at CVS Health, says: “We needed to start
learning very quickly from the changes that were happening in real time: what are the
data elements that capture the new normal of the consumer behavior and how they
have changed?”51
CVS started to find alternative datasets that could be incorporated as proxies into their
demand models. These alternatives included real-estate data from Zillow; infection
data from The Johns Hopkins Hospital and The New York Times; employment data from
the Department of Labor; and news data from The GDELT Project.

29
 Surprisingly, the Zillow data turned out to be one of the most promising proxies for CVS. While the number
of homes sold during “shelter-in-place” dropped, home prices largely held up. As time went on, CVS found a
correlation between strong real-estate prices and customer buying behavior in specific CVS stores. Led by
this data, CVS was able to predict demand within ±2%. Moreover, CVS used this data to create a set of “early
warning” indicators of a likely surge or a dip in sales, enabling the company to take prescriptive action. 52

Data masters are expanding their ability to capitalize on Data masters have started to realize the value in data that
external data. But another option has emerged: data is shared and collaborated on, all without risking their
sharing and exchange. Companies such as BMW have been competitive advantage. Our previous research found that
sharing their data to enable “data for good” applications. organizations participating in data ecosystems have the
BMW recently partnered with Otonomo, an automotive potential to gain financial benefits up to $940 million or over
data-services platform, to share its vehicle data to enable 9% of annual revenue in the next five years.54 Many models
smart-city, real-time traffic information, and other solutions for data sharing have emerged:
that improve the driving experience.53 COVID-19 has also
further demonstrated the need for businesses to work
together to share data to help them innovate, achieve
efficiencies, and build resilience.

Data masters are looking at various data-sharing models to unlock new sources of revenue, introduce
FIGURE 7 new services, and deliver new insights

Commercialization of data, insights & Combining own data with partners'

audience to third parties to deliver new insights, develop
enhanced services or leverage
cross-sell opportunities
monetization

Singtel’s data-analytics
partnerships
1-to-1 data

subsidiary – DataSpark
provides businesses with Using transaction data insights,
access to its repository Commercial International
Data

of anonymized Bank (CIB) Egypt is developing

geolocation data.55 a card-linked loyalty program
with a local retailer to deliver
personalized discounts to
consumers – this is leading to
increased sales turnover to the
retailer, and higher card
balances for the bank.56
DATA SHARING
ECOSYSTEMS

Skywise – an open data JPMorgan Chase, Wells Fargo,

platform designed and and U.S. Bancorp plan to start
developed by Airbus – sharing data on customers’
combines airlines data and deposit accounts to extend
data hubs
powered

Sectorial
services

provide value-added credit to people who have

Data-

services (e.g. predictive traditionally lacked

maintenance, benchmark).57 opportunities to borrow.58

New services, which bring added value Sharing data across the sector to unlock
or complement existing products gains and powering data completeness

Source: Capgemini Research Institute Analysis; Company websites and news articles.

30 Data mastery
Data sharing does not necessarily come naturally to • In data masters, innovation is not only encouraged, but
businesses, however. There are many barriers, from also incentivized, and failure is embraced and celebrated.
establishing mutual trust for proper handling of shared data, P&G gives out a “Heroic Failure” award each year to the
to understanding the sharing mechanisms, and ensuring individual or team who takes the greatest “intelligent”
compliance to regulations when sharing externally. risk at the company. At the heart of the Heroic Failure
award is an assurance to employees that they will not be
From our research, we were able to understand some best
punished for the mistakes they make, building a culture of
practices when accessing data ecosystems:
“psychological safety” and trust.61
• Prioritize data sharing use cases that are aligned with
business goals:
For instance, even before the COVID-19 outbreak, ten large
pharmaceutical companies – including Johnson & Johnson,
AstraZeneca and GSK – undertook collaborative efforts to
train their ML algorithms on each other’s data to promote
drug discovery. They used blockchain to share data without
compromising confidential or commercial secrets. This was
aligned with their target of accelerating the discovery of
drugs and reducing the cost of the process.59
Explore various ways of data sharing:
A recent survey reveals that 49% of organizations that
commercialize data expose an API to that data and 46% sell
an application that enables users to see data trends and
insights.60

