Resources | Tutorials

Action Cohorts

Appsee's action cohorts feature enables you to understand how often your users return to the app and perform certain actions in a given time frame. The information is presented through cohort analysis, which aggregates groups of users based on when they performed certain actions in different time frames. Cohorts can help you understand, for example, how many new users who registered came back the next day to purchase.

Action cohort reports can help you understand if your app meets your users' needs and expectations and enables you to measure how your app optimization efforts impact action retention.

1. Navigate to the 'Actions Cohorts' section in the Appsee dashboard and click 'Create Your First Cohort'.

2. Enter the name of the relevant process you'd like to measure.
For example: 'In-App Purchase', then select 'Device Type' (optional), the time frame which you'd like to include data from, and then click 'Next'.

3. Select the first action that will define the cohort. Users that performed this action will be grouped in the rows of the cohort.
Each step can be either a 'Screen View', a 'Screen Action', a custom event, or a session start.

4. Select the second action that will define the cohort. You will then see how many of the users that performed the first action, came back and performed the second action.


How is action retention calculated?

1. Our retention cohort analysis groups the number of users that performed the first action of the cohort (and not the total number of app users). The cohort will include every user that used the app and performed the first action within the defined time frame and selected date bucket (daily/weekly/monthly).


2. In this case we're looking at weekly cohorts, as shown above, the numbers 1,2,3 along the top of the retention report will indicate weekly buckets. The percentage under each weekly bucket represents the number of users who returned to use the app and performed the second action during that specific bucket.

3. The beginning and ending of each bucket will be different for each new user in the cohort. The bucket marked "1" indicates a timeframe of 7 days following the initial session of the user. For example, if the initial session occurred at 10am on Sunday, then bucket "1" for him/her begins 10am Sunday of the following week. Bucket 1 for this specific user extends between 7 and 13 days from his/her initial session.

4. A user is only be counted once per cohort, but can be included in more than one cohort: For example, if the weekly cohort is based on a "login" event, a customer who logged in at least one item each week will be in every cohort, and not only the cohort for their first purchase.

5. Users may "complete" a cohort within 30 days (when watching daily cohorts), 20 weeks (weekly cohort) or 12 months (monthly cohort).

6. When a user performs the second action, it is counted for every first action that was performed in the timeframe, not just the latest one.

7. Each cohort cell is unique: if a user performs the second action twice in the same bucket, it will only be counted once.

8. When creating a cohort in which both steps are similar (for example: "Log In"), every action will act both as a first and second action. If a user logged in 3 times: on Sunday, Monday, and Tuesday - the action performed on Monday will be counted as the second action for the one on Sunday, but will also open a new cohort that the Tuesday action will complete.

9. When filtering by versions, we will show only the users that performed their first action in the selected version.