Cohort analysis — Comparing groups over time

Comparing groups of customers who started at the same time to see how their behaviour changes.

What it means

Cohort analysis means grouping customers by when they started, such as the week they first visited or the month they first bought, and then tracking each group over time. Instead of mixing all customers together, you compare January joiners with February joiners, and so on.

How it works

You pick two things: what defines the group, and which behaviour you want to follow. For example, group users by their first purchase month, then measure how many come back to buy again in month one, month two and month three.

The result is usually a grid: each row is a cohort, each column a time period after they started, and each cell shows how many are still active.

In GA4, you can build this with the cohort exploration in the Explore section. You choose the inclusion criteria (such as first visit or first purchase), the return criteria, and whether to group by day, week or month.

A simple example

A Kota coaching institute launches a test-series app. In April, 500 students sign up; in May, another 500. By the end of their second month, 200 of the April group are still taking tests, but 300 of the May group are. The team finds they added daily reminder messages in May. Without cohorts, this would be hidden inside one overall number.

Why it matters

Overall totals can hide problems. If you keep adding new users, total activity may look healthy even while each new group drops off faster. Cohort analysis shows whether your product and marketing are really improving, and it helps you understand retention, churn and customer lifetime value.

Beginner tips

Start simple: group by signup month and track one action, like a repeat purchase.
Compare cohorts from similar seasons, since festival months can behave differently.
Note any changes you made to each cohort's experience.
Common mistake: judging a new cohort too early, before it has had as much time as older ones.