The data-driven

enterprise of 2025
Rapidly accelerating technology advances, the
recognized value of data, and increasing data literacy
are changing what it means to be “data driven.”

© Signal Noise

January 2022
 By 2025, smart workflows and seamless interactions The following are the seven characteristics
among humans and machines will likely be as of the data-driven enterprise:
standard as the corporate balance sheet, and most
employees will use data to optimize nearly every 1. Data is embedded in every decision,
aspect of their work. interaction, and process.

We know 2025 isn’t too far off, but that’s the point. 2. Data is processed and delivered in
real time.
Seven characteristics will define this new data-
driven enterprise, and we’ve already seen many 3. Flexible data stores enable integrated,
companies exhibit at least some of them, with many ready-to-use data.
more beginning the journey to do so.
4. Data operating model treats data like
Those able to make the most progress fastest stand a product.
to capture the highest value from data-supported
capabilities. Companies already seeing 20 percent 5. The chief data officer’s role is expanded to
of their earnings before interest and taxes (EBIT) generate value.
contributed by artificial intelligence (AI), for example,
are far more likely to engage in data practices that 6. Data-ecosystem memberships are
underpin these characteristics.1 the norm.

This guide is intended to help executives understand 7. Data management is prioritized and
the characteristics of the new data-driven automated for privacy, security,
enterprise and the capabilities they enable. It also and resiliency.
provides resources to dive deeper on how to embed
them in your organization.

Seven characteristics will define the new

data-driven enterprise. Those companies
able to make the most progress fastest
stand to capture the highest value from
data-supported capabilities.

1
“ The state of AI in 2021,” McKinsey, December 8, 2021.

2 The data-driven enterprise of 2025

 1. Data is embedded in every decision, maintenance and highlight opportunities for
interaction, and process building out the network based on usage.

Today
— Procurement managers regularly apply data-
Organizations often apply data-driven approaches—
driven processes to instantly triage purchases
from predictive systems to AI-driven automation—
for approval so they can focus on building out a
sporadically throughout the organization, leaving
more effective partner strategy.
value on the table and creating inefficiencies.
Many business problems still get solved through
Key enablers
traditional approaches and take months or years
— a vision and data strategy to highlight and
to resolve.
prioritize transformational use cases for data

By 2025
— technology enablers for sophisticated AI use
Nearly all employees naturally and regularly leverage
cases, such as a cloud-based infrastructure;
data to support their work. Rather than defaulting to
architectures that support real-time analytics;
solving problems by developing lengthy—sometimes
and flexible database/data-model tooling to
multiyear—road maps, they are empowered to
support querying of unstructured data
ask how innovative data techniques could resolve
challenges in hours, days, or weeks.
— broad organizational data literacy and a data-
driven culture, where all employees know and
Organizations are capable of better decision making
embrace the value of data
as well as automating basic day-to-day activities
and regularly occurring decisions. Employees are
How to get started
free to focus on more “human” domains, such as
— Read “Winning with AI is a state of mind” for
innovation, collaboration, and communication. The
more about making the shift to an AI-enabled
data-driven culture fosters continuous performance
organization, and learn how to harness the
improvement to create truly differentiated customer
power of data from AI leaders.3
and employee experiences and to enable the growth
of sophisticated new applications that aren’t widely
— Begin upskilling your employees for data use
available today.
and AI, if you haven’t started already. Analytics
academies can help.4
Everyday applications2
— Store managers provide a differentiated
— Learn how to reimagine each workflow, journey,
shopping experience using real-time analytics to
and function to leverage data and AI in “Getting
identify and direct loyalty-program customers
AI to scale.”5
to products they find interesting as they shop,
and streamline or completely automate the
— Articulate your vision for a data-
checkout process.
driven organization.

