A single average churn rate is a check-engine light — it tells you something is wrong but not what. Cohort analysis is the diagnostic that breaks the average apart into stories you can act on. Once you have it set up, it becomes the most important chart in your subscription business.
What a cohort actually is
A cohort is any group of customers who share a defining event — typically the month they signed up. So "January 2025 cohort" means everyone who became a subscriber in January 2025, tracked together as a unit. You can build cohorts on other dimensions too — acquisition channel, first product purchased, discount applied at signup — but signup-month is the foundational view.
The retention curve, explained
For each cohort, plot the percentage of customers still active in each subsequent month. The shape of the curve diagnoses your retention problem:
- Steep early drop, flat tail — Onboarding or fit problem in month 1–2; the customers who survive past month 3 stick. Common in subscription boxes where novelty signups quickly self-select out.
- Continuous decay — Steady erosion every month. Indicates a product-fit or value problem; nothing is creating loyalty.
- Late-stage drop — Customers stay for many months then leave around month 12–18. Usually fatigue or category exhaustion; consider win-back or category extension.
- Smile curve — Decay early, then a stable plateau. The healthiest pattern; your committed subscribers are clearly distinguished from your tryers.
What cohort analysis reveals that averages hide
- Whether retention is improving over time. If newer cohorts retain better than older ones at the same age, your changes are working. Average churn cannot show this because it mixes ages.
- Where the loss is happening. Month 1, month 3, or month 12 — each requires a different fix.
- Channel quality. Cohorts cut by acquisition channel reveal whether one channel produces stickier customers. Sometimes the cheap channel produces customers worth half the LTV of the expensive one.
- Promotion drag. Customers acquired on a heavy discount often churn at 2–3x the rate of organic signups — invisible in averages, obvious in cohort cuts.
How to set up cohort analysis
Most subscription platforms (Joy Subscriptions, Recharge, etc.) include cohort retention reports. For deeper analysis, export to a BI tool — Metabase, Mode, or even a well-built spreadsheet works. The key discipline is running the same cuts every month so you can compare trends. For a related analytical view, see churn rate analysis and the Shopify-specific take in customer cohort analysis Shopify.