Cohort analysis helps you understand how groups of new customers behave over time—specifically, how often they return and how much revenue they generate. This article walks you through how to:

A cohort is a group of customers who share a common trait—typically, the month they made their first purchase. Cohort analysis helps you track and compare the behavior of these groups over time.

This allows you to answer key questions such as:

You’ll find cohort tables and graphs in the Retention tab. These tools help you analyze repeat purchase behavior and identify trends across different customer groups.

🎥 Watch the video below for a walkthrough of the Retention tab and cohort tables.

The Retention cohort table uses a clean, single-metric layout. Each table shows one metric at a time, and every cohort has its own Cohort size and First order columns, followed by Month 0 through Month N—so acquisition and each following period read at a glance.

Select the metric you want from the metric selector and the table redraws for it. The table defaults to Retention Rate, and you can switch between:

Because each metric represents a fixed view, there are no separate cumulative/non-cumulative or absolute/relative toggles. For example, select Retention Rate for a cumulative, relative view of retention, or Customers for period-by-period counts.

Use the period basis control to choose how columns are defined:

When a CAC-based metric such as LTV:CAC is selected, a live indicator tells you whether the CAC being used is blended (across all channels) or scoped to the channel or campaign you've filtered by—so you always know which acquisition cost the ratio reflects.

This shows how a single cohort behaves over time.
​

For example:

When viewing a period-by-period metric such as Customers, each column reflects only purchases made in that specific period. If a customer buys twice (e.g. in November and again in January), they're counted in both periods separately.

This shows how all cohorts behave at a specific time after their first purchase.
​

For example:

Looking diagonally shows trends during specific periods like holidays or promotional campaigns.

For example:

You can take your cohort analysis further by applying filters and customizing how data is displayed.

You can break down your cohorts by:

These filters help answer questions like:

The Cohort Evolution Graph shows how the cumulative lifetime value (LTV) grows over time for a specific customer cohort.

Polar will automatically pre-select the top 3 performing segments to help populate the chart meaningfully. These top segments are typically:

The customer count stays the same across time—only the LTV value increases as more purchases occur.

How cohort periods are defined depends on the period basis you choose:

The rolling-window method reflects customer behavior based on actual timing from first purchase rather than calendar boundaries.