LaunchRanked

GlossaryGrowth & analytics

What is cohort analysis?

Also called: cohort, cohort table, cohort report, cohort exploration

Definition

Cohort analysis groups users by a shared characteristic, usually when they started, such as sign-up week, and tracks each group's behavior over time, so you can compare how different groups retain, activate or pay.

Cohort analysis, explained

Google Analytics' help documentation defines a cohort as a group of users who share a common characteristic identified by an Analytics dimension, for example all users with the same acquisition date. Cohort analysis then follows each group separately instead of blending everyone together.

Why that matters: aggregate numbers hide change. If your total active users are flat, you can't tell whether new users are sticking better and old ones leaving, or the reverse. A cohort table can. Each row is a cohort, such as users who signed up in the week of March 3, and each column is time since starting: week 1, week 2, week 3. Cells show the share still active or paying.

Read it in two directions. Along a row, you see how one cohort decays over time; a row that flattens means those users found lasting value. Down a column, you see whether newer cohorts do better than older ones at the same age. If your week-4 retention climbs from 15% to 25% across recent cohorts, your product or onboarding changes are working.

Cohorts don't have to be by date. Group by acquisition channel, plan, country or first feature used. Comparing a launch-week cohort with a search-traffic cohort shows which channel brings users who stay, often a very different ranking from which brings the most sign-ups.

Keep cohorts large enough to mean something. With 20 users per week, one person changes a cell by five points; use monthly cohorts until volume grows.

Why it matters for founders

Cohorts show whether your product is actually getting better at keeping people, and which channels bring users who stay. Blended totals can't tell you either.

Example

A startup compares cohorts by channel. Users from a launch platform sign up in large numbers but only 10% remain after a month; users from comparison pages retain at 35%. It shifts effort toward comparison content.

Common mistakes

  • Judging retention from total active users.
  • Reading tiny cohorts as meaningful.
  • Only grouping by date and never by channel or plan.

Sources

Checked

← All 274 glossary terms