A former chief innovation officer at a credit insurance

company says: “We collect data on companies to assess the
risk of default – and this data is a key asset to our business We want to create this culture that
for taking decisions such as underwriting or selling insurance.
But now we also extensively share this data to identify new
quality data is a everybody’s matter.
value propositions and for monetization. We have created a Everybody in the data supply chain
product, called ‘TrafficLight,’ to provide an aggregated view needs to be accountable for that
on the financial health of a company.”
clean house because everybody
• Establish governance around sharing data: benefits from it...”
It is crucial to move away from a culture of ownership
to one of data sharing. However, data masters also take
into consideration the government and industry-related DELPHINE CLEMENT
regulations regarding the security and privacy of data. Director business program manager,
Data masters also appoint “data-governance champions”
for looking at these legislative mandates and setting up
Enterprise Data Management,
policies and procedures for effective data sharing. Microsoft
• Data masters incentivize innovation and promote a
“learning” culture:
• Innovative experiments do not always succeed, and a key
component of an innovation culture is the ability to take
risks, accept failure and learn from it. The idea of “failing
fast” and “learning fast” is often discussed and encouraged
in data masters.

31
4.3 Data masters foster a
data-powered culture across all levels
of management
WHY DATA-POWERED CULTURE
IS IMPORTANT?
In previous research we conducted, 62% of organizations said that establishing a
digital culture was one of the top-two hurdles for digital transformation.62 Further,
our research on data-powered enterprises showed that almost all data masters
invest in a data culture.

FIGURE 8 Data masters invest in developing a data culture

We have invested in a data culture by 100%

enabling employees with the skills and
tools to generate and apply insights
48%
57%
We build cross-functional data and
insights teams that work with data 99%
engineers, data scientists, solution 48%
architects, and software developers 57%

Data masters Others Average

Source: Capgemini Research Institute, Data-powered enterprises survey, August 2020; N=87
data masters, N=500 technology executives.

How do data masters build a this resistance, data masters look to identify champions
among the business teams who can promote their cause
data-powered culture? among their peers.
Data masters lead by example
Camilla Schwartz-Björkqvist, global data & analytics
Data masters ensure data-driven behavior by all their team transformation director at Danone, outlines how they
members. Roberto Giménez, digital business manager at identified key executives who were highly motivated by
Roche, says: “I have been working with people with clear the objective to help lead the transition to a data-powered
leadership on data-driven decisions. Whenever they ask for culture. “There’s a cultural transformation [in] a data and
resources or feedback, they also check whether this is based analytics journey,” she explains. “Firstly, we took a very
on data or based on emotions, thoughts, etc. When they have conscious strategy of going with the front-runners. So, the
meetings, they ask that all the feedback that we bring to executives that were excited to do this – who had either
those meetings be supported by something, thereby creating done it before or couldn‘t wait to transform. We focused on
this culture from the top down.” them first – creating business cases together with them and
making sure that we delivered ROI quickly so that we could
Data masters identify champions among business teams use those success stories.”
who can promote their cause
A cultural change in the organization takes time and is often
met with resistance from employees. In order to overcome

32 Data mastery
CONCLUSION
Organizations clearly grasp the value of data, but only a
few are able to make data work for them. These fortunate
few – our data-powered enterprises or the data masters –
enjoy a significant business advantage. Organizations that
want to emulate that advantage need to take a number of
concrete steps:
1. Data strategy – The data masters have long
since realized that business ownership of data
transformation is a must-have. They work hard to
establish a data strategy that is business-driven
and owned. With that in place, they are then able to
capitalize by the opportunities offered by combining
both internal and external data.
2. Data trust – the data trust gap is marked, and
data masters take concrete steps in reducing this
trust deficit. They focus strong data management
capabilities and data quality to build the base trust
foundations in the veracity of data. They then
focus on trusted AI systems and ethics as well as
democratization of data.
3. Modernization of data estate – to ensure that the IT
systems are able to support the increasing demand
for data and analytics, data masters take a number
of steps, including decommissioning legacy systems,
migrating to multi-cloud environments, leveraging
DataOps and developing customizable data-discovery
tools.
4. Activation – this is the final but the critical step in
becoming a data master. Data masters are successful in
building a data organization that focuses on business I have been working with people
priorities and is business-driven. A governance model with clear leadership on data-driven
that supports this data organization – coupled with a
culture focusing on data-powered innovation – helps
decisions. Whenever they ask for
organizations get ahead of their peers. resources or feedback, they also
With concerted and bold action in these four areas, check whether this is based on data
organizations can join the data masters. And, by joining or based on emotions, thoughts, etc.
the ranks of the high performers, they will be better
placed to drive growth, meet fast-changing customer
When they have meetings, they ask
needs, manage risk, and get ahead of shifts and volatility that all the feedback that we bring
in their business environment. to those meetings be supported by
something, thereby creating this
culture from the top down.”

