A single churn rate is a check-engine light. It tells you something is wrong but not what. Churn rate analysis is the diagnostic phase — splitting that one number into the dimensions that reveal the underlying cause so you can act on it.
The dimensions that matter
- By tenure cohort — Are you losing customers in month 1, month 3, or month 12? Different causes, different fixes.
- By acquisition channel — Customers from one paid channel may churn at 2x the rate of organic. That changes the unit economics of the channel entirely.
- By plan or product — A high-churn plan may be a pricing mismatch, a fit mismatch, or just the wrong default.
- By voluntary vs. involuntary — Failed payments need a dunning fix; voluntary cancellations need a product or value fix.
- By cancel reason — Captured from the cancel flow survey. The single most actionable dimension you can collect.
The cohort retention curve
The most important chart in churn analysis is the cohort retention curve — for each signup month, the percentage of customers still active in each subsequent month. Healthy subscriptions show a steep early drop (month 1–3) that flattens into a long stable tail. Unhealthy ones drop continuously. The shape of the curve tells you exactly where to invest: a sharp early drop means fix onboarding; continuous decay means fix product fit.
Common analytical mistakes
- Averaging across cohorts. Mixing month-1 customers with month-12 customers hides the early-churn problem.
- Including new customers in the denominator. Skews the rate downward and makes month-over-month comparisons meaningless.
- Ignoring seasonality. Holiday signups churn faster than year-round signups. Always compare like cohorts.
- Treating revenue churn and customer churn as interchangeable. They diverge when you have tiered pricing, and the gap is informative.
For a deeper view of the underlying metrics, see how to calculate churn.