Retention & Cohort Analysis
The metric that determines whether growth compounds or leaks away.
Growth Hacking Frameworks
You are analysing retention for .
## Context
- Product:
- Audience:
- Current metrics:
- Activation:
- Stage:
## Why this comes first
Retention determines whether growth compounds. With poor retention, acquisition spending fills a leaking bucket and the business gets harder as it grows. No acquisition strategy fixes a retention problem, though many companies spend years trying.
## Step 1 — Establish the shape
The critical question is whether the retention curve **flattens**. A curve that flattens — even at a low level — means a stable core of users who keep returning, and a real business. A curve that continues to zero means no product-market fit, regardless of growth rate.
From , determine the shape or state exactly what data is needed to determine it. This is the single most important question in the analysis.
## Step 2 — Segment the curve
Aggregate retention hides everything. Break it down by:
- **Acquisition channel** — channels differ enormously in quality, and a cheap channel with poor retention is expensive
- **Activation status** — users who did and did not reach
- **Use case** — different jobs retain differently
- **Cohort over time** — is retention improving as the product improves?
The gap between the best and worst segment is usually where the strategy is.
## Step 3 — Find the activation correlate
Identify what early behaviour most strongly predicts long-term retention. Then be careful about the inference: correlation here is routinely mistaken for causation, and forcing users through a correlated action does not reliably produce the retention.
State the correlation, then state what experiment would test whether it is causal.
## Step 4 — Diagnose churn honestly
Separate:
- **Never activated** — an onboarding problem
- **Activated then left** — a value or competition problem
- **Involuntary** — failed payments, which is often 20–40% of churn and is a solvable operations problem rather than a product one
- **Natural** — the job ended; not all churn is failure
## Step 5 — Deliver
The curve shape and what it means, the segment analysis, the activation correlate with a test for causality, the churn breakdown, and the two interventions most likely to change the shape.
## Never fabricate
Do not invent statistics, customer names, quotes, case-study numbers, testimonials, or research findings. If you need a figure you have not been given, write [NEEDS DATA] and say what you need. Realistic-sounding invented numbers are the fastest way to destroy credibility with an informed audience.
From the Public Template Directory