ROBERTO GIMÉNEZ
Digital business manager, Roche

33
APPENDIX 1
OUR DATA MASTERY
FRAMEWORK
Data foundations (tech & tools) Data behaviors

• Data & AI platform • Data activation vision and strategy

• Data identification • Data-powered decision making in business
• Data ingestion, processing and harvesting • Data guiding principles (data access,
interoperability, security, etc.)
• Data governance implementation
• Data governance processes
• Infusion into business for data activation
• Data activation culture
• "Data advantage" tools to leverage
external data • Data advantage strategy

Only 1 in 6 organizations can be categorized as a data master

Data aware Data masters

7% 16%
Data behaviors (people, processes, skills and culture)

71% 7%
Data laggards Data enabled

Data foundations (tech & tools)

Source: Capgemini Research Institute analysis, Data-powered enterprises survey, August 2020; N=500 technology executives, N=504 business
unit executives. Percentages are rounded off to the nearest integer and do not necessarily total 100.

34 Data mastery
APPENDIX 2
CAPGEMINI’S CODE OF
ETHICS FOR AI
At Capgemini, our ethical culture drives our vision of AI. As a leader in digital transformation, we are committed to the adoption
of AI in a way that delivers clear benefits within a trusted framework.
We believe that human ethical values should never be undermined by the uses made of AI by business. We want AI solutions to
be human-centric, which we define as follows:

1 3
AI with carefully Sustainable AI – Fair AI – Transparent and
delimited impact – explainable AI –
developed mindful of produced by diverse
designed for human each stakeholder, to teams using sound data with outcomes that can
benefit, with a clearly benefit the environment for unbiased outcomes be understood, traced,
defined purpose, and all present and and the inclusion of and audited.
setting out what the future members all individuals and
solution will deliver and of our ecosystem, population groups.
to whom. human and non-human
alike, and to address 4
pressing challenges
such as climate change,
CO₂ reduction, health
improvement, and
sustainable food
production.

2
5 7

Controllable AI Robust and safe AI – AI respectful of

with clear privacy and data
including fallback plans
accountability – protection –
where needed.
enabling humans considering data privacy
to make more informed and security from the
choices and final
decisions.
6 design phase, for data
usage that is secure and
legally compliant with
privacy regulations.

For more information, please visit: https://www.capgemini.com/our-company/values-ethics/our-code-of-ethics-for-ai/

35
 RESEARCH
METHODOLOGY
Between February to April 2021, we conducted in-depth interviews with 51 senior executives from 35 organizations
with leading practices in data and analytics, across a range of sectors and countries. Two-thirds of these organizations
reported revenues of more than $20 billion for the last financial year. More than one-fifth of these organizations are
either big-tech companies or digital natives.

Executives by country
The United States Belgium

4% France
14%
29% Germany
4%

20% India
10%
The United Kingdom
2%
6% 6%
Sweden 6%

Spain The Netherlands

Singapore

Executives by function

14% 14%
12%

8% 8%
6% 6% 6%
4% 4% 4%
2% 2% 2% 2% 2% 2% 2% 2%
AI and Analytics

Customer service

CXO

Data and analytics

Data Management

Digital

Digital strategy

General management

Human Resources

Information security

Innovation

IT

Marketing

Operations

Risk Management

Sales

Strategy

Supply chain

Sustainability

36 Data mastery
 Executives by company type

Big-tech/Digital natives
16%

Technical (IT/data) executives 47%

37%
Business executives

Organizations by industry Organizations by revenue

Telecom Automotive

9% 3% Banking
Retail 9% 9%
6%
Life Sciences
6% 14%
Insurance 3% Big-tech/
23% Digital 40%
Industrial 6% natives
Manufacturing 11%
9%
Energy
and Utilities 9% 17%
29%

More than USD $50 billion

Consumer Products
USD $30 billion - $40 billion
USD $20 billion - $30 billion
USD $10 billion - $20 billion
USD $5 billion - $10 billion
Less than USD $1bn

Source: Capgemini Research Institute, Learning from data masters in-depth interviews, N=51 executives from 35 leading organizations in
data and analytics.

