The GameAnalytics Cohorts feature is a framework built on the classical cohort table view.
Cohorts are defined as: groups of users that have completed a specific action within a specific period of time.
A good example of a cohort would be “the players that first launched your game on July 30th, 2022”.
Why you should use Cohorts?
Cohorts are a great way of analyzing your data. One of its key advantages is being able to follow in-game behaviour such as the adoption rate of a new character or the evolution of players’ buying behaviour over a certain period of time
Define A Cohort
To define a cohort you will need:
- A completed action to identify the users by (this is automatically set by GameAnalytics to “Start game” event, and cannot be changed)
- A specific period of time for which the cohort will be computed (you can specify it in the date drop-down table).
This view of the cohort table contains:
- Granularity buttons – each metric has a preset of granularity and aggregations available. If a metric or aggregation is not available for a particular metric, it is most likely not characteristic to it.
- Metric picker – Choose any of the custom events instrumented to analyze them by cohorts. This is ideal for analyzing things like device level retention to how many users reach a certain level.
- Aggregations – For custom events you can choose to display the values by the Mean, Sum or Count aggregation methods.
For cohorts processing we consider days, weeks and months by calendar. All custom events that you send to the GameAnalytics servers under the same user are counted for a cohort (not just their first occurrence for the period specified in the cohort).