Is it possible that a user does not get assigned a...
# announcementsv
victorious-library-28522
08/01/2023, 1:19 PMIs it possible that a user does not get assigned a variation ? I am considering only loggedout users in my experiments
w
white-fireman-22476
08/01/2023, 1:21 PMIs "" the default value you have in your code if the features didn't get loaded?
white-fireman-22476
08/01/2023, 1:23 PMYou might want to change your sql to look at just those 7. Perhaps they all came from the same user, who might be a dev or something trying something out.
v
victorious-library-28522
08/01/2023, 1:25 PMMy default value is “” but shouldn’t the experiment always assign a variation id ?
If I add a default variation Id wouldn’t that create a bias in the experiment?
victorious-library-28522
08/01/2023, 1:26 PMshould the tracking callback be fired if the features didnt load?
w
white-fireman-22476
08/01/2023, 1:30 PMGood point. No the tracking callback shouldn't be fired if the features didn't load.
white-fireman-22476
08/01/2023, 1:33 PMI'm not sure how the "" got set. I can't think of anything within Growthbook that could cause that. I'd examine those 7 a little bit to determine if there is a pattern as to when and with whom it happened.
white-fireman-22476
08/01/2023, 1:34 PMGiven the small numbers it more than likely did not change the statistical significance of your result.
v
victorious-library-28522
08/01/2023, 1:52 PMDidn’t find a variation Id = “” row in my table.
Probably something happens when the query is run
w
white-fireman-22476
08/01/2023, 1:54 PMHmm that is strange. The queries are usually composed of a bunch of subqueries and tables. You could look at those temporary tables that it makes and uses for its sum and compare that sql with the one that you are doing that has no results.
70 Views