Almost every large organization has a data governance program. Almost every large organization's data governance program is, to varying degrees, a fiction.
I don't say this to be harsh. I say it because the gap between governance-on-paper and governance-in-practice is one of the most consistent patterns I've observed across financial services organizations of different sizes, different technology stacks, and different levels of sophistication.
What Governance Theater Looks Like
The tells are consistent:
Data ownership is assigned but not exercised. The RACI says Finance owns the customer revenue metric. In practice, Finance doesn't know what query produces it, doesn't know when the definition last changed, and learns about data quality issues from downstream reports rather than from monitoring. Assigned ownership without operational accountability isn't ownership. It's decoration.
The data catalog exists but nobody uses it. The catalog was populated during a governance initiative two years ago. It hasn't been updated since. New datasets don't get catalogued. The business users who were supposed to consult it don't because the information is out of date and they've learned not to trust it. A catalog that isn't maintained is worse than no catalog — it creates false confidence.
Quality rules are documented but not enforced. The governance policy states that customer identifiers must be non-null, unique, and follow a specific format. The operational reality is that three legacy systems produce non-conforming IDs, there's a reconciliation job that "handles" it, and nobody wants to fix the source because it would require changes to systems that "work."
Governance meetings happen; governance decisions don't. The data governance committee meets monthly. It discusses issues. Tickets are raised. Rarely do those tickets result in anything being changed — because changing things requires resources, and the governance committee doesn't control resources. It advises. It doesn't decide.
Why This Happens
Governance programs are usually launched in response to pain — a failed audit, a regulatory finding, a high-profile data error that reached an executive. The initiative gets funded, a framework gets designed, a structure gets stood up.
Then the crisis passes. The immediate pain is gone. The harder work of actually changing how data is produced, owned, and managed across the organization runs into organizational resistance — because it requires people to change how they work, accept accountability for problems they didn't personally create, and invest time in maintenance that doesn't generate visible output.
The governance framework survives. The operational change doesn't happen. The theater continues.
What Actual Governance Looks Like
Real governance has a few observable characteristics that distinguish it from the theater version.
Data quality issues have owners who feel them. When a quality issue occurs, there's a person (not a committee) who is accountable for both the immediate fix and the systemic resolution. That person's performance evaluation reflects the quality outcomes of the data they own. The accountability is real.
The catalog is a working tool, not a documentation artifact. New datasets get catalogued as a condition of deployment, not as an afterthought. The business users who rely on data consult the catalog because the information there is accurate and current.
Governance decisions can block deployments. This is the real test. If an engineering team can deploy a pipeline that violates data standards because governance is advisory and the deadline is tight, governance doesn't govern anything. Real governance has teeth — not punitively, but structurally.
Quality metrics are visible to leadership. The executives who set priorities see data quality metrics alongside operational and financial metrics. When quality degrades, it's visible at the level where resource allocation decisions get made.
The Hard Conversation
The honest assessment most governance programs need isn't about framework design or tool selection. It's about whether the organization is actually willing to change behavior — or just willing to say it is.
That conversation is uncomfortable because it usually points at decisions made above the data team's level. The people closest to the data often know exactly what needs to change. Getting organizational will behind the change is the harder problem, and it's a leadership problem, not a technical one.
If your governance program has been running for more than a year and the answer to "what has actually changed in how data is produced and owned" is thin, you don't have a governance problem — you have an organizational will problem. Diagnosing that clearly is the first step to actually fixing it.