This guide will help you understand cohorts and give you deeper insights into your tracking data and your user behaviour.
1What are cohorts and how do you work with them?
A cohort is a group of users that share a particular trait, such as that they triggered an event during a specific time span. Within the Adjust cohort analysis, this common denominator is usually the install or reattribution. When defining a cohort in your Adjust dashboard you select all installs and reattributions over your defined time period. In the analysis all subsequent sessions and events are triggered from users within your cohort. Throughout the analysis only this data is taken into account.
What is the benefit of a cohort analysis?
With cohort analysis, you are able to compare apples with apples. Any user that installs your app has an engagement lifespan, and this lifespan expresses trends that you can manipulate. The changes you perform are not really noticable when your user base is constantly changing. So, to clearly see effects of your manipulations you need to look at the same user group over time and this is where the cohort analysis comes in.
Cohort analysis removes the confusion that can arise from running simultaneous improvements and campaigns, so you can get a better sense of what is yielding the best results and is therefore worth investing in.
2User segmentation and user lifespan
Let’s start simple — we segment the users in your mobile app by the date on which they installed the app. The user’s performance data is then aggregated by the install-week segments. This already provides for a whole new definition — pick out the revenue figures from a single install week, on a given number of weeks after they installed.
Once you have these segments, you’ll want to compare their lifespans. Simply line two cohorts up so that you’re comparing metrics for the first week after install, second week after install, and so on.
Since you’re directly comparing users at equivalent times in your app, you’ve removed the interference that comes out of their changing lifespans.
Cohorts give you the capability to track user segments from a specific time period. Engagement rates vary not just between different users but also between different stages of those users’ lifespans. Efforts at optimization — be it marketing, reengagement, or product updates — will often move the needle for some groups and not for others.
3Setting your cohorts
In your Adjust dashboard you will define your cohorts based on the install and reattribution date by using the available filters for time, platform and country.
Further you can segment your data either by your various acquisition channels or by the install day.
The cohort period can be divided by days, weeks, or months after install. When looking at younger cohorts you would go with days after install, for more resolution, and when looking at older cohorts you would select week or month after install. The in-depth daily resolution is presented up to 30 days after install, weekly resolution up to 12 weeks, while the monthly resolution is unlimited.
The cohorts dashboard provides you with two different ways to segment your users:
Segmentation by tracker: Any incoming SDK traffic is always attributed to the corresponding install source: the tracker (marketing channel). Within the cohort analysis we look at the install source as well as at the time the user installed or reattributed to become an active user within the cohort and any post-install sessions and in-app events. The tracker segmentation is not date bound, an user can become an active user at any day within the selected timeframe.
Segmentation by install date: The segmentation is based on the install date - when a user became an active user within the cohort. The user data is aggregated based on the install and reattribution date. This view shows you the activity of users that installed on a certain day and how they develop throughout time, so that you can analyze their lifetime trends and how these change between cohorts.
In the KPI selection, you can fine-tune your analysis. This gives you insights into groups within your cohort, e.g. paying users, or users that triggered a specific event in your app. A full list of Cohort KPIs is available in our KPI Service Glossary.
First you would define the base. The base includes all active users that belong to your defined cohort. You can choose to look at all users or drill down the user base to your paying users. You can also make a more detailed analysis by selecting users that triggered a specific event within your app.
For this cohort, you can now view different metrics like number of sessions, revenue generated, times that the users in the cohort triggered revenue events and non-revenue events. Additionally you have filter options to view the data segmented per user, event or cohort period. The amount of sessions triggered after the install/reattribution in your cohort gives insight into how long your user stay engaged with your app and how deep their engagement is. Over time the number of sessions will decline, but with the help of the cohorts you can pin down this moment and counter act the trend with targeted re-engagement.
Cohort data is not only useful for planning your re-engagement. You can also test if any changes that you have made to your app features affect the behaviour of your users. The idea is to look at a cohort and observe their retention and/or their lifetime value development. In order to see the effect any changes you have made in your app, you create a second cohort of users that have installed after the introduction of your new features. Do the same analysis looking at KPIs like retention, sessions and lifetime value developmemt. This way you can compare directly if the changes you made had the results you were looking for.
All metrics you find in the cohort guide are metrics that apply to the defined cohort and only include data that those users created and will differ from your statistics overview. Within the statistics overview you find all data that has been tracked during a defined timeframe but not solely triggered from users that have installed the app in that timeframe. Cohorts give you exactly that: the in-app actions of a group of users that installed in a given timeframe.
Multiple KPI Selection
Adjust allows you to view multiple KPI’s at once in your cohorts page. In the filter section of your cohorts page, simply click on KPI Selection and select the KPI’s that you would like to view side by side, click ok and then apply. Your page will then refresh to match your changes.
4Cohort data analysis
In your statistics overview in your Adjust dashboard you can see e.g. how many sessions have been triggered within a selected timeframe. Even though you can see developement of activity throughout time you cannot analyse how many sessions have been triggered from the newly acquired users from the selected time period. This is where the cohort analysis comes in.
Within the cohort analysis you can see any metric, such as number of sessions triggered, attributed to the respective marketing channel by which the user was acquired, and for an exact period of install or reattribution dates. Incoming session and any post-install events are attributed to the marketing channel as well as to the day they were triggered.
5Calculated Lifetime Value
In your cohorts you find one important metric that determines if your campaign efforts were successful and you actually acquired the type of quality user that you were looking for: lifetime value or LTV. The lifetime value is the total revenue generated by the average user. In the cohorts you find the lifetime value calculated for all your users that belong to the cohort and, separately, the lifetime value for your paying users.
The lifetime value increases or (in exceptional conditions such as a refunded purchase) decreases over time, and will eventually show you your break-even: the moment when you start making money from your users. When looking at younger cohorts the lifetime value can change depending on how you analyse your data. When you look at data on a tracker segmentation, users who installed at the end of your selected timeframe will mature later, and convert to a paying user later, than users who installed at the beginning of your timeframe.
Also, comparing the lifetime value between tracker-based and date-based segmentation will lead to different results. In each case the data is handled differently, where for one the source is in focus, and for the other the actual install/reattribution date is in focus.
For a full documentation of our Cohort KPIs and how they are calculated, you can review the relevant section of the KPI Service Glossary.