How cohorts work

Cohort analysis is highly customizable and has numerous filters and metrics you can apply that are not available for other data in Adjust. Understanding how these metrics work and are calculated is important to ensure analysis is correct.

Cohort periods

Cohorts can be grouped by days, weeks, or months, and follow the same convention. Adjust considers the moment of install as "0". So the first week or month of install is W0 or M0, and the following week or month is W1 or M1.

Day cohort periods are calculated as follows:

  • D0 = 24 hours after install
  • D1 = 48 hours after install
  • D2 = 72 hours after install

A week is calculated as 7 days. This means that weekly cohort periods are calculated as follows:

  • W0 = D0 to D6
  • W1 = D7 to D13
  • W2 = D14 to D20

A month is calculated as 30 days. This means that monthly cohort periods are calculated as follows:

  • M0 = D0 to D29
  • M1 = D30 to D59
  • M2 = D60 to D89

Cumulative vs non-cumulative

Some cohort metrics are cumulative. This means that if you set the cohort period 6D, the metric is calculated by adding 0D + 1D + 2D + 3D + 4D + 5D.

For cohort metrics that are non-cumulative, the metric provides you data from the cohort time period you set. So, if you set the cohort period 6D, you only see data for the 7th day after install.

When working with non-cumulative metrics, pay attention to how you set the cohort period.
For example: If you want to see data from the whole first week, set the cohort period as 0W. This ensures the data is not only from the seventh day.

Cohort maturity

Cohorts can be described in terms of maturity: those that are mature or immature. These terms refer to the length of time that has passed since the Nth D/W/M of the cohort, determining whether or not all users within the cohort have completed the selected cohort period.

Mature cohort example: The mature cohort only shows data for periods that have elapsed in full. So, if you are looking at the 0W cohort after you release your app, all users must have had your app installed for at least 7 days. This means that today’s date must be no earlier than the 14th date after your app release date.

  • In the graphic, user C installed the app on D6 and still needs a further 7 days to be included on their seventh day after install. When viewing your matured 0W cohort, your cohort size is the same on day 0 as it is on day 6 after install.

Immature cohort example: The immature cohort shows the accumulated data so far for the cohort period.

  • In the graphic, we assume that today is March 6th and you look at your 0W cohort. This means only user A and B are counted, and neither user will complete the full 7 days after install. Using the table below, you can see how this means your cohort size decreases as you look at further days after install.
Cohort size2222110

Impact on metrics

When analyzing your cohort data, it is important to know whether your cohort is mature or not. As already shown, an immature cohort can mean your cohort size decreases as you move further away from the install date.

To help manage this, Adjust offers metrics that are calculated using different definitions of the cohort group. This lets you specify which users to include when making your analysis.

As an example, we can look at the Ad Revenue cohort metrics:

  • N days Ad Revenue

  • N days Ad Revenue Total

  • N days Ad Revenue Total per User

  • N days Ad Revenue Total in Cohort

In this scenario, we recently launched an update to our app and want to look at ad revenue data for March so far. Today is 12pm on March 4th, and we create a report from March 1st.

To start, let's look at ad revenue earned on the third day after install (D2). Here's how each metric would calculate and present that data.

2D Ad Revenue:

This metric is non-cumulative and shows the ad revenue earned on day 2 alone.

User A0
User B$1
User C0

2D Ad Revenue Total:

This metric is cumulative and shows the ad revenue earned on day 0, 1, and 2 added together. A user does not have to spend the full 24 hours in day 2 to qualify for this calculation.

User A$1$1$1
User B$1$1$2
User C0$1$1
Total amount:$2$3$4
Total amount per user:2/2 = $13/3 = $14/3 = $1.33

2D Ad Revenue Total in Cohort:

This metric is cumulative and also shows the ad revenue earned on day 0, 1, and 2 added together. However, ONLY users who have completed the entire 24 hours of day 2 are included in the calculation.

User C has not yet completed their D2, therefore they are not included in the D2 cohort.

User A$1$1$1
User B$1$1$2
User C0$1Not included
Total amount:$2$3$3
Total amount per user:2/2 = $13/3 = $13/2 = $1.5