What data governance is
The rules and checks that make a figure usable for deciding. It is not a tool you buy, it is a state you keep.
Its four pillars are catalogue (what data exists and where it comes from), quality (whether the figure is right), access (who can see it) and ownership (who answers for it).
Data governance and data management: the difference
Data management is the machinery: loading, transforming, storing and serving data. Data governance is the rules and checks that say whether what comes out of that machinery can be used to decide, who answers for it and who may see it.
They get mixed up because they usually live in the same team. The test is simple: if nobody can say who answers for a figure on the dashboard, there is management and governance is missing.
Why KPIs come out wrong
Almost never because of a badly written formula. It is usually a flow that stopped bringing data and nobody noticed, a filter someone changed, a source that changed format or a report still reading a table that is no longer refreshed. The dashboard looks perfect and the figure is false.
It is measured: 26% of an organization data is not reliable, as estimated by its own data leaders.
What a data governance framework contains
A useful framework fits on one page: which data is critical, who answers for each item, which quality rules it must meet, who may access it and which periodic review checks all of the above. Without that check, the framework is one more document.
Tools help to keep it in order: Unity Catalog in Databricks or Microsoft Purview in Fabric and Power BI hold the catalogue, lineage and permissions. But no tool tells you on its own whether last night's load worked or whether a license is paid for and unused: that is measured.
How it is checked, platform by platform
With a periodic health check of what is really there: which pipelines failed silently, which reports read broken sources, who has access to what and which licenses are paid for while nobody signs in. In Salesforce, Microsoft Fabric, Databricks and Power BI, which is where the problem usually lives.
We measure and hand over the list with evidence for each point; the changes are applied by your team or by us, as you prefer.