Skip to main content
Interania

Use relative time window metrics

0votes
5updates
205views

When creating metrics, you can have Interana automatically create previous, current, and next time window metrics. To create these metrics, select Generate previous and next time window metrics when you're creating or editing a metric.

These generated time window metrics compute the specified measure for the two periods prior to the current time window, the previous time window, the current time window, and next time window for a given point in the time series. The metrics are called <user.metricName>_prev2T<user.metricName>_prev, <user.metricName>_curr, and <user.metricName>_next, respectively.  

Requirements

This functionality was added in Interana Version 2.18.1. 

This functionality can only be used with time series queries.

Because Interana calculates activity based on the current time window, an actor must have events in the current window to be included in the computation of other time windows. If the actor doesn’t have events in the current window, that actor will not be included in the other time windows (even if the actor has events in those time windows).

How time window metrics work

For example, if a metric named “foo” does a Count Events calculation for each user ID with a time window of 1 week, the auto-created metrics will compute the following for each point in a time series:

  • The number of events for each user in the current 1-week window ("foo_curr")
  • The number of events in next 1-week window ("foo_next")
  • The number of events in the previous 1-week window ("foo_prev")
  • The number of events in the time window that begins 2 weeks prior to the current date ("foo_prev2T")

This shows how the metrics map to a calendar view of the time windows:

Calculating Retention, Resurrection, and Churn 

Continuing the user activity example, you can use the time window metrics to calculate retention, resurrection, and churn by checking for user events in the various time windows.

The following table shows the time windows where the actor must have events in order to calculate retention, resurrection, and churn.

  Prev2T Prev Current Next
Retention     X X
Resurrection (returning users) X   X  
Churn     X  

Use the following filters to calculate these metrics:

  • Retention: user.foo_curr  > 0 AND user.foo_next > 0
  • Resurrection: user.foo_prev2T > 0 AND user.foo_curr > 0
  • Churn: user.foo_curr > 0 AND user.foo_next = 0
  • Was this article helpful?