How to unlock the full value of data?

Manage it like a product

Creating reusable data products and patterns for piecing together data technologies
enables companies to derive value from data today and tomorrow.

by Veeral Desai, Tim Fountaine, and Kayvaun Rowshankish

© Getty Images

June 2022
 Our recently published article in Harvard Business Review, “A better way to put your data to work,” details
how to establish a sustainable path to value. The key is to manage data just as you would a consumer
product. Here, we present a visual summary of this approach

Today’s predominant—and largely unsuccessful—approaches to data

Organizations typically employ either a grassroots or big-bang data strategy—neither of which enables
them to make the most of their data investments.

Grassroots approach. In a grassroots approach, individual teams must piece together the data and
technologies they need. This approach results in significant duplication of efforts and a tangle of bespoke
technology architectures that are costly to build, manage, and maintain.

Big-bang strategy. At organizations employing the big-bang strategy, a centralized team extracts, cleanses,
and aggregates data en masse. This approach can eliminate some of the rework that occurs, but it’s often
not aligned with business use cases and therefore fails to support end users’ specific needs. End users
often struggle to confirm that the data provide the necessary level of governance and quality, which
limits the time savings. Later work on new use cases that are aligned with business value often triggers a
grassroots approach and its associated problems.

These strategies fail to lay the foundation for current and future use cases that will create value (Exhibit 1).

Exhibit 1
Neither the
Neither the grassroots
grassrootsnor
nor big-bang
big-bangapproach
approachwill
willenable
enableorganizations
organizationstoto
derive sustainable
derive sustainable value
value from
from data.
data.
Example of a complex and ine cient data strategy

Data pipelines designed for Di erent technologies are

batch and real-time delivery are employed for each use case,
Flow of data fragmented and duplicative. adding expense and complexity.

Systems of Data Use-case-speci c Use-case-speci c Use

record platform data sets technologies cases

Digital banking
app
Core
Data
processing
warehouse
systems Investing portal

Predictive
cross-selling model
External Raw data
data lake
Predictive
churn model

Financial
Unstructured Operational report
data data store
Industry data
ecosystem

Data for each domain, such as the

customer, is ine ciently reworked for
every use case; quality, de nitions, and
formats vary.

Source: “A better way to put your data to work,” Harvard Business Review, June 2022

2 How to unlock the full value of data? Manage it like a product

A better approach: Managing data like a product
We find that when companies instead manage data like a consumer product—be it digital or physical—
they can realize near-term value from their data investments and pave the way for quickly getting more
value tomorrow (Exhibit 2).

Exhibit 2
Data products are similar to consumer products in many ways.
Data products are similar to consumer products in many ways.
Examples of similarities

Digital product Physical product

Example: Computer app Example: Car Data product
Product features

Customization of base App enables users to Car buyers may Data products can be
product for di erent personalize the layout, purchase a variety of wired to support
users color schemes, and special options (eg, di erent systems that
content displayed and leather upholstery, consume data, such as
to select plans and tinted windows, advanced analytics or
pricing structures that antitheft systems) reporting systems
meet their needs

Delivery of regular Automatic downloads of New models New data

product enhancements new functionality
Engine modi cations Support for additional
that boost fuel economy consumption archetypes

Production e ciency

Reuse of existing Software developers Automakers use a Organizations reuse

processes, machinery, reuse blocks of code common chassis on vastly blueprints and modular
and components di erent cars technologies for
consumption archetypes
across products

How to unlock the full value of data? Manage it like a product 3

 Data products provide all the data on one entity. A data product delivers a high-quality, ready-to-use set
of data that people across an organization can easily access and apply to different business challenges.
For example, a data product could provide a 360-degree view of an important entity, such as customers,
employees, product lines, or branches. Or it could deliver a given data capability, such as a digital twin that
replicates the operation of real-world assets.

Data products are wired to enable standard types of consumption. Data products incorporate the wiring
necessary for different business systems, such as digital apps or reporting systems, to “consume” the data
(Exhibit 3). Each type of business system has its own set of requirements for how data is stored, processed,
and managed; we call these “consumption archetypes.”

While an organization might have hundreds of use cases on its road map, they typically fit one of five primary
consumption archetypes. Data products built to support one or more of these consumption archetypes can
easily be applied to multiple business applications with similar archetypes.

Exhibit 3
Five
Fiveconsumption
consumptionarchetypes describe
archetypes the primary
describe ways ways
the primary that users
thatconsume
users data.
consume data.

