An enterprise architecture function needs to demonstrate value. Which metrics are meaningful and which are vanity?

Vanity metrics

• Number of reviews conducted. Measures activity, not value, and rewards being a gate.
• Number of artefacts produced or models maintained. Measures cost.
• Standards published. Measures output; adoption is what matters.
• Percentage of applications documented. A complete inventory nobody consults is a maintained liability.
• Compliance rate with standards, taken alone — a high rate may mean the standards are trivial, or that non-compliant work has moved outside the governed estate.

Meaningful metrics

Outcome metrics — what changed because the function exists:

• Applications retired and the cost removed.
• Duplicate capabilities consolidated, with the saving.
• Investment decisions influenced, with the money attached — including decisions to not build something.
• Time-to-first-success on the paved road, and the proportion of new services adopting it voluntarily.
• Reduction in cross-team coordination required for a typical change.
• Incidents attributable to architectural causes — unmanaged dependencies, missing isolation, unowned data — trending down.

Leading indicators — whether the function is positioned to add value:

• Engagement timing. How early in a design teams involve architecture. Early engagement is the strongest predictor of influence.
• Voluntary engagement rate. How often teams seek architecture out before they are required to. This is the single best signal that the function delivers value rather than extracting compliance.
• Review turnaround time, since unpredictable lead time is what causes teams to route around.

The one metric worth leading with

The proportion of the estate that is genuinely governed — following the paved road, covered by guardrails, with attribution and ownership recorded.

A function with rigorous processes covering a third of the estate is providing less assurance than one with light processes covering nearly all of it, and that comparison is what most governance conversations actually need.

The measurement discipline

Capture the baseline before the intervention. EA benefits are indirect and arrive over quarters, so a retrospective baseline is contested and usually unavailable — which is why so many functions cannot demonstrate value they genuinely delivered.