For clauses (+Did [Not] Perform) are a way of filtering a result set based on other actions a user may have taken. Evaluation of users who did For clauses is relative to every single occurrence of the base event down to the millisecond.
It is important to note that the Did Perform and Did Not Perform criteria are not mutually exclusive.
For example with the following segmentation query results:
If there are users who satisfy both criteria, they can be part of both the Did Perform “Open App” event (Green) and Did Not Perform “Open App” event (Purple) results. This applies in a scenario where a user could have downloaded the app, opened the app, and then downloaded the app again, all in one bucket of time. Although the group of users from the Did Perform and Did Not Perform criteria are not mutually exclusive of each other, the group of users within each data point represents a unique group of users. I.e. Users seen on Feb 5 who meet the first row criteria(A) are made up of 1,693 unique users.
Given this non mutual exclusivity, oftentimes, we will see the sum of the For (Did Perform and Did Not Perform) clauses be greater than when you query for the count of users who performed the same event used in the For clause. I.e. The results from a query of Users who performed ‘Open App’ might equal 2,100 rather than the 2,500 count from summing up the For (Did Perform and Did Not Perform) clauses.
To prevent counting sets of users that aren’t mutually exclusive, you can set explicit date ranges following the For clauses (+Did [Not] Perform) In other cases, one can attempt to further qualify an event where a user has downloaded the app more than once. Maybe they downloaded it once on their phone, opened up that app, and also downloaded it on a tablet, but didn’t open that app. Qualifying the download by device will limit, but not likely eliminate, the count of users who exist in both A and B.
Was this page helpful?