37
REFERENCES
1 Capgemini Research Institute, “The data-powered enterprise: 32 Microsoft Azure, “Factory of the future with Microsoft Azure
Why organizations must strengthen their data mastery,” machine learning and AI,” March 2020.
November 2020. 33 Intel Corp special call, April 06, 2021, via AlphaSense.
2 Ibid. 34 Talend, “Enabling customer-centric, data-driven marketing and
3 GDPR-info.eu, “Art. 5 GDPR: Principles relating to processing of sales.
personal data,” accessed May 10, 2021. 35 Facebook Engineering, “Nemo: data discovery at Facebook.”
4 Capgemini Research Institute, “DATA SHARING MASTERS: How 36 Shopify Engineering, “How we’re solving data discovery
smart organizations use data ecosystems to gain an unbeatable challenges at Shopify,” July 2020.
competitive edge,” July 2021. 37 TDWI, “How DataOps is transforming industries,” April 2019.
5 ESRI, “How GM maps and manages supply chain risk,” November 38 Eckerson Group, “Best practices in DataOps: how to create
2018. robust, automated data pipelines,” June 2019.
6 ComputerWorld, “How Munich Re built a data lake fit for all its 39 Streamset, “Humana - making healthcare easier for community
employees,” April 2018. members.”
7 Capgemini Research Institute, “DATA SHARING MASTERS: How 40 Yahoo Finance, “UPS slashes ground delivery transit times,”
smart organizations use data ecosystems to gain an unbeatable November 2020.
competitive edge,” July 2021. 41 Fortune India, “HUL: Eat. Sleep. Innovate. Repeat,” December
8 Capgemini Research Institute, “The data-powered enterprise: 2019.
Why organizations must strengthen their data mastery,” 42 AWS, “The data-driven enterprise.”
November 2020. 43 a. AI Data and Analytics Network, “Ford’s data-driven roadmap
9 Gartner, “How to create a business case for data quality towards future mobility,” February 2021.
improvement,” June 2018.
b. Ford Media Center, “Ford Supports InsureTech Revolution
10 Pragmatic works, “The cost of bad data – infographic,” May
With Insurance Carriers, Ford Insure And Lincoln Motor
2017.
Company Insure, Using Embedded Telematics To Help
11 Capgemini, “How data quality can hurt your data science Customers Save On Auto Policies,” December 2020.
programme… if you’re not careful,” February 2020.
44 Johnson & Johnson, “4 high-tech tools J&J is using to get
12 Capgemini Research Institute, “The AI-powered enterprise: products to you during the pandemic,” September 2020.
Unlocking the potential of AI at scale,” July 2020.
45 World Economic Forum, “Cloud and SaaS technology can drive
13 Stanford University, “Artificial Intelligence index report,” 2021. inclusive banking. Here are 3 reasons how,” December 2020.
14 Capgemini Research Institute, “AI and the Ethical Conundrum: 46 Amazon AWS, “Thorn uses AWS to help law enforcement
How organizations can build ethically robust AI systems and identify child-trafficking victims faster,” November 2017.
gain trust,” September 2020.
47 O’Reilly, “Building a data driven organization.”
15 Capgemini Research Institute, “Why addressing ethical
48 Marketlogic website, “Colgate Palmolive and Market Logic win
questions in AI will benefit organizations,” September 2020.
2021 BIG Innovation Award,” January 2021
16 Siemens blogs, “Achieving ethical AI for industrial applications,”
49 Cision, “Market Logic and Colgate-Palmolive win 2021 BIG
January 2021.
innovation award,” January 2021
17 H&M group, “Meet Linda Leopold, Head of AI Policy,” June 2019.
50 Call transcript, “Colgate-Palmolive Co at Consumer Analyst
18 Reuters, “Amazon scraps secret AI recruiting tool that showed Group of New York Conference,” February 21, 2020.
bias against women,” October 2018.
51 Worldquant Predictive website, “Our CVS Health Predictions
19 Business Insider, “Google empowers 5,000 Cloud employees in Project Featured in Datanami,” March 2021
ethical AI,” July 2020.
52 Datanami, “The data proxy that let CVS see around the COVID
20 Google Cloud, “Explainable AI,” accessed May 28, 2021. corner,” March 2021.
21 CFO, “The growing importance of data democratization,” 53 Forbes, “BMW, Otonomo expanding smart city data-sharing
December 2019. plan,” April 2020.
22 Forbes, “What is data democratization? A super simple 54 Capgemini Research Institute, “DATA SHARING MASTERS: How
explanation and the key pros and cons,” July 2017. smart organizations use data ecosystems to gain an unbeatable
23 Alation, “Munich Re launches enterprise-wide data-driven competitive edge,” July 2021.
platform for analytics,” October 2018. 55 DataSpark website, accessed May 2021.
24 Dataiku, “GE Aviation: from data silos to self-service.” 56 Forrester, “Chief data officers: invest in your data sharing
25 Ibid. programs now,” March 2021.
26 CarMax analyst day transcript, May 6, 2021. 57 Airbus Skywise website, accessed May 2021.
27 Shopify, “Deconstructing the monolith: designing software that 58 Wall Street Journal, “JPMorgan, others plan to issue credit cards
maximizes developer productivity,” February 2019. to people with no credit scores,” May 2021.
28 DoorDash engineering, “Rebuilding our pricing framework for 59 Financial times, “Pharma groups combine to promote drug
better auditability, observability, and price integrity,” March discovery with AI,” June 2019.
2021. 60 Forrester Analytics Business Technographics Data And Analytics
29 Diginomica, “Capital One closes its data centres and goes all in Survey, 2020, N=1,829 global data and analytics business
with AWS,” January 2021. decision-makers whose firms are commercializing data.
30 Amazon Web Services website, “Nationwide selects AWS as its 61 Forbes, “Are you embracing failure or encouraging failure?”
preferred cloud provider to deliver digital experiences to June 2015.
customers,” December 2020. 62 Capgemini Research Institute, “The digital culture challenge:
31 Towards Data Science, “What is a Data Mesh — and How Not to Closing the employee-leadership gap,” 2017.
Mesh it Up,” July 2020.

