Maturity Model
Semantic Governance Maturity Model
A five-level model for self-assessing how an organization governs the meaning of its business terms. Each level builds on the prior one.
Organizations can self-assess against this model. The levels describe observable signals, not aspirations. An organization is at the level whose signals best describe its current state — not the level it intends to reach.
Level 1 — Unmanaged Meaning
Signals
- No central register of business terms.
- Each team relies on tribal knowledge or local spreadsheets.
- Conflicting reports about the same metric are common but unexplained.
Capabilities
- None at the organizational level.
Limits
- Cannot answer 'what does this term officially mean?' for any critical concept.
- AI initiatives inherit the ambiguity and amplify it.
Next step
Inventory the most-contested terms and assign each an owner. The act of naming an owner is the first governance event.
Level 2 — Glossary-Based
Signals
- A central business glossary exists, often in a wiki or data catalog.
- Definitions are written but rarely updated.
- No mechanism distinguishes ambiguity from contradiction.
Capabilities
- Single authoritative source of written definitions for some terms.
Limits
- Glossary entries decay as language evolves.
- No history — only the current state.
- Cannot represent the same term with different governed meanings in different contexts.
Next step
Add governance status (proposed, approved, deprecated), name owners, and require approval before publication.
Level 3 — Governed Definitions
Signals
- Definitions have owners, statuses, and approval workflows.
- Changes are reviewed before publication.
- Drift is sometimes detected but handling is manual.
Capabilities
- Reliable single definition per term within a domain.
- Audit trail of who approved what.
Limits
- Ambiguity is still resolved by forcing one definition per term, losing context.
- No structured learning from prior decisions.
Next step
Introduce Meaning Spaces and context-disambiguated Canonical Concepts. Begin recording immutable governance events.
Level 4 — Semantic Governance
Signals
- Terms resolve to Canonical Concepts within scoped Meaning Spaces.
- Drift is automatically detected and routed to governance review.
- Provenance and governance events are recorded immutably.
Capabilities
- Same term can carry different governed meanings in different contexts without contradiction.
- AI agents resolve terms through the registry instead of guessing.
- Reuse strengthens confidence; contradictions surface as governance work.
Limits
- Learning is recorded but not yet systematically interpreted across the organization.
Next step
Use Organizational Memory and learning signals to inform governance proactively — predict where drift will occur, prioritize review by impact.
Level 5 — Organizational Learning
Signals
- Organizational Memory is queried during governance decisions.
- Semantic Stability is tracked per Canonical Concept and per Meaning Space.
- AI governance recommendations cite prior decisions and provenance.
Capabilities
- Meaning evolution can be reconstructed years after the fact.
- Governance is informed by accumulated evidence, not just current opinion.
- Enterprise AI behaves consistently across teams because it cites the same canonical meaning.
Limits
- Maturity is a continuous practice, not a finish line. Level 5 organizations still face new ambiguity and must govern it.
Next step
Sustain the practice. Treat Semantic Governance as enterprise infrastructure, not a project.