Skip to main content

Measuring user resurrection


Of the users who were active 3 months ago (specifically, in the period from 3 months to 2 months ago), but then had no activity 2 months ago (in the period from 2 months ago to 1 month ago), how many returned in the last month?


For this query, we want to measure user activity during the current month and for the previous three months. One way to do this is to use relative time window metrics. When we build a metric and create previous and next time windows, the metric actually creates four time windows: the "current" window (_curr), the "next" time window (_next, which we are not using in this query), and time windows for the previous (_prev) and "before previous" (_prev2T) time periods. 

So we'll create a query using the current month as the "current" time, and look at how many users were active in the _prev2T time period (between 2 months and 3 months ago) but not active during the month after that (in the time period measured by the _prev metric). 

Building the query

Start by building the metric that we will use. This is a basic count of events for the actor, but with an interesting time override: 


Notice that this metric uses a time override to generate 1 month time windows. As you can see in the upper-right corner of the image, this creates additional relative time window metrics that we can use in queries.

This metric counts the number of events that every actor (in this case, users) performed during the current month. But it also counts the number of events that happened on the next month, last month and two months ago. These metrics are named metricname_currmetricname_nextmetricname_prev, and metricname_prev2T and you can use them just like any other metrics (to build queries or even other named expressions). 

Now we can use these metrics in a query. 

We'll measure activity over the last month; keep in mind that if a user isn't active in the current time period, then they don't exist to Interana. This means that Interana will only consider actors active during the _curr time period.

We can add filters to look at users who were inactive during the previous month, filtering to where the _prev metric is 0, and active in the month before that (where _prev2T is greater than 0).


Tip: add an empty filter for reference

The example uses a Table view, and includes an empty filter (filter B in the image). This second filter shows the count of all active users in the previous month, including those who were active during the previous two months. This allows us to compare the number of resurrected users (filter A) to the total number of users (filter B). 


  • Was this article helpful?