User retention: yes, but which ones?

App2Top.ru together with Apptractor.ru He continues to publish the "Freemium Code", a series of articles dedicated to this business model. Today, an article by Thomas Sommers, marketing manager at AppLift, is being published within its framework.

User retention: yes, but which ones?

Of all the components of the user lifecycle, retention is arguably one of the most important metrics for freemium and free—to—play games. Retention directly affects the calculation of the LTV of acquired users. For example, the retention metric in the early stages (for example, on the Third Day) can serve as an indicator for evaluating the quality of traffic from user acquisition campaigns. Retention is also a reliable indicator of the quality of the application and the experience that the user receives in it.

But there is a problem. And it lies in the fact that there is still no consensus on what retention is/ how to count it.

There are several ways to calculate this metric, and each gives its own results. It often happens that people start comparing apples with oranges.

The purpose of this article is not to state "such and such a model is correct", but to present a list of the main (most popular) methods for calculating retention, as well as to look at their pros and cons. None of the proposed methods is better than the other in terms of absolute numbers, it's just that each of them is better suited for its own analytical tasks.

We compared five different types of retention. Of these, three are the most used, two are less common, which, however, does not make them worse. Please note that the names we have given to the retention types are not official constants, but describe them best.

As a rule, all retention metrics can be calculated as follows:

For each method, we took a 28-day hold as an example.

Here is the legend for the charts:

Basic types of user retention:

Full retention

Answers the question: what proportion of users return to the application every day before D+N (where D is the day and N is the selected day).

Where and for what reason it is used: full retention is a very limited tool and not widely used, but it nevertheless gives an idea of the level of user engagement in the application.

Example 28 days: Only those users who return to the application every day from the first to the twenty-eighth day are considered as retained.

2.       Classic retention

Answers the question: what proportion of users return to the app on Day+N.

Where and for what reason it is used: classical retention is most easily calculated using analytical tools. It is also the most common. It gives an idea of the total retention level of the application.

Example of day 28: Only those users who returned to the app on day 28 are considered retained. It doesn't matter how often they returned to the project before that day.

Rotating retention

Answers the question: what proportion of users return to the app on D+N and any day after.

Where and for what reason it is used: a rotating hold, for example, uses Flurry as a base hold. It gives publishers an idea of the churn rate of users in the application (those users who have not yet stopped using it), since it is usually equal to 100% minus the churn rate.

Example of the 28th day: users who return on the 28th day and any other day after (for example, on the 50th) are considered retained.

Additional types of retention

Return retention

Answers the question: what proportion of users return to the app at least once over the course of N days.

Where and for what reason it is used: chargeback is often used in gambling, because it gives an idea of how many people do not stop using the application after the first launch, which, among other things, allows you to retarget other users.

Example of day 28: All users who returned at least once before day 28 are considered retained.

5.       Range retention (Bracket-dependent return retention)

As it is considered: This is a special, limited-capacity version of the return hold. Day M is defined as the beginning of the segment up to Day N. Range retention measures the proportion of users who return at least once during the period from Day M to Day N. Users who return at least once during the period M-N are considered retained.

Where and for what reason it is used: this method is useful for understanding the behavior (and types of behavior) of users in the application.

Example of day 28: users who returned on the tenth day will be considered retained, but on the fifth or any other Day outside the framework of Day 7 and Day 28 they will not be.

We hope that we have given a comprehensive description of the basic ways to collect user retention in a mobile application or game. Which method is best suited for solving marketing tasks should be decided by the developers or publishers themselves. And remember: retention depends very much on the type and category of the app or game.

The article was published as part of a joint series of materials "The Freemium Code" of mobile development sites App2Top.ru and Apptractor.ru .

Original material: http://www.applift.com/blog/user-retention.html