Cohort analysis is the single most useful analytical framework for subscription businesses. It takes 20–30 minutes to set up and can reveal problems (and wins) that aggregate metrics hide.
Step 1: Define Your Cohort
The most common cohort definition: customers grouped by the month they first subscribed (or activated, or made their first purchase).
Other useful cohort definitions: - Acquisition channel (paid search, referral, organic) - Pricing plan (Starter vs Pro vs Enterprise) - Company size (1–10, 11–50, 51–200 employees) - Onboarding path (self-serve vs sales-assisted)
Step 2: Build the Retention Matrix
Create a table where: - Rows = cohort start month (Jan 2024, Feb 2024, etc.) - Columns = months since start (Month 0, Month 1, Month 2, etc.) - Cell value = % of original cohort still active
Example:
| Cohort | M0 | M1 | M2 | M3 | M6 | M12 |
|---|---|---|---|---|---|---|
| Jan 2024 | 100% | 82% | 74% | 70% | 65% | 60% |
| Feb 2024 | 100% | 85% | 78% | 73% | — | — |
The diagonal shows the current state of each cohort (Feb 2024 cohort is at M2).
Step 3: Identify the Patterns
Healthy pattern: Steep early drop, then flattening curve. Most churn in months 1–2, then survivors are sticky. The tail asymptote is your "power user" baseline.
Problematic pattern: Linear decay that never flattens. Churn continues at the same rate month after month — no core of power users.
Improving cohorts: Newer cohorts have higher retention at the same time point than older ones — your product is getting better.
Worsening cohorts: Newer cohorts churn faster — check for product regressions, channel mix change (acquiring lower-quality customers), or pricing changes.
Step 4: Act on What You Find
| Finding | Action |
|---|---|
| High month-1 churn | Fix onboarding — activation email sequence, in-app guidance |
| High month-3 churn | Improve value realization — what "aha moment" drives retention? |
| High month-12 churn | Contract renewals, QBRs, expansion offers for SMB |
| Channel X retains 2× better | Shift budget to Channel X |
Use the Churn Cohort Calculator to model your retention projections.