How many users were active last month but were not active this month (user churn)?
This query is measuring all who were active in any way two months ago, but who were inactive this month. To do this, we'll use a metric that measures activity over the current month, and see when that equals 0 for users who had any activity during the previous month.
We do this because Interana measures the activity of actors: if an actor (in this case, a user) is inactive, then they do not appear in standard queries (they don't exist to Interana in that time range). We're using the metric to determine "when activity = 0 for this month" because we can't query directly for users that don't exist.
Building the query
Start by building a metric to measure user activity over the previous 1 month. Select Generate previous and next time windows; this creates metrics that use relative times windows, and we need a metric that will measure a period of time relative to the query time. See Use relative time window metrics for more information about how this works.
Now we can build the query. We'll measure user activity over the previous month with a basic "Count Unique user" measure and a time period from 2 months ago to 1 month ago.
A "month" in Interana is a 30-day period
Then to see which users were not active for the last month, we'll filtering by one of the relative time variants of our metric where the value of that filter (again, user activity) is zero.
We're using the
_next variant of the metric because this is the NEXT time period, based on the query time (in this case, 1 month ago to now).
To get a more meaningful result, we'll use a Compare filter, using an empty second filter to compare our measure against a query without filters. Remember that a query with the activity metric filter will return all users who were active in the previous month and not active this month; without the filter, the query will return all users who were active in the previous month, including those who were also active this month.
Tip: adding friendly labels
You can click in the area next to the filter headings (A and B in this case) and enter a descriptive name for the filters.