Playbook
The Product Analytics Playbook
How PMs actually use product analytics — instrumenting, monitoring, diagnosing regressions, and running experiments.
Instrumentation Checklist
Events every product should have
- user_created + user_activated
- session_started + session_ended
- key_action_completed (per your core loop)
- purchase_started + purchase_completed
- error_seen + error_recovered
Dashboards Every PM Should Own
Core set
- Top-line KPI: your team's primary metric
- Funnel: acquisition → activation → retention → revenue
- Cohort retention: D1/D7/D30 by signup week
- Segmentation: usage by user tier / geography / device
Diagnosing Regressions
5-step method
- 1. Confirm the drop is real (not measurement bug)
- 2. Segment by geo/device/cohort — is drop everywhere or specific?
- 3. Check for internal changes (releases, config, experiments)
- 4. Check for external factors (holidays, competitor, PR)
- 5. Form hypothesis + run isolated test
Experimentation Basics
Rules to live by
- Compute sample size before launching (not after)
- Set primary metric and stop-loss metrics
- Run for at least one full weekly cycle
- Beware novelty effects — measure after 2 weeks too
- Always analyze by segment, not just aggregate
Master These with a Coach
The cohort drills every framework and template with live feedback from FAANG PMs.