Hello, some weeks ago we successfully conducted an experiment in Growthbook using a feature flag, but decided to stick with the baseline for now since not all results were satisfactory. After making some adjustments in our code, we have now launched a second test using the same feature flag. We noticed that all users who were already part of the first experiment were immediately reassigned (to the same variant as in the first test) as soon as they launched the app, even though the experiment assignment is actually tied to specific activities. This results in an unhealthy distribution across our test variants 😕 When I filter out these users, I get a much healthier user assignment. How can we ensure that one of the following happens: • Users from the first test are excluded from the second test • Users from the first test are randomly assigned to one of the variants in the second test as usual I look forward to your feedback. Thank you in advance 🙏

that's an interesting case

Can I define the reassignment before the test starts?

@helpful-application-7107 do you know?