Real-Time Cohorts

Last Updated: Nov 28, 2016 03:11PM PST

Real-Time Cohorts

Once you have registered your application events, you can define cohort analytics for your applications.

NOTE: Cohort Analysis reports only show data collected after the creation of the cohort analysis.

What is a Cohort Analysis? Why Use It?

Cohorts are groups of people you choose to track and compare over time. Cohort analyses give you the power to measure user engagement and app monetization with a cohort you create.

There are infinite use cases for cohort analyses. You could track a group of users who downloaded your app during the holiday season in order to measure the retention of acquired users from a marketing campaign. If you run Facebook Mobile App Install Ads, you could create a group of users referred from Facebook campaigns. If you wanted to find out how much revenue was generated from users who had reached a certain level in your game for the first time, a cohort analysis would be the the perfect tool. Cohort analyses are incredible instruments for developers, marketers, and advertisers when it comes to measuring engagement activities and revenue-generating events.

Creating A Cohort

When you visit Cohort Analysis you will see a table of your existing cohort analyses and a button to +Add Cohort Analytics. If you have not set up a cohort, the table will be empty.

To define a cohort:

  1. Go to the Real-Time Cohort page
  2. Click + Add Cohort Analysis
  3. Name the cohort
  4. Select an application from the drop down
  5. Select the cohort event.  This event places the user into the cohort.
  6. Select key indicator. This measures an activity of cohort members.
  7. Click Save.
Here is a cohort table with several cohort analyses:

Each row contains:
  • Cohort Analysis Name – The name entered when creating the analysis. You cannot have more than one cohort analysis with the same name.
  • Application – The application which this cohort analysis will be measuring. The application is identified by the ApplicationName(OS) combination.
  • Cohort Event — The event used to place users into the cohort. The default event is First Use of the application.
  • Key Indicator- The application event that is the metric against which user activity will be measured.  A KI can be:
    • An event that you have instrumented in the selected application
    • ‘All Revenue Events’ — a system event that includes all of application’s events that have revenue assigned to them.  If you had 2 events with revenue, such as InAppPurchase and OfferWallPayment, then these would be treated as a single event.  Hitting either event would affect the data points for the KI, such as Unique Users, Events, Revenue, ARPU, etc.
    • ‘All Engagement Events’ — this is the same as ‘All Revenue Events’, except this would include all of the events for which you have assigned an Engagement Index.  Also, you would not see any metrics with respect to Revenue, only Engagement Index.
  • Status- This shows the activation status of a cohort analysis. (see below for more information on activating, deactivating, and reactivating cohort analyses.)
    • Draft — The cohort analysis has been created but has not yet been activated.
    • Active — The cohort analysis has been activated or reactivated.
    • Disabled — The cohort analysis was active but has since been deactivated.


Select a retention-related KI such as ‘searching for a product’ to measure the number of events users trigger and the amount of time they spend in your app. To optimize monetization, select a monetization-related KI such as ‘completing an in-app purchase’.

Analyzing the behaviors of different cohorts over their usage lifecycle will help you learn how macro changes to your mobile applications such as new offers, UI changes, and increased advertising efforts impact engagement. By understanding and linking engagement and monetization metrics to known activities, you ultimately increase retention and boost revenues.

See the Cohort Analysis report documentation for more information.

 
In addition, the following actions are associated with a defined cohort analysis:
  • Manage Segments — Choosing this will take you to the Segments page for the cohort analysis. You will be able to create segments and view the segment definitions for your cohort analysis that will be used in the cohort analysis report.
Instructions on adding a segment to a cohort can be found here.
  • View Reports –Selecting this will take you to the Report page for the cohort analysis. See the Cohort Analysis Reports documentation for more details.
  • Activate — Selecting this will activate the cohort analysis. When you first create the cohort analysis, it will not be active. Once you activate the cohort analysis, data will be collected and the analysis will begin.
  • Deactivate –The only way that a cohort analysis can become inactive (disabled) is if you deactivate it. You can only deactivate an active cohort analysis. When you deactivate a cohort analysis, no data for the cohort analysis will be collected during the time it is inactive.
  • Reactivate — Selecting this will reactivate a cohort analysis that has been deactivated. Data will be collected from the reactivation date and you will see a gap in data for the cohort analysis.
  • Delete — You can delete a cohort analysis using this action. If you do, you will no longer be able to view any of the reports that were generated for this cohort analysis.  The delete action is irreversible.

Data is collected for the cohort only once it is created and does not retroactively add users into the analysis.

