Design a behavioral cohort diagnosis

Description

Retention curves look flat aggregate and your team has no idea why. This runs a cohort analysis across acquisition channel, feature usage, and behavior pattern so you can identify the 2-3 cohorts that actually drive your numbers — and the ones leaking out.

You are a product analyst diagnosing retention on {{product_name}}. Aggregate retention curve: {{aggregate_curve}}.

## Cohort dimensions
Slice the base into cohorts on:
1. Acquisition channel (organic, paid, referral, etc.)
2. First week behavior (activated core feature vs. not)
3. User segment (role, company size, use case)
4. Signup month (time cohorts to separate seasonality)

## For each cohort
- 7-day retention
- 28-day retention
- 90-day retention
- Feature usage distribution
- Cohort size

## Pattern surfacing
1. Which cohorts have outlier high retention (and why)?
2. Which cohorts have outlier low retention (and why)?
3. What share of total retained users comes from the top cohort?
4. If we only kept the top 3 cohorts, what would our aggregate look like?

## Intervention options
- Acquisition targeting change (more of the winning cohort)
- Activation redesign (fix the onboarding path for leaky cohorts)
- Product scoping (serve a different segment decisively)

## Output
1. Cohort retention table
2. 2-3 insights with evidence
3. The one cohort with highest growth potential
4. The one cohort we might explicitly stop acquiring

Customize This Prompt