Hi guys 🙂 I read this article from Google last year, Overlapping Experiment Infrastructure: More, Better, Faster Experimentation, I wanted to know how do GrowthBook deals avoiding assigning the same user to experiments that might affect each other significantly, that can introduce complex confounders, how to manage domains etc, there is a easy way to use this framework on GB? or is there another strategy in the platform?
07/01/2022, 3:04 AM
Hi Victor - We have the concept of "namespaces" which lets you create mutually exclusive experiments
07/04/2022, 5:48 AM
Hi Graham! Thanks for the reference, interesting, is there any planning on making multiple namespaces for the same experiment?
There was a use case in my company that it would be interesting, we needed to make some revenue-focused experiments isolated from other page/components experiments, so either the user falls into a revenue-focused experiment or a page experiment.
But if the user falls into a page experiment, suppose Home, the user shouldn't fall into two distinct Home experiments, so it's kinda nested/compound namespaces.
Unless it's possible today doing another strategy 🤔
07/04/2022, 5:21 PM
we do have a plan to support multiple namespaces
but we haven't started on that project currently
it is possible to check the state of another experiment when doing your A/B test however