Tips & Tricks for Cohorts

Common cohorts to create:
  • Retention cohort
    • Cohort Event: First Use
    • Key Indicator: App Start
  • Monetization cohort
    • Cohort Event: First Use
    • Key Indicator: Purchase Event

You will be able to view reports for your cohort analyses by selecting View Report from the My Cohorts page or the Cohort Name from Used In Cohorts on the My Segments page.

You will see a cohort analysis similar to the one pictured below:

The analysis will allow you to select the following parameters:

  • Start/End Date — The date range for the analysis
  • Application Version — The application version for which you want to view the data. The application was already chosen when you set up the cohort analysis
  • Segment — You can view the data for any segment that you defined for the cohort analysis
  • Granularity — Allows you to select whether you want to see weekly or daily cohorts.
  • Weeks of Activity — You can choose the number of weeks to view past the start of the cohort.  This option will only show if you are viewing weekly cohorts.
  • Display — you can look at the heatmap display or choose to see the numbers/data used to generate the heatmap color in the cells, as well as other data points.
  • View by — You will be able to select the type of data (for each cohort) that will be shown for the Key Indicator in each weekly cell after the first cohort event. You can choose to look at:
    • Unique User counts –The total number of unique users within the cohort that execute the KI during a specific week.
    • Total number of events — The total number of events that are executed by all users within a cohort within a specific week.
    • Revenue — The Revenue generated by all users within the cohort, if the KI has a Revenue attribute associated with it. In this case, the Revenue for the cohort will be shown as well as each cell showing the percentage increase of the Revenue from the previous week and the Revenue for that week.
    • ARPPU — The Average Revenue per Paying User. The value in the cohort cell is calculated as Total Revenue for that cell period / Total Uniques for that cell period executing the KI, if the KI has a Revenue attribute associated with it. The ARPU will be shown on the left. This is calculated as Total Revenue for that cohort for all time periods / Total Cohort Size. Each cell will also show the percentage increase of the ARPPU from the previous week and the ARPPU for that week.
    • ARPU in Cohort — The Average Revenue per User for the Cohort. The value in the cohort cell is calculated as Total Revenue for that cell period / Total Cohort Size. Each cell will also show the percentage increase of the ARPU from the previous week and the ARPU for that week.
    • Engagement Index — The Engagement Index for all users within the cohort, if the KI has a Engagement Index assigned to it.  In this case, the Engagement Index for the cohort will be shown as well as each cell showing the percentage increase of the Engagement Index from the previous week and the Engagement Index for that week.
    • AEPU in Cohort — The Average Engagement index per User for the Cohort. The value in the cohort cell is calculated as Total Engagement Index for that cell period / Total Cohort Size. Each cell will also show the percentage increase of the AEPU from the previous week and the AEPU for that week.


How To Read A Cohort Report

Each row shows you data for a specific cohort in which users have triggered the cohort event. Each row shows the day or week that users entered the cohort and how many unique users (cohort size) entered the cohort.

Each column will show the data for the selected view by method for members of the cohort for the days or weeks after they entered the cohort. If viewing daily cohorts, you will see data for the first 14 days, the 30th day, and the 60th day. Hovering over any cell will show more detailed data. If a cell is white, then there is not yet data for the cohort for that day or week. In the case of Uniques, hovering over a cell will show you the total number of uniques in that cell as well as the percentage of total unique users in the cohort that this number represents.  NOTE:  Day 1 or Week 1 specifies the same day that the user was placed into the cohort and Week 1 specifies the same week that the user was placed into the cohort.

If you have chosen a KI of ‘All Revenue Events’ or All Engagement Events’, then the data related to the KI will be generated through an aggregation of such events.In other words, you may will be treating the events as a single, combined event. You will not be able to view Revenue-related data if the KI is ‘All Engagement Events’ or view Engagement data is the KI is ‘All Revenue Events’.

The legend on the right shows how to interpret the heat map with respect to Level of Activity, red being very high and black being very low.

The following is an example of a report viewed by Events. Hovering over any cell will show you the number of events in that cell as well as the total events per user this represents users within the cell.

The following is an example of a report viewed by unique users using the cell data display.

 
Once you have registered your application events you can define cohort analytics for your applications.
support@apsalar.com
http://assets1.desk.com/
apsalarinc
Loading
seconds ago
a minute ago
minutes ago
an hour ago
hours ago
a day ago
days ago
about
Invalid characters found
/customer/en/portal/articles/autocomplete