Data Governance for Heads of Data

Narrative intro

Data governance is the operational discipline that converts a data estate into trust. The CDO's strategic platform decisions land here — in the catalog federation that lets workload teams find data, the lineage that makes impact analysis fast, the quality signals consumers can act on, and the data products with named owners that workload teams can actually trust. This map is the Head of Data's operational companion to the CDO's strategic platform map. Multi-platform reality is the norm at enterprise scale — Fabric, Databricks, and Snowflake often all in production, each with native catalog and governance capability. Federation rather than forced consolidation is the working pattern: Purview as the cross-platform layer, native catalogs as authoritative per platform. The discipline lives in the operational practices: lineage as code rather than as documentation, quality as continuous signal rather than periodic audit, data products with named owners rather than abstract teams.

Key takeaways

• Federation beats forced consolidation at enterprise scale — Purview cross-platform plus Unity Catalog plus Horizon, each authoritative for its platform.
• Lineage as a documentation project is accurate once and stale forever. Lineage as a discipline — automated, continuous, used in incident response — is the durable form.
• Quality SLAs without consumer-facing trust signals are governance theatre. Surface freshness, completeness, validity scores in the catalog where consumers actually look.
• Data products with named human owners outperform data as IT-owned assets every time. Move ownership to the business where possible; IT remains accountable for the platform, not the products.

Programme shape

Estimated duration

20–52 weeks

Estimated FTE

Head of Data, data governance lead, data engineering leads, named data product owners across business domains. Governance is an ongoing function; the FTE shape is the standing team.

Spend tier

moderate

Risk level

moderate

Assumes a working data platform foundation (Fabric, Databricks, Snowflake, or a mix). Risk shifts to elevated if catalog federation isn't real or lineage stays as documentation — governance theatre is the canonical failure mode for data governance programmes.

Source references

• Microsoft Purview Data Governance documentation — Canonical Purview Data Governance docs — Data Map, Data Catalog, Data Estate Insights.
• Databricks Unity Catalog documentation — Unity Catalog — governance, lineage, fine-grained access control for Databricks.
• Snowflake Horizon governance — Snowflake Horizon — native catalogue, classification, lineage.
• Data Mesh Architecture (Zhamak Dehghani) — The data product / domain ownership thinking that underpins the data products node.