— Network operations staff at telecommunications

companies leverage autonomous networks that
automatically identify areas requiring

2
E xamples; not exhaustive.
3
See Thomas Meakin, Jeremy Palmer, Valentina Sartori, and Jamie Vickers, “Winning with AI is a state of mind,” McKinsey, April 30, 2021; and
Mohammed Aaser, Jonathan Woetzel, and Kevin Russell, “Five insights about harnessing data and AI from leaders at the frontier,” McKinsey
Global Institute, March 25, 2021.
4
Solly Brown, Darshit Gandhi, Louise Herring, and Ankur Puri, “The analytics academy: Bridging the gap between human and artificial
intelligence,” McKinsey Quarterly, September 25, 2019.
5
Tim Fountaine, Brian McCarthy, and Tamim Saleh, “Getting AI to scale,” Harvard Business Review, May–June 2021, Volume 99,
Number 3, pp 116–23.

The data-driven enterprise of 2025 3

 2. Data is processed and delivered example, by using internet-protocol data and
in real time website behavior to personalize web
experiences for specific customers in real time).
Today
Only a fraction of data from connected devices
— Financial analysts use alternative visualization
is ingested, processed, queried, and analyzed in real
tools, potentially leveraging augmented reality/
time because of the limits of legacy technology
virtual reality (AR/VR) to visualize analytics for
structures, the challenges of adopting more modern
strategic decisions involving multiple variables
architectural elements, and the high computational
rather than being limited to the typical two-
demands of intensive, real-time processing jobs.
dimensional dashboards common today.
Companies often must choose between speed and
computational intensity, which can delay more
Key enablers
sophisticated analyses and inhibit the implemen­
— a view of the full business architecture to
tation of real-time use cases.
understand integration across assets, processes,
insights, and interventions and to enable the
By 2025
identification of real-time opportunities
Vast networks of connected devices gather and
transmit data and insights, often in real time. How
— more powerful edge-computing devices (IoT
data is generated, processed, analyzed, and
sensors, for example), so that even the most
visualized for end users is dramatically transformed
basic devices generate and analyze usable data
by new and more ubiquitous technologies, such as
“at the source”
kappa or lambda architectures for real-time analysis,
leading to faster and more powerful insights. Even
— advanced-connectivity infrastructures, such as
the most sophisticated advanced analytics are
5G, to support high-bandwidth, low-latency data
reasonably available to all organizations as the cost
from connected devices
of cloud computing continues to decline and
more powerful “in-memory” data tools come online
— in-memory computing for faster and more
(for example, Redis, Memcached). Altogether,
effective computations for intensive
this enables many more advanced use cases for
analytics jobs
delivering insights to customers, employees,
and partners.
How to get started
— Take advantage of a road-tested reference data
Everyday applications
architecture that enables the modularity,
— Maintenance teams for physical assets, such as
flexibility, and scalability needed to support
those in factories, regularly leverage networks of
these capabilities.6
connected sensors to detect maintenance
needs in real time.
— Evolve to a cloud-enabled data platform to meet
future data and analytical needs, such as real-
— Product developers use unstructured data and
time capabilities.7
unleash unsupervised machine-learning
algorithms on web data to detect deeply hidden
— Learn about the future of cellular-enabled
patterns and develop a much richer under­
computing devices.8
standing of customers than is possible today (for

6
Sven Blumberg, Jorge Machado, Henning Soller, and Asin Tavakoli, “Breaking through data-architecture gridlock to scale AI,” McKinsey,
January 26, 2021.
7
“Three actions CEOs can take to get value from cloud computing,” McKinsey Quarterly, July 21, 2020.
8
Ferry Grijpink, Kasia Jodlowska, Mark Patel, and Rutger Vrijen, “Reliably connecting the workforce of the future (which is now),” McKinsey,
April 14, 2021.

