The DAMA Data Management Body of Knowledge, the reference most courses and job descriptions draw on, treats governance as the knowledge area that oversees the others: it sets the rules under which data is modelled, stored, integrated, secured and measured for quality. In practice the rules reduce to a small number of questions asked of every dataset the organisation depends on. Who is accountable for it. What each field means, in one agreed definition. Who may read it, who may change it, and under what approval. How long it is kept and when it is deleted. How its quality is measured and who is told when the measure falls. An organisation that can answer those questions for its principal datasets is governed; the method by which it arrived at the answers is secondary.
- Owner: the role, not the person, accountable for a dataset's definition, quality and access decisions.
- Steward: the person who maintains the definitions and the catalogue entry day to day and resolves disputes about meaning.
- Catalogue: the register of datasets with their owners, definitions, sources, freshness and access rules; the artefact that makes the rest usable.
- Lineage: the record of where each figure came from and what was done to it on the way, so that a number in a report can be traced to its source system.
- Policy: the written rules on classification, access, retention and quality that the catalogue entries apply.
The reasons an organisation adopts governance are usually external before they are internal. Data-protection law requires it to know what personal data it holds, why, for how long and who can see it, and the UK framework for public-sector data quality frames the same requirements around fitness for purpose and clear ownership. Financial reporting requires figures to be traceable to their source. A data warehouse project discovers that three departments define a customer differently and needs an arbiter. Each of these produces the same register, and the organisations that build it once, for all three reasons, spend less than those that build it three times.
| Question | Without governance | With governance |
|---|---|---|
| What does active customer mean? | Three definitions in three reports | One definition in the catalogue, referenced by every report |
| Who may see salary data? | Whoever has the database password | A named role, enforced by row-level access and reviewed |
| Where did this revenue figure come from? | An analyst's memory | Lineage from the report to the source table and the transformation |
| When is this data deleted? | Never, by default | At the end of the retention period stated in the policy |
The discipline connects to data quality, which it defines the measures for and assigns the ownership of, and to security, where the access rules it sets are enforced. The platforms now carry much of the machinery, a catalogue, lineage captured from the pipelines, access rules attached to tables, which is why the AWS Data Engineer Associate, Databricks Data Engineer Associate and Microsoft Fabric Data Engineer exams each include a governance and security domain. The governance, quality and privacy courses cover the organisational side that the platforms cannot supply: the decision about who owns what.