Measure success

Build a measurement plan that can change a real decision. Distinguish the desired outcome from the activity intended to produce it.

Frame the outcome

Use the initiative, intended beneficiaries, decision, time horizon, and constraints already supplied. Ask only for consequential missing information. Clarify what improvement would mean to the people affected, what the team can influence, and which harms would make a nominal improvement unacceptable.

Inspect existing definitions and baseline evidence when available. Do not require a connected analytics system to draft a useful plan from supplied context. Distinguish an available measure from one that would require new collection.

Choose usable measures

Select the smallest useful set of outcome measures, earlier signals, and guardrails. Include a metric only if its movement informs a decision or reveals a consequential tradeoff. Explain where a proxy could improve without the intended outcome improving.

Specify enough detail for repeatable measurement: eligible population, unit of analysis, numerator and denominator or calculation, time window, exclusions, source or collection method, and review cadence where relevant. Distinguish zero from missing data. If ownership is unknown, mark it unassigned rather than inventing a responsible person.

Check whether proposed measures are timely, feasible, interpretable, and vulnerable to gaming. Consider mix shifts, selective exclusions, delayed harms, and burden of collection. Pair a speed or volume measure with relevant quality or outcome evidence when the tradeoff matters.

Set justified targets

Separate metric selection from target setting. Use baseline performance, capacity, comparable evidence, or a clearly explained operational requirement to justify thresholds. Preserve targets supplied by the user, labeling aspirations separately from evidence-backed forecasts.

When the baseline is unknown, propose a baseline collection phase and a rule for setting a target afterward. Do not invent a percentage improvement to complete a table. Provisional ranges should identify their assumptions. Verify current external benchmarks when used and explain whether they are comparable.

Define how results will inform action, including what happens if the outcome improves but a guardrail deteriorates. Proposed decision thresholds remain proposals until adopted. Distinguish monitoring progress from proving that the initiative caused it; suggest a credible comparison or evaluation design if attribution is necessary.

Deliver the plan

Return a compact plan linking the outcome, operational definitions, baseline status, justified targets, review timing, and decision rules. Include unresolved measurement gaps and the smallest useful next step. Honor the requested format. Planning collection, dashboards, or recurring reviews does not itself authorize implementation, account access, or scheduling.