4 The data-driven enterprise of 2025

 3. Flexible data stores enable integrated, — Transportation and logistics companies use
ready-to-use data real-time location data and sensors embedded
into vehicles and transportation networks to
Today
develop digital twins of supply chains or trans­
Though the proliferation of data is driven by
por­tation networks, enabling a range of
unstructured or semistructured data, most usable
potential use cases (such as what-if simulations,
data is still organized in a structured fashion using
interaction monitoring, and real-time
relational database tools. Data engineers often
location insights).
spend significant time manually exploring data sets,
establishing relationships among them, and
— Construction teams crawl and query unstruc­
joining them together. They also frequently must
tured data from sensors on buildings to derive
refine data from its natural, unstructured state
insights that allow them to streamline design,
into a structured form using manual and bespoke
production, and operations; for instance, they
processes that are time-consuming, not scalable,
can simulate the financial and operational impact
and error prone.
of selecting different types of materials for
construction projects.
By 2025
Data practitioners increasingly leverage an array of
Key enablers
database types—including time-series databases,
— a modern data architecture to support more
graph databases, and NoSQL databases—enabling
flexible data stores
more flexible ways of organizing data. This allows
teams to query and understand relationships
— development of data models and digital twins to
between unstructured and semistructured data
replicate real-world systems
easier and faster, which accelerates development
of new AI-driven capabilities and the discovery
How to get started
of new relationships in the data to drive innovation.
— Implement culture and technology changes to
Combining these flexible data stores with advances
modernize your data architecture.9
in real-time technology and architecture also
enables organizations to develop data products,
— Identify critical data sets (such as customer
such as “customer 360” data platforms and
purchase frequency, customer attributes) that
digital twins—real-time-enabled data models of
could later be organized into data assets (for
physical entities (such as a manufacturing facility,
example, a complete view of the customer) and
supply, or even the human body). This enables
develop a taxonomy for these data assets
sophisticated simulations and what-if scenarios
(for example, a business-data product such as
using traditional machine-learning capabilities
“customer 360”).
or more advanced techniques such as reinforce­
ment learning.
— Explore flexible ontologies and knowledge
graphs to map the relationship between
Everyday applications
different classes of data and data points.
— Financial institutions regularly use graph-
database technology and a common data model
— Upgrade existing digital simulators,
to stream and integrate customer data from
replatforming them onto a cloud environment
multiple sources (marketing systems, enterprise-
and updating APIs, to support more
resource-planning systems, web data) into a
sophisticated AI capabilities such as
single, unified, 360-degree view of the customer
reinforcement learning.10
that can be modeled in real time.

9
“Breaking through data-architecture gridlock,” January 2021.
10
Jacomo Corbo, Oliver Fleming, and Nicolas Hohn, “It’s time for businesses to chart a course for reinforcement learning,” McKinsey, April 1, 2021.

The data-driven enterprise of 2025 5

 4. Data operating model treats data like Everyday applications
a product — Dedicated teams at retail companies develop
data products, such as “product 360,” and
Today
ensure that the data asset continues to evolve to
An organization’s data function, if one exists outside
meet the needs of critical use cases.
of IT, manages data using top-down standards,
rules, and controls. Data often has no true “owner”
— Healthcare organizations, including payers and
ensuring that it is updated and ready for use in
healthcare analytics firms, stand up product
various ways. Data sets are also stored—sometimes
teams to develop, maintain, and evolve “patient
in duplication—across sprawling, siloed, and often
360” data products to improve health outcomes.
costly environments, making it difficult for users
within an organization (such as data scientists looking
Key enablers
for data to build analytics models) to find, access,
— a data strategy that identifies and prioritizes
and integrate the data they need quickly.
business cases for data

By 2025
— understanding of the organization’s data
Data assets are organized and supported as
sources and the types of data they hold
products, regardless of whether they are used by
internal teams or external customers. These
— an operating model that establishes a data-
data products have dedicated teams, or “squads,”
product owner and team—which could include
aligned against them to embed data security, evolve
analytics professionals, data engineers,
data engineering (for example, to transform data
information-security specialists, and other roles
or continuously integrate new sources of data), and
as needed
implement self-service access and analytics tools.
Data products continuously evolve in an agile manner
How to get started
to meet the needs of consumers, leveraging DataOps
— Embed AI teams in the business and empower
(DevOps for data) and continuous integration
them to design, develop, deploy, and continually
and delivery processes and tools. Altogether, these
enhance new AI-driven products using these
products provide data solutions that can more
data products.11
easily and repeatedly be used to meet various
business challenges and reduce the time and cost
— Employ a data-governance operating model
of delivering new AI-driven capabilities.
that ensures data quality and treats data like
a product.12

11
“Getting AI to scale,” May–June 2021.
12
Bryan Petzold, Matthias Roggendorf, Kayvaun Rowshankish, and Christoph Sporleder, “Designing data governance that delivers value,”
McKinsey, June 26, 2020.

