Data Governance and Stewardship
Data governance is the decision-rights and accountability system for data; stewardship is the day-to-day execution of those decisions. The exam tests operating models, role design, policy lifecycle, escalation, and how governance is measured rather than merely charted.
Core rules
- Governance sets direction (policy, standards, priorities); management executes. A council that only reviews slides is not governing.
- Define decision domains explicitly: quality rules, access approval, reference data, issue prioritization. Each needs an owner, a decision right, and an escalation path.
- Separate accountabilities using a RACI or similar: data owners approve, data stewards define and monitor, IT delivers, auditors assure.
- Stewardship must be funded and embedded in workflows; a steward who is only a mailbox for complaints cannot enforce definitions.
- Publish policies with controls, exceptions, and review dates. Measure adoption, issue cycle time, and policy exception rates.
Worked example
Sales and billing define “active customer” differently: sales counts any account with an open opportunity; billing counts accounts with an invoice in 12 months. A steward documents both definitions, maps them to source fields, and convenes the owner and finance lead. The council approves a single enterprise definition for reporting, permits a sales-specific variant for pipeline, and records the decision, owner, effective date, and exception process. The metric is added to a glossary and monitored monthly. The reasoning: governance resolved a decision right, not a data defect, and documented allowed variants to prevent rework.
Traps
- Confusing a governance council with a project steering committee.
- Appointing stewards without time, mandate, or decision rights.
- Writing policies with no control, exception, or metric.
- Treating data ownership as IT ownership.