38 Data mastery
ABOUT THE AUTHORS

Zhiwei Jiang Anne-Laure Thieullent

CEO, Insights and Data Global Business Line Managing Director, Artificial Intelligence &
[email protected] Analytics Group Offer Leader
[email protected]
Zhiwei is the CEO of I&D Global Business Line for
the last two years. Before that, Zhiwei was the Anne-Laure leads Perform AI, Capgemini’s Artificial
Head of Insights & Data for Financial Services intelligence & Analytics group offer. She works with
Business Unit. Prior to Capgemini, he spent 20 clients to scale Artificial Intelligence, so it infuses
years in banking. His passion has been in data everything they do, and they become AI-driven,
privacy, security, sustainability, and now more into data-centric and innovative. With 20 years of
AI ethics and social impact. experience in big data, analytics and AI systems,
from design to production roll-out, her passion is
to bring companies what they need to transform
themselves into intelligent enterprises. Anne-
Laure is committed to guiding clients to increase
activating their data, while cultivating the values of
trust, privacy and fairness.

Ron Tolido Valérie PERHIRIN

EVP, CTO, Insights and Data Global Managing Director, Data, Analytics and AI
[email protected] Capgemini Invent
[email protected]
Executive Vice President and Chief Technology
Officer, Capgemini Insights & Data global business Valérie leads the data, analytics and artificial
line. AI Futures domain lead for Capgemini’s Intelligence consulting activities. She advises
Technology, Innovation & Ventures (TIV) council. Capgemini Invent clients to design their data
Lead author of the TechnoVision trend series. strategy and designs roadmap and execution plan.
She helps our clients to drive their data powered
transformation in order to reach operational
excellence and leverage data and AI to invent
new sources of revenue. She has over 19 years of
experience in data centric transformation powered
by data, analytics and AI.

