Process family

Data governance process: six templates from operating model to disposal

How data governance runs as a sequence: the operating model, issue-to-closure decisions, the catalog, quality rules, lineage and lifecycle, with the glossary, access request and legal hold templates alongside.

Data governance is the set of processes that decide who owns a dataset, what it means, how good it has to be, where it came from and when it may be deleted. The six templates in this family follow that order: an operating model gives stewards their mandate, a decision process settles disputes, and the catalog, quality, lineage and lifecycle processes then run against that authority.

The sequence starts before any single dataset is touched. In the operating model an executive sponsor names domain owners, maps stewardship roles and fixes decision rights and forum cadence; every later step assumes those roles exist. When a definition is disputed, an owner is missing or a policy needs an exception, the governance process takes the issue from intake through a delegated or council decision to evidenced closure. With ownership settled, the steward registers, enriches and certifies the asset in the data catalog, the record every downstream process points back to.

Quality, lineage and lifecycle then run against that record. Data quality management turns a consumer's objective into owned rules, tolerances and monitoring, and retires the rule when it is no longer needed. Lineage documentation starts from a report or model, traces it back through transformations to its source fields, and routes sensitive or critical paths past a control reviewer. The lifecycle process closes the loop: classification on storage, sharing approval on request, a retention trigger checked against hold status, and disposal recorded as evidence rather than assumed.

The usual failure is a catalog with no decision rights behind it: assets get registered, but nobody can settle who may change a definition or grant access, and quality scores pile up with no owner. Three neighbors are listed as decisions governance owns rather than steps in the chain: the business glossary publishes the meaning a catalog entry references, the data access request lets the data owner, not the credential holder, grant time-boxed access against the stored classification, and the legal hold is why the lifecycle checks for an active hold before disposal. There is no standalone data policy or retention schedule template; both are stages inside the operating model and lifecycle.

Defect handling for one faulty dataset and the shared reference records several systems read are covered next door in data quality and master data; the pipelines and migrations that move governed data between systems are in data engineering and migration. Access to systems rather than datasets, the joiner-mover-leaver side, belongs to access governance.