Scorecard
A one-page document, agreed before build, that names the AI programme's outcome metric with a baseline, target, review cadence and written stop condition.
A scorecard is the single most important artifact a scoping engagement produces. It is the one-page document that names the programme's outcome metric, its baseline, its target, the review cadence, the cost envelope, the named owner, and - critically - the written stop condition.
What it includes
- The business outcome as a measurable metric. Cycle time, cost avoided, revenue uplift, risk reduced, capacity created - pick one.
- A baseline and a target with a dated review cadence.
- A cost envelope per bet.
- A named owner, not a committee.
- A written stop condition - the circumstances under which the programme would be rescoped or halted.
Why the stop condition matters more than it looks
A programme that cannot articulate, in writing, the circumstances under which it would be paused is not being managed - it is being hoped for. The stop condition is the single clearest signal of whether an AI programme is scoped honestly.
Who signs it
Programme sponsor and finance partner, at minimum. In regulated programmes, add a risk or compliance partner. Without those signatures, the scorecard is an internal document that will not survive contact with a budget review.
How it prevents theatrical scopes
A good scorecard prevents the vendor-led workshop pattern where the scope produces a list of "use cases" without owners, outcomes, or numbers. If the scope doesn't end in a signed scorecard, the programme will drift. See How to scope an AI deployment in two weeks.