Archetype Requirements Example uses

Digital applications Specific data cleaned and stored in a A marketing trends app or vehicle
particular format and frequency (eg, tracking app
delivering access in real time to event
streams of GPS or sensor data)

Advanced analytics Data cleaned and delivered at a certain Simulation and optimization engines
systems frequency and engineered to allow processing
by machine learning and AI systems

Reporting systems Highly governed data with clear definitions— Operational or regulatory
managed closely for quality, security, and compliance dashboards
changes—aggregated at a basic level and
delivered in an audited form

Discovery sandboxes A combination of raw and aggregated data Ad hoc analysis for exploring new
use cases

External data-sharing Adherence to stringent policies and Banking systems that share
systems agreements about where the data sit and fraud insights
processes for managing and securing data

4 How to unlock the full value of data? Manage it like a product

Data products enable more speed and efficiency. Teams using data products don’t have to waste time
searching for data, processing it into the right format, and building bespoke data sets and data pipelines—
an effort that ultimately creates an architectural mess and governance challenges (Exhibit 4).

Exhibit 4
A data
A data product
product approach
approach results
resultsin
in standardization
standardizationthat
thatsaves
savestime
timeand money.
and money.
Example of an efficient data product approach

Flow of data

Systems of Data Data Consumption Use

record platform products archetypes cases

Digital Digital banking

Vendor applications app
Core
Data
processing
warehouse Investing portal
systems Advanced
Customer analytics Predictive
cross-selling model

External Raw data Branch Reporting

Predictive
data lake churn model

Financial
Product/ report
service External data
sharing
Industry data
Unstructured Operational ecosystem
data data store Employee/
agent Discovery
sandbox Data exploration

Source: “A better way to put your data to work,” Harvard Business Review, June 2022

The benefits of this approach can be significant:

— New business use cases can be delivered as much as 90 percent faster.

— Total cost of ownership, including technology, development, and maintenance costs, can decline by 30
percent.

— The risk and data-governance burden can be reduced.

How to unlock the full value of data? Manage it like a product 5

 Getting started with data products

Success in product development requires an operating model that ensures dedicated management and
funding, the establishment of standards and best practices, performance tracking, and quality assurance
(Exhibit 5). Success with data products is no different.

Exhibit 5
Managing data like a product requires the right operating model.
Managing data like a product requires the right operating model.

Human resources

Use case Data

teams utility Dedicated team:
Product manager
Employee data product Data engineers
Marketing
Data center of Contractor data product Data architects
Use case Data excellence Data modelers
teams utility
Data platform engineers
Site reliability engineers
Customer data product
Manufacturing
Branch data product Provides: Cross-functional resources:
Standards Use case Data Source system SMEs¹
Architectural blueprints for teams utility
Data stewards
consumption archetypes
Vendor data product IT experts
Tooling
Factory data product
Best practices
Specialized talent

Enterprise IT Provides platform for:

Modular, extendable
technologies for each
consumption archetype
Security

¹ Small and medium-size enterprises.

— Dedicated management and funding. Each data product should have a product manager and a team
consisting of data engineers, data architects, data modelers, data platform engineers, and site reliability
engineers who are funded to build and continually improve their product and enable new use cases.
These teams should sit within a data utility group inside business units. This organizational structure
gives them ready access to the experts they need (including business subject-matter, operational,
process, legal, and risk experts) to develop useful and compliant data products. In addition, it gives the
teams access to user feedback, which helps them continue to improve products and identify new uses.

— Standards and best practices. We find organizations are most successful when they institute standards
and best practices for building data products across the organization. This work is typically handled
by a data center of excellence. Establishing standards and best practices includes defining how teams
will document data provenance, audit data use, and measure data quality, as well as designing how the

6 How to unlock the full value of data? Manage it like a product

 necessary technologies should fit together for each consumption archetype so they can be reused
across all data products.

— Performance tracking. To confirm that their products meet end-user needs and are continually
improving, data product teams should measure the value of their work. Relevant metrics may include
the number of monthly users for a given product, the number of times a product is reused across the
business, satisfaction scores from surveys of data users, and the return on investment of use cases
enabled.

— Quality assurance. Because quality issues can erode end-user trust and retention, data product teams
closely manage data definitions (for instance, whether the definition of customer data is limited to active
customers or includes active and former customers), availability, and access controls that meet the right
level of governance for each use case. To confirm data integrity, they work closely with data stewards
who own data source systems.

For a deeper look at how leaders can manage data as they manage a product, read “A better way to put your
data to work,” on hbr.org.

Veeral Desai is a senior expert in McKinsey’s Sydney office, where Tim Fountaine is a senior partner; Kayvaun Rowshankish
is a partner in the New York office.

Copyright © 2022 McKinsey & Company. All rights reserved.

How to unlock the full value of data? Manage it like a product 7