7 characteristics of the data-driven enterprise: McKinsey’s 2025 forecast revisited

Updated on October 19, 2026: data reviewed and updated.

7 characteristics of the data-driven enterprise: McKinsey’s 2025 forecast revisited

In a forecast published in early 2022, McKinsey predicted that by 2025 the data-driven enterprise would be the norm: smart workflows and seamless interaction between people and machines would be as standard as a company’s balance sheet, and most employees would use data to optimize nearly every aspect of their work.

2025 has now passed, so this is a good time to hold that forecast up against reality and see where your own business stands.

McKinsey set out 7 characteristics that define this new data-driven enterprise. At the time the forecast was written, many companies already showed at least a few of these characteristics, while many others were only at the very beginning of the journey.

The companies that move fastest will capture the most value from data-supported capabilities. For example, according to McKinsey, January 2022, companies that get at least 20% of their EBIT (earnings before interest and taxes) from AI are more likely to have adopted data practices that serve as a foundation.

This guide is meant to help executives understand the characteristics of the new data-driven enterprise and the capabilities they can put in place to get there. It also points to resources for going deeper into how to apply data across the organization.

Characteristic 1: Data is embedded in every decision, interaction and process

The reality in 2022

Organizations tended to apply data-driven approaches, from predictive systems to AI-driven automation, in a patchy way across the organization. That left value on the table and led to inefficiency. Many business problems were still being tackled through traditional approaches and took months or even years to resolve.

The forecast for 2025

Nearly all employees use data naturally and regularly to support their work. Instead of solving problems by default through building long roadmaps that can sometimes take years to deliver, they are empowered to ask how new and innovative data techniques could solve the challenge in a few hours, a few days or a few weeks.

Organizations are able to make better decisions, automate the basic activities that happen every day, and make routine decisions far more frequently than before. Employees are freed up to focus on more “human” areas such as innovation, collaboration and communication. A data-driven culture drives continuous performance improvement, so that customer and employee experiences start to stand apart, and it also makes room for sophisticated new applications that are not yet widely used today.

How to get started

• Learn how to transform into an AI-enabled organization, and learn from AI leaders how to harness the power of data.
• Start upskilling employees on how to use data and AI.
• Learn how to reimagine each workflow, journey and function so it can make use of data and AI.
• Build a vision for the organization that is grounded in data.

Characteristic 2: Data is processed and delivered in real time

The reality in 2022

Only a small share of the data coming from connected devices was ingested, processed, queried and analyzed in real time. The reasons were the limits of legacy technology structures, the difficulty of adopting more modern architectural elements, and the heavy computing demands of intensive or real-time processing jobs. Companies often had to choose between speed and computational intensity, which could delay more complex analyses and hold back the rollout of use cases that need real-time data.

The forecast for 2025

Vast networks of connected devices gather and transmit data and surface insights in real time. The way data is generated, processed, analyzed and shown to end users changes dramatically thanks to new and more widely available technologies. For example, kappa or lambda architectures are used for real-time analysis, producing faster and more powerful insights. Both are designs for handling streams of data as they arrive: lambda runs a real-time layer alongside a batch layer, while kappa treats everything as a single stream. Even the most complex advanced analytics become reasonably available to every organization as the cost of cloud computing keeps falling and more capable “in-memory” data tools come online (for example, Redis and Memcached). Taken together, this opens up many advanced use cases for delivering insights to customers, employees and partners.

How to get started

• Take advantage of tested reference data architectures, and make use of the modularity, flexibility and scalability needed to support these capabilities.
• Evolve toward a cloud-enabled data platform that can meet future analytics and data needs, such as real-time capabilities.
• Learn about the future of mobile-enabled computing devices.

Characteristic 3: Flexible data stores make integrated, ready-to-use data possible

The reality in 2022

Although the growth in data was driven by unstructured or semi-structured data, most usable data was still organized in a structured way using relational database tools. Data engineers often spent most of their time exploring datasets by hand, establishing relationships between them and joining datasets together. They also frequently had to refine data from its natural, unstructured state into a structured form using custom, manual processes. The result was work that was time-consuming, could not scale, and was prone to errors.

The forecast for 2025

Data practitioners make growing use of many types of databases, including time-series databases (built for data points recorded over time), graph databases (built for relationships between entities) and NoSQL databases (which do not require a fixed table structure). All of these allow more flexible ways of organizing data than traditional tables. Teams can then query and understand the relationships within unstructured and semi-structured data more easily and more quickly, which speeds up the development of new AI-driven capabilities and the discovery of new relationships in the data that lead to innovation. Combining these flexible data stores with advances in real-time technology and architecture also lets organizations develop data products, such as “customer 360” data platforms and digital twins. Digital twins are real-time data models of physical entities, such as a factory, a supply chain or even the human body. They make it possible to run complex simulations and what-if scenarios using traditional machine learning or more advanced techniques such as reinforcement learning.

How to get started

• Make the cultural and technological changes needed to modernize your data architecture.
• Identify the important datasets (such as customer purchase frequency and customer attributes), organize them into data assets (for example, a complete view of the customer), and build the ability to classify those data assets (for example, a business data product such as “customer 360”).
• Explore flexible ontologies and knowledge graphs to map the relationships between different data classes and data points.
• Upgrade your existing digital simulators, re-platform them in the cloud and update their APIs, so they can support more complex AI capabilities such as reinforcement learning.

