I am trying to build a segment of visitors that visited us in the last 30 days and to find out how many of them visited the app in the months before.
I.E If we were to use today's date as reference Mar 13, I want to segment a group of unique visitors that visited us from Feb 11 - Mar 13, and find out how many of them visited us in Jan too.
I tried building the segment using visitor in last 30 days date range but it will only return visitors for last 30 days, and I am not able to find out how many of them visited us in Jan. Thank you!
Have you considered using a Cohort Table?
If you use the above (yellow highlights), this will build a table based on Month Over Month Visitor Retention. (Cohorts are based on Visitors).
If your panel is looking at last 6 months, for instance, it will result in 6 rows of data (for each month of the panel):
The cohort column is the "month that is considered the start of visitor calculation - within the row"...
So Sep 1 - Sep 30 had 363K visitors; 1 month after that 29K Visitors came back in Oct; from that 28K visitors came back in Nov; from that 19K came back in Dec, etc
Now, since we are looking at "last 6 months", the Oct row can only calculate out "4 months beyond", and Nov can only calculate out "3 months beyond", etc
Now, going back to building the cohort, you would have noticed the red arrow pointing to "rolling calculation"...
In the non-rolling that I showed you; the visitors did not have to come to the site in every subsequent month... a user who was on the site in Sept, may have only returned in Dec (they would still be counted in the +3 Month column).
However, if you enable rolling calculations; this means that in order to be included in the each column, the visitor must have visited the site every previous month in the cohort...
If I enable rolling, you can see how different the values become:
For more information on Cohort Tables:
Is there a way to build a segment instead? As I would like to break it down by specific dimensions which a cohort table does not allow this option. Thanks!