Data Management and Governance: Accountability, Metadata, and Quality
What to master
Governance is the decision-rights system around data: who owns it, who stewards it, who may change it, and how quality is measured. Know the distinction between data ownership (accountable business role, often a domain leader), data stewardship (day-to-day custodianship of definitions and quality), and data custodianship (technical operation of platforms). Understand the three metadata types-business (glossaries, definitions), technical (schemas, lineage, models), and operational (run stats, job logs)-and how a business glossary differs from a data dictionary. Study data quality dimensions (completeness, accuracy, consistency, timeliness, uniqueness, validity) and the DAMA-style lifecycle versus a control framework like COBIT.
Traps in questions
- Conflating accountability with technical ownership; the accountable role is usually business-side.
- Assuming a glossary and a data dictionary are interchangeable-one is semantic, one is structural.
- Treating governance as a one-time policy artifact rather than an operating process with metrics and escalation.
- Confusing data quality rules with data quality dimensions.
Worked example
A reporting table shows 4% null customer IDs. The steward defines the rule (customer ID must be non-null for active accounts), the custodian instruments a completeness check in the pipeline, and the owner accepts the residual risk threshold. The dimension is completeness; the rule is a null check; the threshold is a governance decision, not an engineering default.
Rule of thumb: every quality metric should have an owner, a threshold, and a remediation path-otherwise it is just monitoring.