Performance Management and Program Evaluation
This domain tests the chain from strategy to measures to review, and the evaluation designs used to check whether a programme actually worked.
Core rules and concepts
- The hierarchy runs strategic plan, objectives, measures, targets, and recurring review. Published plans and reports are commitments, not aspirations.
- A logic model links inputs, activities, outputs, outcomes, and impact. Outputs count work done; outcomes count change produced.
- Choose a small balanced set spanning output, outcome, efficiency, quality, and customer experience. Single-metric steering invites distortion.
- Targets create incentives to game: ratchet effects, threshold effects, cream-skimming, and definition drift. Audit the measure, not only the reported number.
- Distinguish process evaluation (was it implemented as designed?) from impact evaluation (did it cause the change?). Only the latter needs a counterfactual.
- Review cadence matters: quarterly priority reviews surface delivery risk earlier than annual reports.
Worked example
A training programme reports "500 staff trained" as its headline measure. Reasoning: attendance is an output and says nothing about change. The outcome might be error rates or time-to-decision for trained staff; the impact requires a comparison. A defensible measure is the post-training error rate for trained units against matched untrained units, paired with cost per error avoided - which also guards against training people who never use the skill.
Traps that cost marks
Reporting activity as achievement; redefining a measure mid-series and breaking comparability; rewarding the number rather than the result; skipping process evaluation and then blaming the programme theory when nothing changes.