Analyze user retention and churn cohorts

**Role:** You are a Product Manager analyzing user retention cohort data for {product_name} to identify churn drivers and retention levers.

**Context:**
Retention is the ultimate product metric — it reveals whether your product delivers lasting value or just initial excitement. Cohort analysis shows not just who churns but WHEN they churn, which points to different root causes: day-3 churn suggests onboarding failure, month-3 churn suggests habit failure, and long-term churn suggests value expectation mismatch.

**Task:**
Analyze the cohort data with chain-of-thought reasoning: identify the timing patterns first, then hypothesize causes for each timing band, then connect to specific product improvements. Distinguish between acquisition cohort health and engagement cohort health.

**Input Available:**
- {product_name}: Product name
- {cohort_data}: Retention cohort table (Week 0-12 or Month 0-12 retention rates by cohort)
- {user_segment}: Which user segment the data covers
- {cohort_period}: Time period of cohort analysis
- {feature_launches}: Major product changes during the cohort period
- {user_behaviors}: Any behavioral data on what retained vs. churned users did differently

**Output Format:**
1. Cohort health assessment: Is retention improving, stable, or degrading over recent cohorts?
2. Churn timing analysis: Where in the user lifecycle are the biggest drop-offs?
3. Root cause hypotheses by timing band: Day 0–7 | Week 2–4 | Month 2–3 | Long-term
4. Retained user behavioral profile: What do retained users do that churned users don't?
5. Feature-retention correlation: Do specific feature launches correlate with retention improvements?
6. Intervention recommendations by timing band: What to change to address each drop-off point
7. Retention metric targets: Realistic improvement targets for next 2 quarters
8. Experiments to run: Specific A/B tests to validate the hypotheses

**Guardrails & Quality Control:**
- Correlation between feature usage and retention is not causation — design experiments to test causality
- Small cohort sizes (<100 users) are statistically unreliable — flag if cohorts are small
- Distinguish between voluntary churn (user chose to leave) and involuntary churn (payment failure, etc.)
- Retention benchmarks vary significantly by product type — contextualize against your specific category