[NEW] Ways of Measuring Events (Cohort vs. Actual, Unique, First)

Terminology:

  • A cohort is a group of users with a shared conversion date. Usually, this means a group of users who installed an app on a specific date. Learn more in the Cohorts FAQ.
  • An event is any post-conversion/in-app user interaction, such as a sign-up, a purchase, a video view, leveling up in a game, etc. Learn more in the Events FAQ.
  • Cohort events are events that can be measured in a cohort-based manner, meaning you take a specific cohort (e.g., all the users who installed the app on March 1) and count how many times these users performed an event within a given period after conversion.
  • Actual (actual date analysis) is the opposite of cohort analysis. It means counting metrics and events based on the actual date in which they occurred, regardless of the users' conversion date.
  Meaning Instructions

Cohort Events

How many times was an event performed, by users who installed the app in a specific range of dates, in the first x days after the install?

Example: How many purchases were made by users who installed the app between March 1 and March 7 in the first 7 days after the install?

To see cohort events:

  1. Set up event tracking in your SDK integration.
  2. Define events in Settings > Events.
  3. In your report, select one or more of the events you defined.
  4. Select one or more cohort periods.

The date range you choose for the report is then interpreted as the conversion date.

Unique Cohort Events

How many distinct users who converted in a specific range of dates performed an event in the first x days after the install?

Example: How many distinct users who installed the app between March 1 and March 7 made an in-app purchase in the first 7 days after the install?

To see unique cohort  events:

  1. Set up event tracking in your SDK integration.
  2. Define events in Settings > Events and make sure you define them as unique.
  3. In your report, select one or more of the unique events you defined.
  4. Select one or more cohort periods.

The date range you choose for the report is then interpreted as the conversion date.

First Cohort Events

In cohort analysis, first events = unique events. It's just asking the same question in a different way, e.g., How many users who installed the app between March 1 and March 7 made an in-app purchase for the first time within the first 7 days after the install?

Same as Unique Cohort Events.

Actual Events

How many times was an event performed in a certain date range (regardless of the users' conversion dates)?

Example: How many purchases occurred between March 1 and March 7 (regardless of the users' conversion dates)?

To see actual events:

  1. Set up event tracking in your SDK integration.
  2. Define events in Settings > Events.
  3. In your report, select one or more of the events you defined.
  4. Instead of a cohort period, select "Actual".

The date range you choose for the report is then interpreted as the actual date you want event counts for.

Unique Actual Events [Not Supported]

How many distinct users (regardless of their conversion date) performed an event in a certain date range?

Example: How many distinct users made an in-app purchase between  March 1 and March 7 (regardless of the users' conversion dates)?

See First Actual Events for an alternative.

First Actual Events

In a given date range, how many users (regardless of their conversion date) performed an event for the first time?*

Example: How many users made an in-app purchase for the first time between March 1 and March 7 (regardless of the users' conversion dates)?

Supported for Enterprise Customers only. 

To see first actual events, talk to your Customer Success Manager or open a ticket for Singular Support, detailing the metric you are interested in. 

The Singular team will set up a custom metric for you that will appear in the Reports page.

Notes:

  • If you're tracking users across different devices, note that first events are counted per device. So if a user signs up on one device and then signs up on another device, these are recorded as two "first actual events".
  • In addition to installs, there's another type of conversion called re-engagement (see What is re-engagement?). This affects cohort analysis because when a user is re-engaged, the user's conversion date for the purpose of cohort analysis is the date of re-engagement (not the date of the original install).
  • Re-engagement also triggers a new process of attribution. If the attributed network is different from the network that was credited for the original install, the re-engagement itself and any subsequent activity by the user in the app (including revenue) are attributed to the new network.
Was this article helpful?