Marie-Caroline Baerd Steve Jones

Executive Vice President, Data & AI – Financial EVP, Chief Data Architect
Services, Capgemini Invent [email protected]
[email protected]
Steve Jones is the Chief Data Architect for
Marie-Caroline leads Capgemini Invent’s Data & Capgemini’s Group Portfolio. With over a decade
AI offer for the FS sector, developing thought in Big Data and Cloud, he’s been at the front of
leadership, new frameworks and solutions. She has helping our clients adopt new technologies to drive
a track record of more than 20 years in business new business value and become more data driven.
management consulting, and is applying her Steve was one of the first people to propose using
experience in helping her clients leverage the mobile applications and analytics to help track
full potential of Data & AI to innovate, enhance and manage individuals carbon footprint and has
their revenues and efficiency, supporting their continued to be at the front of Capgemini’s CSR
transformation journey from awareness and AI/ and Diversity initiatives.
Data strategy to implementation at scale within
the organization.

39
 Jerry Kurtz Eric Reich
Executive Vice President, Insights & Data Offer Leader and Global Head of AI & Data
[email protected] Engineering Insights & Data, Capgemini
[email protected]
Jerry has more than 30 years of IT and
management consulting experience. He currently Eric is having 20+ years of experience in the
leads Capgemini’s Insights and Data business in Industry and is part of Capgemini Insights and
the Americas. During his career, Jerry has led Data Global practice where he leads globally AI &
a number of global enterprise transformation Data Engineering team and portfolio. With 16+
programs with a particular focus in Manufacturing, years of experience in the FS sector and 8 years+
Hi-tech, Consumer Products, Retail, and Logistics in the data space, Eric had the opportunity to
industries. Some of his competencies include Big work with some of Fortune 500 customers in the
Data & Analytics, Artificial Intelligence, Internet Assessment, Design and Execution phases of their
of Things (IoT), Enterprise Transformation, SAP, data driven transformation. He is also engaged in
Supply Chain Management, Shared Services, and the development of new offers and solutions on
Business Process Services. He has also lived and the data, Analytics & AI space with the major cloud
worked in Japan on two separate occasions for five platform providers and ISV on the market.
years.

Jerome Buvat Nancy Manchanda

Global Head of Research and Head of Capgemini Senior Manager, Capgemini Research Institute
Research Institute [email protected]
[email protected]
Nancy is a senior manager with Capgemini Research
Jerome is head of Capgemini Research Institute. He Institute. She keenly tracks the patterns of digital
works closely with industry leaders and academics disruptions across industries to help clients
to help organizations understand the nature and understand the business impact of emerging
impact of digital disruptions. technologies.

Ramya Krishna Puttur Gaurav Aggarwal

Assistant Director, Capgemini Research Institute Manager, Capgemini Research Institute
[email protected] [email protected]
Ramya is an Assistant Director at Capgemini Gaurav is a manager at the Capgemini Research
Research Institute. She studies the growing role of Institute. He likes to assess how technology
digital technologies in transforming the boundaries impacts businesses and understand how they
of traditional business consortiums to help them respond to it. He is eager to learn about emerging
develop informed strategies. business models, technologies, and trends across
sectors.

The authors would like to thank Paul Hermelin, Dinand Tinholt, Maarten Deceuninck, Sandeep Bhatia, Gagan Mehta, Sarah
McNeil, Subrahmanyam KVJ, and Mansi Gupta for their contribution to the research.
The authors would also like to thank Ashwin Yardi, Thomas Svahn, Nick Clarke, Yashowardhan Sowale, Chris Dickey, Monish Suri,
Ingo Finck, Sergey Patsko, Christian Kaupa, Pierre-Adrien Hanania, Isabell Schastok, Matthew Jones, Sandra Ficht, Seth Rachlin,
Nilesh Vaidya, Rüdiger Eberlein, Giulia Carsaniga, Johanna Sundh, and Priyabrata Ghosal for their contribution to this research.

