What is cohort retention?
How does cohort retention work?
You group customers by when they joined, then track each group forward in time. The March cohort might have 100 customers in its first month. If 85 are still active a month later, month-one retention for that cohort is 85 ÷ 100 = 0.85, or 85 percent. A month after that, 78 remain, so month-two retention is 78 percent. Repeating this for every cohort produces a grid, often shown as a set of curves, that reveals how each starting group holds up over time.
What does a retention curve tell you?
The shape matters more than any single point. A curve that falls sharply in the first month and then levels off says you lose people early but keep a committed core afterwards. A curve that declines steadily every month says customers drift away throughout their life, which is harder to fix with onboarding alone. Watching whether newer cohorts hold a higher line than older ones tells you if changes to the product are actually improving retention.
How does this relate to churn?
Cohort retention and churn rate are two views of the same behaviour. Churn compresses a period's losses into one number; cohort retention spreads those losses across time and across groups so you can see where they come from. A stable overall churn figure can still hide the fact that recent cohorts are leaving faster than earlier ones, and only the cohort view makes that visible.
Reading cohorts from an export
Cohort retention is usually built from an export that records each customer's join date and their activity or subscription status over time, such as a CSV from Stripe or Shopify. Assembling the grid by hand is slow and error-prone. Instead you can open the file and ask for retention by signup month in plain English. Paperswift builds the cohorts in your browser, and only your column names and their types are sent when you ask, never the values in your rows.
Frequently asked questions
How is cohort retention different from churn rate?+
Churn rate gives one number for a period across all customers. Cohort retention breaks customers into groups by join date and tracks each group separately over time. This reveals whether losses cluster in the first month or continue for years, a pattern a single blended churn figure hides. The two metrics complement rather than replace each other.
What is a cohort?+
A cohort is a set of customers grouped by something they share, most often the month they signed up. The January cohort is everyone who joined in January. Following that group forward shows how many remain active after one month, three months, and so on, without mixing them with customers who joined later.
How do I read a retention curve?+
A retention curve plots the share of a cohort still active against the months since they joined. A curve that drops steeply then flattens means early losses but a loyal core. A curve that keeps sliding downward means customers leave steadily over their whole life. Comparing curves across cohorts shows whether retention is improving.
Why track cohorts instead of one retention number?+
A single blended figure mixes long-standing customers with brand-new ones, which can mask a real problem. If recent cohorts leave faster than older ones did, overall retention may still look stable for a while. Splitting customers by join date surfaces that decline early, when there is still time to respond to it.
Bring your own export and see it in seconds.
Free to start · your data stays in your browser.
Start free todayLast updated · by Stefan