6 The data-driven enterprise of 2025

 5. T
 he chief data officer’s role is Key enablers
expanded to generate value — data literacy among business-unit leads and
their teams to create energy and urgency to
Today
engage with CDOs and their teams
Chief data officers (CDOs) and their teams function
as a cost center responsible for developing and
— an economic model, such as an automated
tracking compliance with policies, standards, and
profit-and-loss tracker, for recognizing and
procedures to manage data and ensure its quality.
attributing data and costs

By 2025
— top data talent with an eye for innovation
CDOs and their teams function as a business unit
with profit-and-loss responsibilities. The unit, in
— adoption of venture-capital-style incubator
partnership with business teams, is responsible for
operating models to support experimentation
ideating new ways to use data, developing a
and innovation
holistic enterprise data strategy (and embedding it
as part of a business strategy), and incubating
How to get started
new sources of revenue by monetizing data services
— For CDOs, begin conversations with business-
and data sharing.
unit leaders to identify opportunities for
leveraging data to drive business value.
Everyday applications
— Healthcare CDOs work in partnership with
— Develop holistic priorities, underpinned by
business units to deliver new subscription-based
scorecards and metrics, that cover
services for patients, payers, and providers
organizational health, talent, and culture, as
that can improve patient outcomes. Such
well as data quality.
services might include tailoring treatment plans,
more accurately flagging miscoded medical
— Reinforce the ethical use of data to ensure that
transactions, and improving drug safety.
new revenue-generating data services align with
corporate values and culture.13
— Bank CDOs commercialize internal data-
oriented services, such as fraud monitoring and
anti-money-laundering services, on behalf of
government agencies and other partners.

— Consumer-products CDOs partner with the

sales team to use data to drive sales
conversion and share responsibility for meeting
target metrics.

13
Tech: Forward, “Ethical data usage in an era of digital technology and regulation,” blog entry by Ewa Janiszewska-Kiewra, Jannik Podlesny, and
Henning Soller, McKinsey, August 26, 2020.

The data-driven enterprise of 2025 7

 6. D
 ata-ecosystem memberships are researchers and anonymized patient data
the norm collected by the healthcare provider) so that
each company can better achieve its goals.
Today
Data is often siloed, even within organizations. While
— Financial-services organizations tap data
data-sharing arrangements with external
exchanges to create new capabilities—
partners and competitors are increasing, they are
for example, to support socially conscious
still uncommon and often limited.
investors by providing an environmental,
social, and governance (ESG) score to publicly
By 2025
traded companies.
Large, complex organizations use data-sharing
platforms to facilitate collaboration on data-driven
Key enablers
projects, both within and between organizations.
— adoption of common data models to facilitate
Data-driven companies actively participate in a data
ease of data collaboration
economy that facilitates the pooling of data to
create more valuable insights for all members. Data
— development of data alliances and sharing
marketplaces enable the exchange, sharing, and
agreements; several data-sharing
supplementation of data, ultimately empowering
platforms have emerged in recent years to
companies to build truly unique and proprietary data
facilitate the exchange of data both within
products and gain insights from them. Altogether,
and among institutions
barriers to the exchange and combining of data are
greatly reduced, bringing together various data
How to get started
sources in such a way that the value generated is
— Read more about the different types of data
much greater than the sum of its parts.
ecosystems and best practices for a successful
ecosystem. There are examples in financial
Everyday applications
services, retail, and healthcare.14
— Manufacturers share data with their partners
and peers through open manufacturing
— Choose the data-ecosystem archetypes that will
platforms to build a more holistic view of
be most important for your organization.15
worldwide supply chains.
— Adopt data-sharing tools, protocols,
— Pharmaceutical and healthcare organizations
and procedures.
pool their respective data (for example, clinical-
trial data gathered by pharmaceutical

14
See Violet Chung, Miklós Dietz, Istvan Rab, and Zac Townsend, “Ecosystem 2.0: Climbing to the next level,” McKinsey Quarterly, September 11,
2020; “Financial data unbound: The value of open data for individuals and institutions,” McKinsey Global Institute, June 24, 2021; Julien Boudet,
Jess Huang, Phyllis Rothschild, and Ryter von Difloe, “Preparing for loyalty’s next frontier: Ecosystems,” McKinsey, March 5, 2020; and
Shubham Singhal, Basel Kayyali, Rob Levin, and Zachary Greenberg, “The next wave of healthcare innovation: The evolution of ecosystems,”
McKinsey, June 23, 2020.
15
Mohammed Aaser, Kumar Kanagasabai, Marcus Roth, and Asin Tavakoli, “Four ways to accelerate the creation of data ecosystems,”
November 23, 2020.

