What Is Cohort Analysis? How to Read It in GA4 and Use It to Improve Repeat Rate

"We keep acquiring new users, but they just won't stick around"—for site operators facing this challenge, cohort analysis, which visualizes how well users are retained, is an essential technique. This article clearly explains the basics of cohort analysis, how to read it in GA4 (Google Analytics 4), and how to use it to improve your repeat rate.

What Is Cohort Analysis?

Cohort analysis is a technique that divides users into groups based on a shared characteristic and then tracks and analyzes how those groups behave over time. The word "cohort" originally means a "group" or "band," and in marketing it refers to a set of users acquired during the same period, for example.

For instance, if you treat "users who visited for the first time in the first week of January" as a single cohort and track how many of them return in weeks 2, 3, and 4, you can grasp the retention rate and the timing at which users drop off.

Why Cohort Analysis Matters

If you only look at your total number of active users, you cannot separate whether "the numbers are growing because of new acquisition" or "existing users are being retained." Because cohort analysis lets you break down user behavior by acquisition period, it enables the following kinds of judgments.

• You can identify at which point users are most likely to drop off.

• You can compare the retention impact of individual campaigns and initiatives.

• You can determine the optimal timing to launch measures that nurture repeat users.

How to Read Cohort Analysis in GA4

GA4 provides a dedicated template called "Cohort exploration," which lets you build a report in just a few clicks from the Explore feature. Let's first check how to access it.

Steps to Open Cohort Exploration

1. Log in to GA4 and click "Explore" in the left-hand menu.

1. Open the "Template gallery."

1. Select "Cohort exploration" from the list of templates.

1. Configure each item under "Tab settings" on the left to build your report.

Understanding the Settings

In Cohort exploration, you mainly configure the following items. Understanding what each one means helps you avoid misreading the report.

• Cohort inclusion: the action that groups users. The default is "First touch (user acquisition date)."

• Return criteria: the action counted as a return visit. The default is "Any event," and you can also narrow it to a specific event such as a conversion.

• Cohort granularity: the unit used to divide groups. Choose from daily, weekly, or monthly.

• Calculation: how continuation is counted. There are three types—standard, rolling, and cumulative—and the meaning of the metric changes accordingly.

• Breakdown: adds a dimension such as device or acquisition channel to segment the cohort further.

• Value: the metric shown in the table. The default is "Active users," but you can also choose event count, revenue, and more.

Watch Out for the Three "Calculation" Types

The "Calculation" setting in particular directly affects how the numbers are interpreted, so it is important to understand the differences.

• Standard: counts users who met the criteria in each period. The figure may rise or fall depending on the period.

• Rolling: counts only users who met the criteria in every consecutive period. It necessarily decays as the weeks progress.

• Cumulative: counts users who met the criteria at least once during the measurement period. It accumulates continuously.

How to Read the Table

The Cohort exploration table is structured with rows representing "cohorts (the week or month of acquisition)" and columns representing "elapsed periods (week 0, week 1, week 2, and so on)." Week 0 is the baseline immediately after acquisition (equivalent to 100%), and the further right you go, the more it shows how much of that cohort remains. Reading vertically reveals differences in retention tendencies by acquisition period, while reading horizontally reveals how drop-off progresses over time.

Using Cohort Analysis to Improve Your Repeat Rate

Cohort analysis is meaningless if you simply look at it and stop there. Its value emerges only when you read the challenges from the numbers and connect them to action. Here are some representative ways to put it to use.

1. Identify When Drop-off Occurs and Take Action

On many sites, retention drops sharply in weeks 1 to 2 right after acquisition. If you can pinpoint the elapsed period at which drop-off accelerates, you can see concrete moves such as sending a reminder email, a push notification, or a coupon that encourages a return visit just before that point.

2. Compare Retention by Acquisition Channel

Adding an acquisition channel to the "Breakdown" lets you compare retention rates by channel. A channel with many new acquisitions but low retention may have poor cost-effectiveness. Conversely, shifting budget toward channels with high repeat rates lets you improve the efficiency of your overall acquisition cost.

3. Verify the Effect of Measures and Campaigns Over Time

By comparing the cohort acquired during the week you ran a particular measure with cohorts from weeks when you did not, you can judge whether the measure ended as a "temporary spike in acquisition" or "contributed all the way through to retention." You gain a perspective that measures effectiveness not just by short-term traffic gains but by mid- to long-term retention.

4. Analyze CVR by Repeat Timing

If you change the "Value" to a conversion-related metric, you can grasp the conversion rate by cohort and by elapsed period. If the CVR is markedly low at a specific repeat timing, you can judge that there is room to improve the user experience or content during that period.

Summary

Cohort analysis is a technique that groups users by a shared characteristic such as acquisition period and tracks their subsequent retention and drop-off over time. In GA4, using "Cohort exploration" lets you visualize retention without any specialized aggregation.

The key is to correctly configure the inclusion criteria, the return criteria, and the calculation method, and to read the table from both angles—vertically (differences by acquisition period) and horizontally (the passage of time). Then connect it to concrete actions such as identifying drop-off timing, comparing channels, verifying measures, and analyzing CVR by repeat timing. Start by opening the Cohort exploration for your own site and checking at which elapsed period your users are leaving.