One of the most effective analytics tools is cohort analysis. It allows us to track changes in your products by grouping users for any attribute. Let’s talk about cohort analysis, its features, and how it works in app analytics.
For a start, we will figure out what is “cohort’’. A cohort is a group of users that performed a definite action (installed an app in our case) at any time period. Cohort analysis compares only groups, which allows us to observe their activity in time. It doesn’t matter if it’s a day, a week, a month, or a year. It’s much more important for cohort analysis to compare only those cohorts that are using an app for the same amount of time. Otherwise, the metrics of the groups will be very different.
Some metrics, such as cumulative ARPU and Retention, show changes in a product over time so they can only be calculated using cohort analysis.
We are using the cumulative ARPU metric to count the average revenue from a user for a certain amount of time in the app. If we look at the chart, we can see that this value is growing, but over time it turns into a horizontal line and stops changing. This means that users from this cohort stop using the app or pay for full access, which does not require additional payments. This allows us to calculate their LTV. We will use it to predict revenue from new users and other cohorts. For example, we know that the day 30 ARPU is $2. If we acquire 1,000 users for $1,500, then most likely we recoup this spend within a month (under the condition that the characteristics of those players are similar to the analyzed cohort). If we analyze this cohort in 3 months and find out that the ARPU is already $3.5, then we can calculate the revenue from the users who we acquired over this period:
1,000*$3.5-$1,500 = $2,000
If we have changed our product and need to assess their impact but we don’t have the opportunity to run A/B tests, cohort analysis will be the solution. We will compare the metrics of an old version cohort and a cohort formed after changes. For example, we have redesigned the in-game store which could affect the conversion into purchase rate and ARPU. If we compare the performance of the two cohorts, we can see the impact this change had on the audience. We can notice that the second cohort’s key metrics in the picture below are higher so we make a conclusion that the experiment was successful.
To run cohort analysis in devtodev, use a special section on the left menu bar named “Cohort analysis”. On the Custom Cohorts tab, you need to select multiple cohorts and metrics that you want to compare. You can evaluate not only financial metrics but also analyze, for example, behavioral metrics such as successful completion, retention, the number of sessions per user. Focus on the goal of the experiment to choose the necessary metrics.
With the help of cohort analysis, we can compare the same metrics from different users. The main thing to remember is to create the same conditions for the compared users.