Skip to content
Log inRegister

Framework · Governance

Building a Data Governance Operating Model: Ownership, Stewardship, and Decision Rights

A practical blueprint for who decides what about data: the council, owners, and stewards, how they work together, and how to start with one business problem instead of a big program.

Dr. Abeer Alshammari8 min readPublished November 7, 2026Updated November 7, 2026

The short answer

Data governance is accountability made explicit. It answers three questions for every important data set: who is accountable for it, who decides how it may be used, and how disagreements get resolved. Start small, tie it to a real business problem, and build roles before tools.

What governance is, and is not

Governance is the set of decision rights, roles, and policies that control how data is defined, protected, and used. It is not a software product, and it is not a one-off policy document. The DAMA body of knowledge describes it as the exercise of authority and control over the management of data assets, and ISO/IEC 38505-1 applies the same governing principles to an organization use of data: responsibility, strategy, acquisition, performance, conformance, and human behavior.

The three-layer model

Data Governance Council Sets policy and priorities, resolves escalations Data Owners Accountable for a data domain and its use Data Stewards Maintain definitions, quality, and access in daily work
Decisions flow down as policy; issues and evidence flow up as escalations and metrics.
RoleAccountable forTypical decisions
Governance councilPolicy, priorities, funding of the programApprove policies, settle cross-domain disputes
Data ownerA domain such as customer, product, or financeClassification tier, who may access, acceptable use
Data stewardQuality and definitions in daily operationsResolve data issues, approve changes to definitions
Data custodianTechnical storage and securityBackups, encryption, access implementation

Standing up the model in six steps

  1. Pick one business problem. Choose a pain that leaders feel, for example inconsistent customer counts across reports, or an audit finding about access to personal data. Governance that solves a visible problem earns support.
  2. Write a short charter. One page: purpose, scope, roles, decision rights, and how often the council meets.
  3. Name owners and stewards. Choose people who already make decisions about the data, and give them explicit time in their role descriptions.
  4. Define decision rights. For each type of decision (definition change, access request, retention exception) record who proposes, who approves, and who is informed.
  5. Create an escalation path. Stewards resolve what they can; owners decide what they cannot; the council decides cross-domain conflicts. Publish response times.
  6. Run a steady cadence. A monthly council meeting, a weekly steward check-in, and a quarterly report on a small set of measures.

A worked example

Two departments report different totals for active customers. The steward for customer data finds that one counts anyone who has ever purchased and the other counts purchases in the last twelve months. The steward proposes a single definition (a purchase in the last twelve months) to the customer data owner, who approves it. The council records it in the business glossary and both reports are corrected. Cost: one meeting and one definition. Benefit: one number that leaders trust.

Common pitfalls

  • Starting with tools. A catalog without owners becomes an unmaintained list.
  • Making governance a committee that only reviews. Give it decision rights and a budget.
  • Treating stewardship as extra work with no time allocated.
  • Trying to cover every domain at once. Prove the approach on one domain first.

Implementation checklist

  • Business problem chosen and sponsor identified
  • One-page charter approved
  • Owners and stewards named for the first domain
  • Decision rights recorded for the main decision types
  • Escalation path and response times published
  • Business glossary started with the first agreed definitions
  • Three to five measures reported quarterly

Measures that matter

Track the number of critical data elements with a named owner, the age of open data issues, the percentage of policy decisions made within the agreed time, and the number of definitions agreed and published. Resist counting meetings or documents; count decisions made and problems closed.

Practical note

The most common failure is not a lack of policy but a lack of named people. If you can only do one thing this month, name an owner for your most important data set and record what that owner is accountable for.

Interactive: escalation walkthrough

Step through how a data issue travels from the people who find it to the people who decide it.

  1. Data Governance Council
  2. Data Owners
  3. Data Stewards

Step 1 of 4

A steward notices that two reports count active customers differently.

References and further reading

  1. DAMA-DMBOK: Data Management Body of Knowledge (DAMA International)Accessed November 7, 2026
  2. The DGI Data Governance Framework (Data Governance Institute)Accessed November 7, 2026
  3. ISO/IEC 38505-1:2017 Governance of data (ISO)Accessed November 7, 2026
Back to Governance