User Retention Measurement Using Cohorts
Connect With Us

User Retention Measurement Using Cohorts

March 23rd, 2013, By

Why Retention is Important

Retention is a measure of the percentage of users that remain active from one month to the next (i.e. monthly retention).

As owners and marketers of web services and mobile apps, we are usually focused on two major metrics:

  1. Acquisition – bringing in new leads / registrations/ downloads
  2. Activation – convincing new visitors to go premium, buy, etc.

However, retention is an issue often overlooked.

In order to grasp the importance of retention, consider the fact that if your monthly churn (users lost over time) is higher than your monthly user acquisition, then your active user base will decrease rather than increase over time.


How to Measure User Retention

Retention is typically measured using a method called cohort analysis. A cohort is a group of users that started using a service or app at a specified date range. When we collect and keep our users’ cohort, we can then measure their retention over time.

Few tools include retention measurement out of the box (e.g. Mixpanel). Measuring retention in Google Analytics requires some work. Here is a general explanation of how it can be done:


Data Collection

For each user’s initial interaction (e.g. first use of mobile app), keep their start-date in a user-scope custom dimension (for mobile apps) or custom variable (for websites) as a “yyyymmdd” formatted string.


Looking at Retention Data

After enough data has been collected, we can use it to see the rate of user retention (or churn) from month to month.

We’ll start with the first full calendric month that cohort data was collected – for instance August 2012. For this, we’ll create an Advanced Segment that includes only visits where the custom variable containing the user’s start date begins with “201208”.

We’ll set our graph resolution to monthly, and measure the Unique Visitors metric from month to month.


User Retention

We repeat this for each subsequent cohort, and with a little Excel work, we chart for each cohort (vertical axis) how many users are still active month over month (horizontal):




Taking the above chart as an example, we see:

  • For the August cohort, 51% of the users were retained over to the next month, 40% were retained over to the 3rd month, and so on.
  • Retention ratios have improved gradually from cohort to cohort, probably due to product improvements.
  • However, we see low retention rates for the December cohort – perhaps the user acquisition strategy that was used in December brought users of lower level of intent for using our solution.

Evidently, retention measurement requires some implementation and Excel work, but it’s a must for any website or app that intends to keep visitors engaged over time.