Feature Spotlight: Localytics Event-Based Cohort Retention

At Localytics we’re constantly working to make it easier for you to identify and retain high value app users. A great example of this is event-based cohort retention, a feature that enables you to measure retention based on engagement with a particular event.

It’s a highly useful tool for any app owner who wants to understand user retention based on their in-app actions. This insight will help you make better decisions around the types of app updates to make to promote user engagement.

In this post, we’ll dive into event-based cohort retention: what it is, how it works, and how to use it as part of your broader app strategy.


Understanding Event-Based Cohort Retention

Simply defined, event-based cohort retention is a measure of retention for engagement for a specific event, i.e. any notable user action in your app. Cohorts are groups of users who did a specific event in a specific time frame: day, week, or month. Event-based cohort retention shows the percentage of users from each cohort who then went on to do the event in each subsequent day, week, or month.

In Localytics, you can take your analysis a step further by defining a cohort based on one event, and measuring retention for another event. This enables you to view the percentage of users who progressed from one event to a subsequent event. For example, you could create cohorts based on a “message viewed” event, and then measure the percentage of those users who completed an “in-app purchase” event in each subsequent time period.


Event-Based vs. Session-Based Cohorts

Localytics tracks two types of cohort retention: session-based and event-based.

Session-based weekly cohort analysis:

Screen Shot 2014-06-02 at 9.41.19 AM


Event-based weekly cohort analysis:

Screen Shot 2014-06-01 at 3.35.05 PM

Session-based cohorts give a high-level overview of app user retention based on app opens, while event cohorts allow you to take a more granular look at which specific events continue to engage users over time.

These two retention analyses can offer complementary insights. For example, if session-based retention is declining, delving into event-based cohorts can provide insight into the specific areas where usage is dropping off in your app. In a media app, for example, you may see that retention for the “article viewed” event has fallen, while retention for the “video played” event has been improving. You could try serving up more video content as a way to keep users coming back to the app, boosting session-based retention.


Using Event-Based Cohorts in Localytics

In the Cohorts dashboard, you can select event-based cohorts and then choose two events. You can also select attributes, i.e. details that describe your event, for each of the events.

Screen Shot 2014-06-01 at 3.34.10 PM

The first event you select indicates the cohort: the group of unique users who did that event for the first time (during the selected time period) in a given day, week, or month. The second event, or return event, indicates the retention measure: the percentage of users from each cohort who did the second event in each subsequent day, week, or month. You can choose the same event twice, or choose two different events.

Here are the three primary uses of this feature:

1. Comparison of event cohorts

You can easily compare multiple cohort indicators (first events) against its corresponding conversion event (second event) in Localytics. For example, if you wanted to understand whether users who read an article were more likely to buy a subscription vs. users who watched a video. To compare, you would set up a report using the first event “article read” and the second event “subscribed.” You would then save this report to your dashboard in Localytics. Then, you would set up a second report using the first event “video play” and the second event “subscribed.” You would also save this report to your dashboard, where you could view the two reports side-by-side.

2. Single event cohort retention

Choosing the same event twice will show retention for a single event, independent of other events in the app. It’s a useful measure for events that a user might do on regular basis, such as article viewed, photo taken, or add to cart. Declining retention for a key event in your app could be a pre-indicator of app user churn and an opportunity to re-engage those users.

3. Multiple event cohort retention

Choosing two different events will show the percentage of users in a cohort who did the first event then went on to do the second event. It’s useful if you want to identify possible correlations between events and compare the behavior of users who did one event vs. another. For example, you could see the percentage of each user cohort who registered through Facebook and then went on to make an in-app purchase in each week after registration. You would not only be able to see whether users who registered through Facebook event were likely to make a purchase, but which days/weeks/months after registration they were most likely to purchase.


Leveraging Event-Based Cohorts in Your App Strategy

Event-based cohorts provide a new level of insight into user actions within the app and the connections between those actions. By understanding which events are more and less likely to lead to a desired conversion event, you can take action to influence desired outcomes.

We already discussed how session-based retention and event-based retention can be used in tandem to understand how engagement with specific features and events is impacting your overall user retention – valuable input into your product strategy.

Lastly, I’d like to dive into how event-based cohorts help complete the picture with another feature that measures the relationship between events: funnels. Funnels enable you to track event-based processes in the app, tracking the percentage of users who complete each step in the funnel through to the conversion event. This helps you identify key-drop off points in the conversion process.

By using event based cohorts in addition to funnels you can see how long (in days, weeks, or months), after completing the funnel start event, users are most likely to complete the conversion event, and whether users who started the funnel at different times are more or less likely to convert. With this information, you can time your marketing campaigns more intelligently, using push and in-app messages to nudge users towards conversion at the times they’re most likely to convert.

So, if you’re seeking more granular insight into user behavior and actions in the app, event-based cohort retention can definitely help complete the picture.


New Call-to-action