Hi there. I am just starting to use Growthbook to run experimentation and also as a feature flag. I might have missed this somewhere - but whats the difference between
• Setting a Experiment override for feature flag
• Creating an experiment that uses the tracking key of a feature flag
02/07/2022, 10:05 PM
Hi! Feature flags and overrides rules are for actually running and implementing experiments in your application. The "Experiments" section in GrowthBook is for the data and analysis side once an experiment is already running.
We're thinking of renaming the "Experiments" section to "Analysis" since this seems to be a pretty common source of confusion for people.
02/07/2022, 10:24 PM
Gotcha. I don't completely follow how variants in "Analysis" would be useful.
If I have a feature flag with 50/50 experiment override to variant A&B
And I have an analysis for that tracking key with Variants X/Y/Z, how would that work?
Let me know if I am getting this right. In the above scenario, users would be assigned to A or B. Why would anyone want to create Analysis around another set of variants? Also how would that be tracked?
02/07/2022, 11:04 PM
The analysis variants just tell GrowthBook which ids to look for in your data source. You always want to make them equal to what's actually being tracked. If you specify a different set of variants in your analysis or add extra ones that don't exist, you'll see a warning in the UI
02/07/2022, 11:32 PM
Makes sense. Thanks for explaining that Jeremy. 🙂
02/08/2022, 1:54 AM
+1 for renaming "Experiments" to "Analysis". The naming was a bit confusing for me as well.