Measuring adoption & value

Six months in, someone senior will ask 'is the AI program working?' — and the difference between an answer and an anecdote is a measurement stack you designed at funding time. Three layers, each answering a different question:

• Adoption (are people using it?) — weekly active use per rolled-out team, not org-wide averages that blend pilots with bystanders. Watch the distribution, not the mean: 80% adoption in CS and 5% in sales-ops is a diagnosis (whose rollout stalled? which manager wasn't won?), while '42% overall' is a shrug. Adoption is a leading indicator — it moves months before value does, and its stalls are your earliest warning.
• Operation (is it working as designed?) — the run-level health your teams already track if they've built things properly: auto-handle rates, unsure-lane volumes, approval rates (100% = rubber stamp; investigate), error catches, latency. This layer is where 'working' gets its evidence — and where quiet degradation gets caught before it reaches the value layer. You don't review these numbers; you review that someone does.
• Value (did the line move?) — the Module 2 baselines, revisited on schedule: minutes per exception now vs. the frozen before-number, quote turnaround, dispute deflection, the two planned hires not made. Value claims cite baselines or they're stories. And value gets harvested, not assumed: freed capacity becomes value when it's redeployed to named work or absorbed growth — a management action, on someone's objectives, or the savings evaporate into slightly-longer coffee breaks and the CFO notices nothing changed.

The quarterly strategy review (one hour, five questions)

1. Portfolio: every initiative against its kill criterion and review date — continue, re-scope, or kill, decided out loud. (The zombie hunt. Protect it from agenda creep; it's the hour that keeps the program honest.)
2. Adoption: which teams moved, which stalled, and what the stalled ones share (a manager? a training gap? a genuinely bad tool?).
3. Value: baselines revisited, harvests confirmed — what did we actually bank this quarter, in the CFO's own ledger?
4. Risk & governance: the calibration gauges, incidents and their lessons, anything the regulation radar picked up.
5. Pipeline: what did the options teach? which red-flagged inventory items got unblocked? what enters the portfolio next — and what has to leave to make room? (Change capacity is still finite; the review enforces the budget.)