Microsoft Purview Data Governance: Unified Catalog, Data Map, and Lineage

An enterprise deploys Purview data governance. The data engineering team configures scanning across Azure SQL databases, Synapse workspaces, and a few on-premises SQL Server instances. The catalog populates with 15,000 data assets. The project is declared complete. Six months later, a review reveals: nobody outside the data engineering team uses the catalog, the business glossary has 12 terms (all created during the implementation), data stewards were never designated, scanning hasn't been updated as new data sources were added, and the CDO who sponsored the project can't demonstrate business value to the CFO asking about the Azure consumption cost. The catalog has data. Nobody consumes it. The governance program is technically deployed and practically abandoned.

Data governance that sustains requires three things the technology alone doesn't provide. First, a governance operating model — designated data stewards with specific responsibilities, glossary ownership mapped to business domains, classification standards that reflect how the business organizes data, and a governance council that reviews and evolves standards. Second, catalog adoption — connecting the catalog to the workflows where people actually search for data (analysts looking for datasets, engineers checking lineage before changes, compliance officers verifying sensitive data location). Third, ongoing scanning and quality — new data sources scanned as they're created, data quality rules monitoring the metrics that matter, and the cadence that keeps the catalog current. Done with all three, governance becomes institutional capability. Done as a technology project, it becomes the catalog nobody opens.