Cohort Analysis: The “What” and “How” to Improving User Retention

September 6, 2022

This article was initially published to LinkedIn.

Acquisition and retention: two vastly different pieces of the same marketing pie.

Digital products live or die by their ability to retain users. For product marketers, creating and curating a strong user experience – both within and outside of a platform – is the key to driving retention, reducing churn, and improving conversions that drive long-term revenue.

There are many ways product teams measure user retention. One common approach to understanding retention is through what’s called a cohort analysis – a high-level look into how users engage with a particular product and that product’s features, over time.

Cohort analysis helps product marketers understand their current user engagement, and identify the area(s) where the product can be improved to foster deeper engagement and reduce customer churn.

My goal in writing this introductory guide was to present cohort analysis in the most digestible way possible. I wanted this article to be something that would’ve helped me out when I first learned about cohort analysis, and I genuinely hope you will find it helpful, too.

What is a cohort analysis?

Let’s start with that first word, “cohort.” Cohorts are simply groups of users who share a single commonality. One example of a cohort is “all new users who signed up in month X,” and another example could be “all users who made a purchase.”

In marketing, a cohort analysis is the measurement of a specific trend or user behavior over an indefinite period of time. Cohorts can be created daily, weekly or monthly (whichever cadence is most appropriate for your product) to illustrate changes throughout a typical user’s lifecycle:

An example of a month-by-month cohort analysis table
A cohort analysis of New User retention for a paid subscription service. The “Monthly Cohort” column indicates the month the New User signed up.

In the above table, each month’s cohort consists of users who signed up during that month. For their entire user lifespan with the product, they will always be members of this specific cohort. Typically, the percentage of active users within that cohort will decrease over time (eg, users are inactive or churn completely). For any product, the goal is to eventually reach a point where the percentage “plateaus” out (meaning, strong retention!).

According to both Amplitude and CleverTap, there are two basic types of cohorts that product marketers use:

How to Read a Cohort Analysis

The rows in a cohort analysis represent a group of customers or users (aka, a “cohort”). All users within a single cohort share one commonality, like the month of their sign-up or month of first purchase:

No alt text provided for this image
The December New User cohort is highlighted above.

The columns in a cohort analysis represent how much time has passed since the cohort’s creation. This period can be a week, a month, a quarter or even a year:

No alt text provided for this image
Months 1 & 2 of each New User cohort are highlighted above.

On the table, each cell where a row and column meets displays the % of active users within that cohort – it decreases over time, as an increasing number of users will likely display no activity. For marketers, your tactics and strategies should aim to slow or plateau this drop-off, and improve long-term retention.

Sample Scenarios: What a Cohort Analysis Can Tell You

In learning about cohort analysis, the topic fully “clicked” for me once I saw it in use. Below, I’ve drafted a few product-specific instances where a cohort analysis would be useful for measuring retention and analyzing user insights:

Q: “Did the updates to our product’s welcome tutorial help us improve user retention?”

User onboarding tutorials are common in many successful online platforms. Their purpose is to acquaint new users with the most valuable features and functions within the software, so users can start leveraging the platform’s benefits immediately (thus, become less likely to churn).

Let’s say the product marketing team at Company X has noticed a steep user drop-off in their new user cohorts after month 1. To reduce churn and improve user retention, the product marketing team revamps their platform’s build-in user onboarding tutorial with more personalization and greater relevance to the user's unique persona:

No alt text provided for this image

With the introduction of this more-personalized onboarding tutorial, we can see that user retention improved 10-20% by months 1-3 (quite significant!). By month 4 onwards, the cohorts of March, April and May saw a significant percentage increase in monthly active users – in other words, more users are successfully finding long-term value in Company X’s platform!

This is one case where marketers can apply data to inform retention tactics. Perhaps, in addition to this new welcome tutorial, this strategy was supplemented by refreshed customer-facing resources and other content. Strong retention is marked by the value a user extracts from your product, always – a cohort analysis presents the data that backs this up.

Of course, any thorough retention analysis needs to be vetted in context; is there enough evidence that this tutorial alone is the key driver for improved retention / reduced churn, or are there any other factors in play (other features, seasonality, marketing campaign, etc)? This is not only true of cohort analysis, but almost all measurement techniques, as critical thinking puts data points into context.

Q: “How can we improve our DAU (daily active users) on our mobile game app?”

Earlier, we established that behavioral cohorts are used to measure *any* possible decision a user will make— as small as a button tap. As marketers, this behavioral data can assist us in determining when the best time is to re-engage and re-market to users, based on actions taken or not taken.

Let’s say your product is a mobile game app, in which players participate in short online matches. You want to observe two cohorts: 1) total users who started at least one game, and 2) total users who have opened the app but not started a game – both on a daily cadence:

No alt text provided for this image
No alt text provided for this image

It’s clear that users who have played at least one game are “sticky;” there is a sizable amount which open the app the next day, with this percentage peaking on the weekends.

It’s also clear that users who open the app but do not start a game are very quick to fall un-engaged. If on “Day 0” (when they first open the app and become part of the cohort) they don’t play, the second table above shows that it’s extremely unlikely the user will return to the app organically.

As a marketer, what tactics can be deployed to drive more activity and engagement among the second cohort? Whichever tactic(s) we choose, they have to be quick; as in, 24 hours since the first engagement with the app.

One option for inactive users may be implementing a re-marketing “drip” sequence, perhaps through push notifications or emails. These communications might invite players to try out game modes they haven’t tried before, or let them know just how close they are to a goal.


A cohort analysis can help marketers and product teams optimize retention through user experience and messaging. The method is most effective in helping marketers identify specific drop-off points for users (eg, inactivity or churn), and allowing marketing teams to implement strategies to drive stronger retention.

Additionally, there are much more complex use cases for a cohort analysis. I’ve linked some resources which I found helpful below, if you wish to read further or get another perspective on cohort analysis, and how it can help your business:

Peel: “What is a Cohort? How to Read a Cohort Analysis Chart…”

Userpilot: Cohort Analysis vs. Segmentation: What’s the Difference and How To Use Them To Drive Retention?

Adjust: “The definition of a cohort”

© 2023 Alex Bialek and 1FiveFour Creative LLC. All right reserved.