About the Capgemini Research Institute

The Capgemini Research Institute is Capgemini’s in-house think tank on all things digital. The Institute publishes research
on the impact of digital technologies on large traditional businesses. The team draws on the worldwide network of
Capgemini experts and works closely with academic and technology partners. The Institute has dedicated research
centers in India, Singapore, the United Kingdom, and the United States. It was recently ranked number one in the world
for the quality of its research by independent analysts.
Visit us at www.capgemini.com/researchinstitute/

40 Data mastery
For more information, please contact:

Global

Anne-Laure Thieullent Valerie Perhirin Steve Jones

Vice President, Artificial Intelligence and Managing Director in charge of Insight-Driven- Chief Data Architect
Analytics Group Offer Leader Enterprise, Capgemini Invent [email protected]
[email protected] [email protected]

Eric Reich Padmashree Shagrithaya Sébastien Guibert

AI & Data Engineering Offer Leader AI, Analytics & Data Science Offer Leader Intelligent Process Automation Offer Leader
[email protected] [email protected] [email protected]

APAC France Germany

Dr. Jing Yuan Zhao Marion Gardais Fabian Schladitz
AI Centre of Excellence Leader AI Centre of Excellence Leader AI Centre of Excellence Leader
[email protected] [email protected] [email protected]

Europe India
Robert Engels Chandrasekhar Balasubramanyam
AI Centre of Excellence Leader AI Centre of Excellence Leader
[email protected] [email protected]

North America United Kingdom

Dan Simion Joanne Peplow
AI Centre of Excellence Leader AI Centre of Excellence Leader
[email protected] [email protected]

41
EMPOWER YOUR BUSINESS
LEVERAGING DATA & AI
WITH CAPGEMINI
Data promises infinite possibilities. Now, more than ever, Our seamless approach, that connects business strategy and
organizations need to activate its full potential. To become data operations, enable growth, performance, innovation
a Data Master, you must be at the helm, ready to harness with tangible benefits. We empower some of the largest
data in critical dimensions, to outperform cohorts on data & AI transformations at scale in the market – activating
financial performance and make crucial decisions from a Data & AI for CxOs. We improve organization’s business and
qualitative base. A Data Master is a high-performer who is cultural change, data engineering and data science skills, and
able to harness the full suite of data capabilities across its execute at scale with hybrid cloud Data & AI platforms and
organization and ecosystem — turning insights into actions, Intelligent Process Automation. Get ready to shift into data-
and beyond. mastery to sustain a continuous competitive advantage with
trusted business outcomes.
Data Masters significantly outperform their peers, generating
+70% higher revenue per employee and up to 22% higher
profitability and lead in top-line benefits increasing sales of
traditional or new products and services.1 More over front-
runners who are participating in data ecosystems have the
potential to gain financial benefits up to $940 million or
over 9% of annual revenue over the next five years.2 There’s
an ocean of business opportunities to create new intelligent
experiences, products and services leveraging data from
your ecosystem.

Become a Data Master: Manage

your data as a strategic asset
Capgemini’s Perform AI is a portfolio of Data and Artificial
Intelligence services to help organizations take control of
their data as a strategic asset and transform their business to
be a leader in the data economy.

1 Capgemini Research Institute, “The data-powered enterprise: Why organizations must strengthen their data mastery,” November 2020.
2 Capgemini Research Institute, “DATA SHARING MASTERS: How smart organizations use data ecosystems to gain an unbeatable
competitive edge,” July 2021.

42 Data mastery
PERFORM AI:
Activate data. Augment intelligence. Amplify outcomes.
Speed up your business outcomes with trusted Data & AI at scale
With Perform AI, we deliver high-performing, tangible business outcomes, across industries and sectors.

INCREASED
SALES 25%
FASTER
CUSTOMER
INSIGHTS
2x
impact
BETTER
CUSTOMER
SERVICE
24
languages

Organizations successfully scaling A global consumer product firm One of the world’s largest biotech
AI have seen more than a 25% successfully engaged with 1billion companies used our award-winning*
increase in sales of their products people using customer data & virtual assistance to automate
and services insights with double the impact in outbound telephone calls in 24
half the time for half the cost. languages

OPERATIONAL
EXCELLENCE cheaper
40% COST
SAVINGS €100 M
FRAUD
DETECTION ROI
10x
A leading producer of packaging A major international retailer A European government
and paper used our AI solution to used our AI-powered solution to agency has delivered a 10x
reduce its cost of handling invoice optimize sales forecasting which return on investment thanks
processing queries by 40% overall. directly drove €100M of inventory to our AI-powered fraud
costs savings. detection systems.