Characteristic 4: The data operating model treats data like a product

The reality in 2022

An organization’s data function, if it existed outside IT at all, managed data through top-down standards, rules and controls. Data often had no real “owner” responsible for making sure it stayed up to date and ready to use in different ways. Datasets were also stored, sometimes in duplicate, across sprawling, siloed and costly environments. That made it hard for users inside the organization (such as data scientists looking for data to build analytical models) to quickly find, access and integrate the data they needed.

The forecast for 2025

Data assets are organized and supported as products, whether they are used by internal teams or by external customers. These data products have dedicated teams, or “squads”, attached to them. The squads build in data security, carry out data engineering (for example, to transform data or to keep integrating new data sources) and roll out self-service access and analytics tools. Data products keep evolving in an agile way to meet the needs of the people who consume them, using DataOps (DevOps for data) along with continuous integration and delivery processes and tools. Taken together, these products provide data solutions that can be reused many times and more easily to meet different business challenges, while cutting the time and cost of delivering new AI-driven capabilities.

How to get started

• Integrate AI teams into the business and give them the authority to design, develop, deploy and continuously improve new AI-based products using these data products.
• Use a data governance operating model that ensures data quality and treats data as a product.

Characteristic 5: The chief data officer’s role expands to generate value

The reality in 2022

The chief data officer (CDO) and their team operated as a cost center. Their job was to develop data management policies, standards and procedures, and to monitor compliance with those rules, all in order to ensure the quality of the company’s data.

The forecast for 2025

CDOs and their teams operate as a business unit with profit and loss responsibility. Working with the business teams, the unit is responsible for coming up with new ways to use data, developing a comprehensive enterprise data strategy (and embedding data as part of the business strategy), and creating new revenue streams by monetizing data services and data sharing.

How to get started

• If you are a CDO, start conversations with business unit leaders to identify opportunities to use data to increase enterprise value.
• Develop overall priorities, backed by scorecards and metrics, that cover organizational health, talent and cultural factors as well as data quality.
• Promote the ethical use of data to make sure that data services generating new revenue stay in line with the company’s values and culture.

Characteristic 6: Membership in data ecosystems is the norm

The reality in 2022

Data was often siloed, hidden away even within a single organization. Data-sharing agreements with external partners and even competitors were on the rise, but they were still uncommon and limited in scope.

The forecast for 2025

Large, complex organizations use data-sharing platforms to support collaboration on data-driven projects, both within and between organizations. Data-driven companies take an active part in a data economy that makes it easier to pool data and generate more valuable insights for every member. Data marketplaces allow data to be exchanged, shared and supplemented, which ultimately lets companies build truly unique, proprietary data products and to draw deeper insights from their data. Overall, the barriers to exchanging and combining data are greatly reduced, bringing together many different data sources in a way where the value created is much greater than the sum of the individual parts.

How to get started

• Learn more about the different types of data ecosystems and the best practices for building a successful one.
• Pick the data ecosystem archetypes that matter most to your organization.
• Adopt data-sharing tools, protocols and processes.

Characteristic 7: Data management is prioritized and automated for privacy, security and resiliency

The reality in 2022

Data security and privacy were often seen as compliance issues, driven by new regulatory data protection mandates and by consumers starting to realize how much of their information was being collected and used. Data protection, security and privacy measures were often inadequate or applied in one blanket way, instead of being tailored to individual datasets. Giving employees secure access to the data they needed was a highly manual, error-prone and lengthy process. Manual data recovery processes made it hard to restore data quickly and completely, which created a risk of prolonged data loss that hurt employee productivity.

The forecast for 2025

The organization’s mindset shifts to seeing data privacy, ethics and security as required capabilities. The shift is driven by regulations such as Vietnam’s Personal Data Protection Law No. 91/2025/QH15 (in force from January 1, 2026), the EU’s GDPR, and the CCPA and VCDPA in the US, along with consumers’ growing awareness of their data rights and of security incidents. Self-service provisioning portals manage and automate how data is provided. They use predefined “scripts” to give users secure, safe access to data in real time, and that greatly improves user productivity.

Automated, near-continuous backup processes ensure data resiliency. Faster recovery processes can be set up quickly and restore the “last good copy” of the data in minutes instead of days or weeks, which reduces the risk when technology problems occur. AI tools become available to manage data more effectively, for example by automating how data quality issues are identified, corrected and fixed. Taken together, these efforts let organizations build greater trust in both their data and the way it is managed, which in the end speeds up the adoption of new data-driven services.

How to get started

• Consider adopting a data ethics framework to understand and assess the potential ethical and regulatory implications of data and analytics work, especially where consumer data is involved.
• Consider using cloud tools to store, manage and secure priority data. For data that is already in the cloud, make use of the automated backup and recovery tools and capabilities as part of your cybersecurity policy.

Build a roadmap toward automated provisioning and resiliency as systems evolve

Take a regular, iterative approach to developing, reviewing and revising governance protocols and controls, so you can take advantage of upcoming opportunities to automate how databases are governed. Examples include setting up a self-service provisioning portal and requiring automated backup and recovery procedures for compatible data on each platform.

This article was compiled by Adtek based on the original McKinsey content (2022), updated on October 19, 2026.