How Do You Measure Data Governance Maturity?
Key takeaways
- Conventional maturity models measure process completeness: policy coverage, roles, tooling, quality measurement, escalation.
- None of those dimensions asks what a term means.
- A meaning-first model runs five levels: undocumented, documented, owned, scoped, and governs meaning in context.
- The jump that matters is three to four. Ownership without scope produces four owners who each believe they hold the company definition.
- Most organizations sit between two and three, with a prose glossary and partial ownership.
- Score one contested term rather than the whole programme, which gives a next action instead of a number.
How do you measure data governance maturity? Most models score the programme: policies written, roles filled, tools deployed, coverage achieved. Useful for a budget conversation. They can also rate a programme highly while the company holds four uncoordinated answers to the same question.
A maturity model worth running measures one thing at the top level. Can the organization state every definition of a contested term, where each applies, and how they reconcile.
What do existing maturity models measure?
Process completeness, mostly. The dimensions repeat across every published model.
- Policy coverage and documentation
- Role definition and staffing
- Tooling and automation
- Data quality measurement
- Issue resolution and escalation
Each is measurable and each improves real things. None of them asks what a term means.
What does a meaning-first model measure instead?
How much of the meaning layer is written down and machine-readable. The progression runs from memory to register.
What are the five levels?
| Level | Name | Test |
|---|---|---|
| 1 | Undocumented | Definitions live in people's heads. Disputes settle by seniority. |
| 2 | Documented | Definitions are written somewhere. Scope is absent, so each one claims the whole company. |
| 3 | Owned | Each definition has a named owner who may change it. |
| 4 | Scoped | Each definition states which systems, processes and reports it governs. |
| 5 | Governs meaning in context | Scope, ownership, change history and reconciliation are held machine-readably. Agents resolve against them. |
The jump that matters is three to four. Ownership without scope produces four owners who each believe they hold the company definition.
Where do most organizations sit?
Between two and three. Definitions exist in a glossary. Some have owners. Almost none carry scope, and the glossary is prose a person reads.
A programme can score well on a conventional maturity model from exactly this position. That is the gap the levels above are drawn to expose.
What does level five look like?
Take revenue, moved up the levels.
- Level two. A glossary entry says revenue is income from sales. Four teams read it and carry on as before.
- Level three. Finance owns the entry. Sales still reports booked value and nobody calls it a breach.
- Level four. Four definitions are registered. Recognised revenue governs statutory reporting. Booked value governs commission. Recurring revenue governs retention. Invoiced amount governs collection.
- Level five. Those scopes are machine-readable, the relationships are stated, and an agent asked about revenue names which definition it used.
Level five is the point at which an AI system stops guessing.
How do you move a level?
- Two to three. Assign one owner per definition. A week of meetings.
- Three to four. Write the scope for each. The hard part is admitting several definitions are legitimate.
- Four to five. Move the register into a system agents can query, and state the reconciliation rules.
Teams usually expect step two to be the fight. It settles quickly once nobody has to surrender a definition.
How should you use this model?
Score one term, never the whole programme. Pick the term that causes the most arguments and place it on the five levels honestly.
That gives a specific next action instead of a maturity score. Data governance for AI covers why the top level matters now, a governance framework covers the structures, and governance platforms covers what to look for in tooling.
So what should you take away?
A high maturity score and four uncoordinated definitions can coexist comfortably. Most programmes prove it every quarter.
Score one contested term against these five levels. Then move it one level.
TopQuadrant builds the systems that hold definitions, scope and ownership together. One term, one level, this quarter.