*Capgemini wins 2020 Artificial Intelligence Breakthrough award for best virtual agent solution

43
 CAPGEMINI PERFORM AI
Accelerated time to value for business outcomes
with trusted Data & AI at scale

CP & RETAIL FINANCIAL SERVICES MANUFACTURING,

& INSURANCE AUTO & LIFE SCIENCES
Augmented operations
Augmented operations for
and immersive customer Right Product,
operational efficiency &
engagement Right Price
risk reduction

TELCO ENERGY UTILITIES &

PUBLIC SECTOR
MEDIA TECH CHEMICALS
Deep Customer Sustainability powered by Infusing AI to better
Engagement and Data & AI and Operational serve citizens and
Operational Excellence Excellence governments

DATA ACTIVATION DATA ADVANTAGE

CUSTOMER INTELLIGENT ENTERPRISE DATA SHARING
FIRST INDUSTRY MANAGEMENT ECOSYSTEMS

DATA ENABLERS

Strategy & MVPs Build, Deploy, Manage, Operate

INTELLIGENT
AI ACTIVATE AI, ANALYTICS PROCESS
DATA SCIENCE AUTOMATION
Transformation Target Scalable, Trusted AI and Data & AI powered
and foundations for Augmented BI & Data Viz Frictionless Entreprise
execution of AI @Scale solutions

AI REIMAGINE AI & DATA

ENGINEERING
Invent and Implement Hybrid Cloud Data & AI platforms, Data Estate
“the next” Data & AI modernization, Intelligent Master Data
powered business Management & Data Trust

AI4Good Trusted AI Sustainable AI

44 Data mastery
Discover more about our research:

The data-powered AI and the

enterprise Ethical Conundrum
How organizations can build ethically
robust AI systems and gain trust

€ $

$ €
DATA
€ $ SHARING
$ € MASTERS
How smart organizations
use data ecosystems to gain an
unbeatable competitive edge

Why organizations must

strengthen their data mastery

The data-powered Data sharing masters: The Age of Insight: AI and the Ethical
enterprise: How smart organizations use How Consumer Products Conundrum:
Why organizations must data ecosystems to gain an and Retail organizations can How organizations can build
strengthen their data unbeatable competitive edge accelerate value capture ethically robust AI systems
mastery from data and gain trust

The AI-powered enterprise:

Digital
The art of
customer-centric
Unlocking the potential of AI at scale
Mastery
How organizations have progressed
artificialintelligence in their digital transformations over
the past two years

How organizations can unleash Scaling AI in

the full potential of AI Manufacturing Operations:
in the customer experience A Practitioners’ Perspective

The Art of customer centric The AI-powered Digital Mastery: Scaling AI in

Artificial Intelligence enterprise How organizations have Manufacturing
Raising the bar on AI in Unlocking the potential of progressed in their digital Operations:
customer experience AI at scale transformations over the A Practitioners’
past two years Perspective

Why addressing The Digital

Agile at Scale ethical questions in AI Culture Challenge:
Four ways to gain will benefit organizations Closing the Employee-Leadership Gap
enterprise-wide agility

What’s the
Big Idea?
Why most innovations fail to scale and what
to do about it

Digital
Transformation
Institute
By Capgemini Digital Transformation Institute

Agile at Scale Scaling Innovation: Why addressing ethical The Digital Culture
Four recommendations to What’s the Big Idea? Why most questions in AI will benefit Challenge:
gain enterprise-wide agility innovations fail to scale and organizations Closing the Employee-
what to do about it Leadership Gap

45
46 Data mastery
SUBSCRIBE TO LATEST
RESEARCH FROM CAPGEMINI
RESEARCH INSTITUTE
Receive advance copies of our reports by scanning the QR code or visiting
https://www.capgemini.com/capgemini-research-institute-subscription/

47
About Capgemini
Capgemini is a global leader in partnering with companies to transform and manage
their business by harnessing the power of technology. The Group is guided everyday
by its purpose of unleashing human energy through technology for an inclusive
and sustainable future. It is a responsible and diverse organization of 270,000 team
members in nearly 50 countries. With its strong 50 year heritage and deep industry
expertise, Capgemini is trusted by its clients to address the entire breadth of their
business needs, from strategy and design to operations, fuelled by the fast evolving
and innovative world of cloud, data, AI, connectivity, software, digital engineering and
platforms. The Group reported in 2020 global revenues of €16 billion.

Get the Future You Want | www.capgemini.com

Copyright © 2021 Capgemini. All rights reserved.