Analytics
1.7K members online now
1.7K members online now
Ask questions about filter set-up and issues with using filters in Google Analytics reports
 
Guide Me
star_border
Reply

Question on session vs user filter

[ Edited ]
Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

Hi,

 

We want to test two remarketing vendors against each other, so we set up some javascript to randomly assign each visitor to one of the two vendors.  If you're assigned to vendor A, then you'll get vendor A's pixel fired each time you come to the site, and likewise with vendor B.

 

When a cookie assignment is made, we push an event to GA with a label indicating which vendor was fired.  Everything seems to be working fine, nice 50/50 split in the assignments when I look at event counts.

 

Here's where things get weird.  When I create a new audience segment, and filter on sessions for sessions that include an event fired from both vendors, I get a substantial number, maybe 10-20% of visitors, both for sessions and users.

 

But then when I change to filter on users, it goes to 0.  No overlap at all.

 

I've never seen this before - normally, the user count I see when I filter on sessions is the count I see when I switch to filter on users.  Session count just goes up because you're now including all the sessions from the users that met that particular criteria.

 

Has anyone seen this behavior before?  I'm at loss for an explanation: if you have overlap at session level, and every session has an associated user, wouldn't you therefore have overlap at user level?

 

Thanks!

Re: Question on session vs user filter

Explorer ✭ ✭ ☆
# 2
Explorer ✭ ✭ ☆
Sounds like you used javascript to perform A/B testing. You could be experiencing a cookie issue, where a returning user is not getting filtered back to the original landing page. That would explain why the users overlap, but the sessions to not.

However, if I read that wrong and it is the sessions that overlap and not the users, then you might need to perform a comprehensive review of how the event is configured and firing. It could be that in some instances the events for both vendors is firing on the same page somewhere.

Re: Question on session vs user filter

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭
Hi Miguel,

Yes, the sessions are overlapping and not the users. You mentioned that in some instances the event for both could be firing on the same page. In that case, wouldn't the users overlap too? User A in session A somehow triggers both vendor A and vendor B events. So on a session or user basis, there'd be overlap. That's what's weird, is when I look at session A, i see overlap, but when I look at user A, there's no overlap.

Re: Question on session vs user filter

[ Edited ]
Explorer ✭ ✭ ☆
# 4
Explorer ✭ ✭ ☆

Users and sessions are two separate metrics. Have you checked to see if the overlap is isolated to either new or returning visitors? This might be tedious, but you could try tracking users by client ID. This might help narrow down when this is occurring. https://developers.google.com/analytics/devguides/collection/analyticsjs/cookies-user-id

If the events tracking code was done manually or dynamically (PHP), I would check the source code on the front end of one of the pages where it is occurring to see if both codes got put there somehow. If the tracking code was done with Google Tag Manager, then look at the Fire On settings for the Tag.