The Cohort module of MoEngage Analytics, as the name suggests, helps you understand user cohorts & retention. Retention & cohort analysis is essential to understand product health with respect to loyal returning users.
A cohort is a set of users, who are identified by their common behavior. The cohort for web and app analytics is a group of users who have performed the same events in a given duration. For example, newly acquired users on a music streaming app who are actively playing music for 15 days.
You can get graphical cohort reports for easy visualization. There are also tabular reports if you’d like to download it for deeper analysis. These reports are helpful for analysing cohorts over time.
You can use Cohorts to answer questions like
- How many users are sticking around with my platform after they sign up?
- What percentage of users who visited a particular product on my e-commerce app, came back and bought the product?
- What is the quality of users from one paid channel compared to another?
Let’s see how it works.
To create a cohort, go to the Analytics menu and click on Cohorts. Select the First Event and Return Event for cohort analysis from the drop-down menu.
You can use filters to narrow down the cohorts to specific attributes. These can be Event Attributes, User Attributes, Device Attributes or a combination of them.
Filters can be added to both first and return event of the cohort depending on what data you want to segment. To do that, click on Add Filters and then select appropriate filters from the Filter By section. Select the Case Sensitive checkbox if you want the report to match the exact case of the value you specify.
If you add multiple filters, you can choose whether they’re all applied or just one of them. Toggle the switch between AND/OR based on what you’re looking for.
You can also compare different cohorts by a specific attribute. You can compare cohorts for any property of the first event. To do that, click on Split By and select the attribute from the drop-down menu. You can see the top 1000 values of the selected event property. The table has all the cohorts and the chart shows the top 5 cohorts.
Setting the data Time Frame
Once you’ve defined the funnel steps, window, and order, select the time frame for which you need the report Duration section. You can choose from “Yesterday”, “Last 7 Days”, “Last 30 Days”, “This Month”, “Last Month”, or select your own custom date range.
MoEngage Analytics provides two different types of cohort analysis - Retention Cohort and Return Visits. You can access this on the chart.
Retention Cohorts provide the percentage of users retained (at least) till the selected day. Some of these users can also be retained for more duration. These are the users who performed the return event until the selected day or later. You can choose to replace ‘day’ with any level of granularity such as Hour, Week or Month.
Let's understand this with the image provided below - User has installed the app (first event) on Week 0. Now user has opened the app (return event) at least ones on Day 0, on Day 1 and on Day 3 but not opened app anytime on Day 2. She will be counted as retained for Day 0, Day 1, Day 2(as well) and for Day 3. User will be counted retained for Day 2 as she was active on the platform for the later period Day 3.
N-Day Retention provides the percentage of users retained on the selected day. These are the users who performed the return event on the selected day. You can choose to replace ‘day’ with any level of granularity such as Hour, Week or Month.
Let's understand this with the image provided below - User has installed the app (first event) on Day 0. Now user has opened the app (return event) at least ones on Day 0, on Day 2 and on Day 4 but not opened app anytime on Day 1 and Day 3. She will be counted as retained only for Day 0, Day 2 and for Day 4. User will not be counted retained for Day 1 and Day 3 as she was not active on these days.
Return Visits show the percentage of users, who come back and perform the return event on a specific day. This can also be seen as, the percentage of users, who did not perform the return event until the selected day. This implies how much time users are taking to perform the return event. Again, you can choose to replace ‘day’ with any level of granularity such as Hour, Week or Month.
For the example provided in the image given below, user installed the app (first event) on Day 0. User did not perform the Purchase (return-event) on Day 1 & Day 2, she is counted as inactive on Day 1 & Day 2. On Day 3, when she performed Purchase (return-event) for the very first time, she is counted as a returned user on Day 3. We do not consider what happens after Day 3 as this analysis only provides how long the user was away from the app, came back and performed the returned event.
The cohort line chart is the weighted average of the percentage of cohort users for the selected day or selected granularity. This calculation works the same way for Retention and Return Visits.
By default, MoEngage Cohorts shows you your App/Site Open retention cohort for the last 6 weeks with weekly granularity.
You can break down the report into the granularity of Hourly, Daily, Weekly, or Monthly for easy visualization. Cohort users will be displayed based on the granularity you select.
There are a few limitations for the cohorts granularity - Hourly cohorts can only be viewed for the duration of a day. Daily cohorts can only be viewed for 90-day durations. Weekly cohorts can only be viewed for 180-day durations.
Cohort Table Report
Table column ‘Cohort’ shows the number of users who performed the first event on a specific day. The rows ‘Day 0’, ‘Day1’,.... ‘Day n’, provide the percentage of users according to Retention and Return Visit explained above. Here, Day can be replaced with any granularity like Hours, Week or Month. The percentage numbers are color graded for a better holistic view.
Once you’ve viewed the report, you might need to share it with your team. You can download the report in order to do that.
Charts can be downloaded in PNG format and you can export a CSV file of the tabular version.