Retention Cohort Shock Analysis Builder

Analyze sudden retention drops or spikes in specific cohorts and isolate likely causes, impacted segments, and next investigative steps.

Prompt Template

You are a product analyst investigating an unexpected retention change. Build a cohort shock analysis from the details below.

Business type: [SaaS, app, marketplace, subscription ecommerce, etc.]
Retention metric: [day 7, week 4, month 3, renewal rate, etc.]
Affected cohorts: [signup dates, acquisition source, plan, geography, device, etc.]
Size of change: [drop or spike magnitude]
Relevant product or business changes: [pricing, onboarding, release, policy, campaign, bug, seasonality]
Available data: [cohort tables, event logs, support tickets, billing events, NPS, etc.]
Hypotheses already discussed: [list]

Provide:
1. Clear problem framing and scope
2. A ranked hypothesis list with what evidence would confirm or reject each one
3. Segment cuts to inspect first
4. Confounders and measurement traps to rule out
5. A step-by-step investigation plan
6. Possible business actions for each likely root cause
7. An executive summary in plain English

Focus on separating true behavior change from instrumentation noise.

Example Output

Problem Frame

Month-2 retention fell from 41% to 33% for self-serve US cohorts acquired in February after the new onboarding flow launched.

Highest-Priority Hypotheses

1. Onboarding flow increased early confusion for low-intent users

2. Trial-to-paid billing bug removed some users from the retained population incorrectly

3. Paid social mix shifted toward weaker-intent traffic

First Cuts to Review

- Device: mobile web vs desktop

- Acquisition source: branded search vs paid social

- Activation completion within first 24 hours

Executive Summary

The sharp drop may be partly real and partly measurement-related. Check billing-state definitions first, then compare activation rates and acquisition mix before escalating product changes.

Tips for Best Results

  • 💡Include recent launches and tracking changes, because retention shocks love wearing disguises.
  • 💡Ask for confirm-or-reject evidence next to each hypothesis so the analysis leads to action, not just theories.
  • 💡Mention the exact affected cohorts instead of saying retention fell in general.