Are they coming back?

Measuring user retention

Anuradha Sridharan
Aug 9, 2013 · 3 min read

Every product organization would love to have users coming back to them. But in this age where we talk a lot about cognitive overload, casual users and distribution / discovery challenges (especially in the case of mobile apps), retaining users seem to be a bigger challenge. I will leave it for a subsequent post to write about product strategies that might lead to increased retention or engagement.

First, I think it makes more sense to understand how to measure and monitor user retention. I would like to talk about two different quantitative techniques that will help you in this direction. This will be more useful for someone in a product management or business analyst role or just someone who loves to crunch numbers to get some meaningful insights.


When you talk about user retention, you should have a clear definition of who an active user is with respect to your product. I would like to use the following three factors to define an active user:

  • Novelty factor — When did the user register?
  • Recency factor — When did he perform a specific activity or use a particular feature?
  • Frequency factor — How frequently the user has used the feature?

For example, for a task management app like any.do, an active user is one who

  • Registered atleast a week earlier (this will prevent new users from the scope of engagement analysis)
  • Created a task in the last one week (this ensures the user is still active and not dropped off)
  • Has created atleast 20 tasks (this ensures the user is fairly active and is not someone who is just casually checking out the app)

Once we identify these active users, two metrics need to be tracked on an ongoing basis:

  • Ratio of active users to new users
  • Ratio of active users to total users

The ideal benchmark for these two ratios would depend on your product, industry and context. But instead of an absolute metric, it would be wise to track the trend of these metrics on a daily/weekly basis and ensure they are on the rise.


Another technique by which you could measure user retention is to use a “cohort analysis”. Josh Kopelman has given a useful example on how to perform cohort analysis on a month-on-month basis. To get deeper insights, it is much better to analyze retention on a weekly basis.

This gives a clear vision of what percentage of users were you able to retain the subsequent weeks after they registered. If your product features are targeted towards improving retention, you can also get a good understanding of whether these features created a positive impact over the weeks.

Apart from getting an overall analysis, it also makes sense to split the cohort analysis based on multiple factors such as

  • Demographic based (For example, retention of users aged 30-35 years are higher as compared to that of users aged 20-25 years)
  • Device specific (For example, iOS app users are much more active than Android app users)
  • Other user profile information that is relevant to your product and target segments (For example, a health and fitness app would rely on BMI of users to segment which BMI range users engage better with the app)

I’m sure there are other ways by which user retention could be measured much more deeply but I found these two techniques to be a good starting point. I would love to hear your views on how you measure and monitor user retention on an on-going basis.

    Anuradha Sridharan

    Written by

    A mom, Product Manager at HealthifyMe, IIMB Alumnus, ex-Cleartriper, ex-Yahoo, ex-Oracleite, dreamer, product & technology enthusiast