Data Governance
What governance actually is
Data governance is the exercise of authority and control over data management - planning, monitoring, and enforcement. The single most tested distinction is this: governance decides who decides; management does the work. Governance is about decision rights, accountability, policy, standards, and oversight, not about running ETL or fixing records.
Know the DAMA governance components: data principles, data policies and standards, data strategy alignment, the data governance operating model (roles, decision rights, processes), data stewardship, governance metrics, and change management.
Roles you must be able to separate
- Data Owner - business executive, accountable for a data domain; approves definitions, policies, access decisions.
- Data Steward - business-side, day-to-day: definitions, quality issues, issue triage.
- Data Custodian - technical: implements controls, storage, backup, access provisioning.
- Governance bodies - Data Governance Council/Steering Committee, Data Stewardship Council, Data Governance Office.
Traps
- Describing an operational task (cleansing a table, approving a schema change) and asking if it is governance - usually it is management or execution.
- Assuming governance must be centralized; operating models can be centralized, decentralized, or federated/hybrid.
- Assuming IT owns the data. The business owns; IT custodies.
- Swapping Data Owner and Data Custodian - the most common role trap.
Worked example (rule of thumb)
For any new data domain, write a RACI. Who is Accountable for the business definition? Data Owner. Who does the daily work? Steward. Who implements the technical control? Custodian. Who is informed of policy changes? Council. So: "Who approves a change to the definition of 'active customer'?" → Data Owner, on the Steward's recommendation - never the DBA.