Measurement Plan: Checklist to keep metrics useful

Use this checklist to ensure your metrics are reliable, actionable, and tied to decisions.

1. Define outcomes vs outputs. Write the outcome you care about (customer retention, uptime, throughput) and list supporting output measures.
2. Choose 1–3 primary metrics. Primary metrics should reflect the decision's ultimate goal—not internal activity.
3. Document baselines and targets. Record current values and what change would count as meaningful.
4. Identify data sources and ownership. For each metric list the source system and a named owner responsible for correctness.
5. Specify calculation logic. Write the exact SQL or formula used to compute the metric (including filtering rules).
6. Set measurement cadence and reporting channels. How often will metrics be updated, who receives them, and where are they displayed?
7. Data quality checks. Define simple sanity checks (ranges, sudden jumps, missing data) and who investigates anomalies.
8. Alert thresholds & action plan. For critical metrics, set thresholds that trigger an assigned action or huddle.
9. Link metrics to decisions. For each primary metric, note which decisions it informs and where the decision is recorded (Decision Brief ID).
10. Schedule reviews. Set review cadences—ad-hoc for experiments and regular for operational KPIs. Consider using the KPI Huddle template for recurring reviews.

Example: "Primary metric = weekly active users (WAU); baseline = 12,000; target = +8% in 6 weeks; source = product events DB; owner = analytics lead; calculation = unique user IDs with >=1 event per week; cadence = weekly; alert if drop >10% week-over-week."