Hi guys! We have some questions about running holdouts: What if the feature won in an experiment in the end of Q1? So it was running in a holdout, for example, only for a week. Should we move this feature to Q2 holdout? Or we just drop it and start an empty new Q2 holdout? How it is usually handled? We are thinking about moving winning features from Q1 to Q2, then winning features in Q2 moving to Q3 (removing Q1 features), etc. So every feature will be running in a holdout for at least 3 months and max for 6 months.

that sounds like an okay plan. There is no one right way to do it.

ok, thank you