https://www.growthbook.io/ logo
#ask-questions
Title
# ask-questions
r

rhythmic-chef-84815

03/12/2024, 1:08 PM
Hello, we are currently running a few experiments and in only one of them we see an SRM warning I checked the implementation and the setup is the same as we always do but this time the warning is there, I am a bit lost on how to debug this. Could someone maybe help out?
f

fresh-football-47124

03/12/2024, 2:37 PM
Hi Rodolfo - SRM errors usually occur when: • You change the split percentage of a running experiment • You have some caching that can cause one variant to be served more than it should be
r

rhythmic-chef-84815

03/12/2024, 2:40 PM
When you say cache you mean that it should be something we did on our side to cache for example? The fact that this one experiment has this errors is what bugs me the most, I checked the code and saw that we are doing things as we always did so I really can not tell what could be causing. we did had more then one phase but both phases were 50/50 and they both had the same SRM issue, one tending more to variation 1 and the other one to control
f

fresh-football-47124

03/12/2024, 2:41 PM
how much off was it?
r

rhythmic-chef-84815

03/12/2024, 2:42 PM
Screenshot 2024-03-12 at 15.42.00.png
The difference is still there and I really can not find out why this is happening exactly, our setup is the same as before and only this experiment has this problem 🤔 It looks like this now
6 Views