A cohort is a group of people that have common characteristics and are considered as a group.
AppsFlyer's Cohort Report provides advertisers with the ability to view and compare different metrics per multiple cohorts over different time frames. For example, an advertiser can create different cohorts according to different GEOs and compare the average number of sessions per user for each cohort over the first 30 days of each user's activity.
The Cohort Report is comprised of 3 layers:
- Defining the cohort
- Selecting the metric to be analyzed
- Selecting the time frame to be looked at
Define the Cohorts
1. Define the cohort groups you would like to view:
1.1 Select the date range in which the users first launched the app.
1.2 Apply additional filters to divide these users to different cohorts by clicking Filter, defining the filters and groups-by (up to 4 parameters are allowed) and then clicking Apply
Select the Metric You Want to Analyze
2. Click the Metric dropdown list to select the metric you would like to view per the different cohorts.
The available metrics are:
- Average sessions per user
- Average revenue per user
- Average # of revenue events per user
- Any in-app event reported by the advertiser (average number of occurrences per user for each event)
- You can also select "All" to view a snapshot of all metrics per cohort
Select the Time Frame You Want to Look At
1. The different metrics are calculated per different time frames, which represent the average cumulative activities in the first X days after install per user. The provided time frames are 1, 3, 5, 7, 14, 30, 60 and 90 days. Example: cohort day 3 of network A shows 10 sessions, meaning the average user attributed to network A performs 10 sessions in the first 4 days after installation (install day, day 1,2 and 3).
AppsFlyer considers the counting of a day based on the UTC calendar date and not based on the number of hours since install. For example, if an install occurs on Feb 1st at 22:44 UTC and a purchase takes place on Feb 5th at 01:12 UTC - the purchase is shown on Day 4 even though it occurred 3 days after the install.
The generated report provides a graph and a data table. The example below is for users from Germany and the UK, grouped by campaign and by the attributed media sources:
Note that the table can be sorted by any timeframe (columns), in this example it is sorted according to the First 14 days timeframe.
1. The filter values are not dependant. Example: Even when selecting Facebook as a media source, the campaign names from ALL the sources are displayed.
2. Data Freshness: AppsFlyer aggregates the cohort data on a daily basis. For applications in the UTC - (minus) time zone it may take up to 48 hours for the report to be populated with data. For applications in UTC + (plus) time zone it may take up to 24 hours for the report to be populated.
3. Cohort data can be calculated for 90 days or less. The data displayed will be based on the time offset between the current date (excluding the 12 hour delay) and the later date in the selected range.
4. If cohort size is less than 10 the appropriate row is not displayed.
5. If you change the default time zone of your data on AppsFlyer's dashboard, the Cohort data is also displayed according to the newly selected time zone.