Data strategy

Data strategy defines how an organization will govern, create, share and use data to support operations, decisions, analytics and artificial intelligence.

Entry type: Knowledge article

Field: Data and Technology Advisory

Last reviewed: 24 August 2026

Definition

Data strategy is the set of choices, governance arrangements and investments through which an organization aligns data assets, architecture, quality, access and analytical capabilities with business objectives and risk obligations.

In this entry

Components
Strategy process
Data governance
Analytics and AI readiness
Related concepts
Sources

Components

Value and use

Priority decisions and products
Analytics and AI use cases
Data-sharing opportunities
Value measurement

Foundation

Architecture and platforms
Models and metadata
Quality and lineage
Integration and access

Governance

Ownership and stewardship
Privacy and security
Retention and lifecycle
Standards and accountability

Strategy process

Business priorities: identify decisions, operations and services dependent on data.
Current state: assess assets, quality, architecture, skills and governance.
Target state: define data products, ownership, platforms and control principles.
Roadmap: sequence foundational work and high-value use cases.
Adoption: embed accountability, literacy, measurement and continuous improvement.

Data governance

Role	Responsibility	Common weakness
Business owner	Meaning, quality and permitted use	Ownership assigned only to IT
Data steward	Standards, definitions and issue resolution	Responsibility without authority
Technology team	Platform, integration, reliability and security	Architecture detached from business use
Risk and privacy functions	Policy, rights and control challenge	Controls introduced after design

Analytics and AI readiness

AI readiness is not achieved by centralizing every dataset. It requires usable, lawful and sufficiently reliable data for defined purposes, with lineage, access controls, evaluation data and feedback from production. The architecture should serve the operating model rather than become an end in itself.