Hey all! I’m using a feature flag with a 50%-50% rollout configuration. However, after implementation, the results are closer to a 30%-70% split. I tested this with around 250 samples collected within a single day. Could this be related to GrowthBook’s rollout algorithm, where the distribution becomes more balanced over a longer duration or with a larger sample size? Or is a sample size of 250 generally sufficient, which might indicate an issue with the implementation instead?

is it a URL redirect by any chance?

No, this is not a URL redirect setup. I’m simply using a feature flag rollout to split users between two payment gateways: • 50% see Payment Gateway 1 • 50% see Payment Gateway 2

ya, that doesn't seem right

Okay, any suggestions? Basically, I want to analyze which payment gateway users prefer.

@fresh-football-47124 is there anything we could share with you and someone from GB can help

@fresh-football-47124 is there anything we could share with you and someone from GB can help

can you share the code you're using for implementation?

Sure, Give me a moment.