Cohort analysis

Grouping people by when they started, such as the month they subscribed, and tracking how each group behaves over the same elapsed time.

By Azande PorterUpdated 3 min read

What is cohort analysis?

Cohort analysis groups people by a shared starting point, usually the week or month they subscribed or first bought, and compares how each group behaves over the same elapsed time. It shows whether newer audiences behave differently from older ones.

A single blended number hides that. A list-wide click rate can fall because recent signups are less engaged while long-time readers are as active as ever. Cohorts separate the two.

How does a cohort table work?

Each row is a starting cohort, each column is time since joining, and each cell is the share of the cohort still doing the thing you care about. In this illustrative example, the cells show the share still subscribed:

Signup month Joined Month 1 Month 2 Month 3
January 200 92% 88% 85%
February 450 80% 71% 66%
March 220 91% 87% not yet

Read down a column to compare cohorts at the same age. February brought more than twice as many people, and by month three a third of them were gone, against 15% of January's group. If February was the month a giveaway or one widely shared post drove signups, that is a reason to look closer. It is an observation, not proof that the source caused the drop.

Read across a row to see when people leave. If most of the loss happens in month one, the welcome sequence and the first few editions are where to look.

What can cohort analysis show creators and small teams?

  • Whether a growth spike was durable. Compare cohorts from a launch, a giveaway, or newsletter recommendations with your usual months.
  • When list churn happens, and whether a change to onboarding moved it for the cohorts that joined afterward.
  • How engagement holds up. Track the share of each cohort that clicked in a given month, which tells you how many engaged subscribers each wave really added.
  • When people buy. If most first purchases happen in month two or three, that shapes your offer timing and your lifetime value estimates.
  • Which sources hold up. Cohorts split by the acquisition source your provider stored can compare channels on retention, not just on volume.

How do you run a cohort analysis without a data team?

  1. Export subscribers with their signup date, status, and any stored source from your newsletter provider.
  2. Group by signup month in a spreadsheet and count how many from each month are still subscribed or still active at each age.
  3. Define "active" once. Clicks are more reliable than opens, because some email clients register opens automatically.
  4. Compare the same elapsed time. A one-month-old cohort and a six-month-old cohort are not comparable on cumulative numbers.
  5. Watch cohort size. A cohort of 30 people swings wildly; combine small months into quarters.

The guide to identifying engaged newsletter subscribers covers the engagement side. The newsletter growth rate calculator gives you the overall trend to set the cohorts against. Retention by cohort also feeds payback math for customer acquisition cost.

Try it

Free Newsletter Growth Rate Calculator

Calculate net growth rate, gross growth rate, and churn for one newsletter period from your starting subscribers, new signups, and unsubscribes.

How Distinctful helps

Distinctful connects beehiiv, Kit, or Ghost and reports subscribers and editions beside your publishing history, so you can see which stretches of posting lined up with each wave of signups before you build a cohort table. Treat any link between a post and a cohort's later behavior as an observation to test. See newsletter analytics.

Related terms

Go deeper

Compare signup waves with what you published

Read subscriber growth beside your posting history before you build a cohort table.