8 The data-driven enterprise of 2025

7. Data management is prioritized data-quality issues. Altogether, these efforts enable
and automated for privacy, organizations to build greater trust in both the
security, and resiliency data and how it’s managed, ultimately accelerating
adoption of new data-driven services.
Today
Data security and privacy are often viewed as Everyday applications
compliance issues, driven by nascent regulatory — Retailers with an online presence specify the
data-protection mandates and consumers data from consumers that they collect and
beginning to realize how much of their information is develop consumer portals to obtain consent
collected and used. Data-security and -privacy from users and allow them to “opt in” to
protections are often either insufficient or monolithic, personalized services.
rather than tailored to individual data sets. Providing
employees with secure data access is a highly — Healthcare and governmental institutions
manual process, making it error prone and lengthy. with highly sensitive data institute advanced
Manual data-resiliency processes make it data-resiliency protocols that automatically
difficult to recover data quickly and fully, creating back up data multiple times daily and, when
risks for lengthy data outages that affect needed, identify the “last good copy” and
employee productivity. restore it seamlessly.

By 2025 — Retail banks automatically provision credit-card

Organizational mindsets have fully shifted toward data needed to support customer-facing
treating data privacy, ethics, and security as areas applications, specifically during development or
of required competency, driven by evolving regulatory testing, to improve developer productivity
expectations such as the Virginia Consumer Data and provide access to data more efficiently and
Protection Act (VCDPA), General Data Protection securely than is possible with traditionally
Regulation (GDPR), and California Consumer manual efforts today.
Privacy Act (CCPA); increasing consumer awareness
of their data rights; and the increasingly high Key enablers
stakes of security incidents. Self-service provisioning — elevating the importance of data security
portals manage and automate data provisioning throughout the organization
using predefined “scripts” to safely and securely
provide users with access to data in near real time, — rising consumer awareness of, and active
greatly improving user productivity. involvement in, individual data-protection rights

Automated, near-constant backup procedures — adoption of automated database-administration

ensure data resiliency; faster recovery procedures technologies for automated provisioning,
rapidly establish and recover the “last good copy” processing, and information management
of data in minutes rather than days or weeks, thus
minimizing risks when technological glitches occur. — adoption of cloud-based data-resiliency and
AI tools become available to more effectively -storage tools to facilitate automatic backup and
manage data—for example, by automating the restoration of data
identification, correction, and remediation of

The data-driven enterprise of 2025 9

 How to get started — Create a road map for migrating to new
— Consider adopting a data-ethics framework to automatic provisioning and resiliency
understand and evaluate potential ethical capabilities as they evolve.
and regulatory ramifications of data and analytics
activity, especially involving consumer data.16 — Adopt a frequent, iterative approach to
developing, reviewing, and revising governance
— Consider leveraging cloud tools to store, and control protocols to take advantage of
manage, and secure priority data, and, for data forthcoming opportunities to automate database
already residing on the cloud, leverage administration—for example, by setting up a self-
automated backup and resiliency capabilities service provisioning portal and mandating
and tools as part of cybersecurity policies.17 automated backup and restoration procedures
on compatible data platforms.

Neil Assur is an associate partner in McKinsey’s Philadelphia office, and Kayvaun Rowshankish is a partner in
the New York office.

Designed by McKinsey Global Publishing

Copyright © 2022 McKinsey & Company. All rights reserved.

16
See “Ethical data usage,” August 2020; and Venky Anant, Lisa Donchak, James Kaplan, and Henning Soller, “The consumer-data opportunity
and the privacy imperative,” McKinsey, April 27, 2020.
17
“Security as code: The best (and maybe only) path to securing cloud applications and systems,” McKinsey, July 22, 2021.

10 The data-driven enterprise of 2025