#ask-questions
Title
a
adorable-solstice-28735
07/14/2023, 4:59 PMHi everyone, I have a question about variant rotation: is it normal to see such gap between variants? I know GA4 will take time to aggregate those events (it hasn't been ~48h yet), but I wonder if you guys see the variants rotating closely
my id attribute is a client_id parsed from Google cookie, just FYI
h
helpful-application-7107
07/14/2023, 5:14 PMAre these users that have seen each event or just events total? If it is the latter, is it possible that clients in control hit the tracking callback more frequently?
a
adorable-solstice-28735
07/14/2023, 5:15 PMthose are users that have seen the experiment and one of the variants
ideally, the gap should be minimal, but maybe I just wait for more data to come in
h
helpful-application-7107
07/14/2023, 5:15 PMso its 56 users (
client_ida
adorable-solstice-28735
07/14/2023, 5:15 PM56 events, sorry
h
helpful-application-7107
07/14/2023, 5:16 PMYeah, so if control users are more likely to hit the tracking call back that could also explain this.
And that would be normal, and would just indicate that control users are, for whatever reason, hitting the tracking callback more. But you can test this by instead seeing unique
client_ida
adorable-solstice-28735
07/14/2023, 5:18 PMI don't have a datasource set up (yet), sending those events to GA4 for testing. Let me process this, I think I am getting users vs events confused in this case
maybe it's better to set up the GA4 audiences instead of this event table
h
helpful-application-7107
07/14/2023, 5:21 PMEach time user hits a feature evaluation, a tracking callback gets fired. So it's possible that a user can fire multiple events for the same experiment. That's just a key thing to start with.
a
adorable-solstice-28735
07/14/2023, 5:22 PMcorrect, that's that what I see in this table. Let me try to segment it by client_id.. or creating the audiences instead. This table is not a valid presentation as I understand
h
helpful-application-7107
07/14/2023, 5:23 PMYes, this view can be informative, but it isn't necessarily indicative of a problem randomizing between variants.
a
adorable-solstice-28735
07/14/2023, 5:23 PMthanks for explaining, it makes sense
I don't think I can achieve this with a simple table, I think the only way to correctly display the results in GA4 would be using the audiences. Or/and connecting the sources of course.
2 Views