Run a retention analysis and diagnose the leak

Description

Retention graphs are red and nobody knows where to start. This runs a structured retention analysis — cohort segmentation, event funnel, feature usage correlation — so you find the 1-2 leak points that matter most instead of debating 10 interventions at once.

You are a retention analyst diagnosing {{product_name}}'s retention. Target retention bar: {{retention_target}}.

## Step 1 — Define the retention metric
- Unit: daily, weekly, monthly active?
- Event: logged in, performed core action, completed outcome?
- Cohort boundary: signup date, first-use date, purchase date?

## Step 2 — Segment the leak
Three cuts:
1. By cohort (signup month → seasonality or onboarding change?)
2. By acquisition channel (paid vs. organic — channel quality?)
3. By first-week behavior (activated vs. not — onboarding flow?)

## Step 3 — Funnel analysis
For the day 1 → day 7 drop-off:
- Event 1: {{first_core_event}}
- Event 2: {{second_core_event}}
- Drop-off rate per step

The biggest % drop is the leak candidate.

## Step 4 — Hypothesis generation
For the leak candidate:
- Why might users drop here? (value prop unclear, friction, misaligned expectation)
- What evidence would confirm each hypothesis?
- What intervention could move the metric?

## Step 5 — Intervention priorities
Rank interventions by:
- Expected lift (rough)
- Time to ship
- Evidence required to validate
- Reversibility

## Output
1. Retention metric definition
2. Top 2 cohort patterns
3. Biggest funnel drop-off with hypotheses
4. Top 3 interventions with expected lift

Customize This Prompt