https://www.growthbook.io/ logo
#ask-questions
Title
# ask-questions
n

narrow-lion-95606

12/05/2023, 3:29 PM
Hello everyone šŸ‘‹ I hope you're doing well. I'm currently exploring GrowthBook, and I was wondering if anyone could help me with some questions I have about percentage rollouts. I've searched through the documentation, but I couldn't find the answers I was looking for. I have a use case regarding releasing a new feature to a specific group of users. ā€¢ If I want to release the feature to only 10% of users named John, will the list of users affected by the feature flag persist? In other words, if John A once gets access, will he always be included in that 10% of affected users, or will the access randomly change according to the traffic? ā€¢ Also, is there a way for me to keep track of which users are affected by the feature flag? I noticed in the documentation that you do not store any attributes on your side, so I'm curious if and how this would be possible. Any assistance would be greatly appreciated. Thank you! šŸ™
b

brief-honey-45610

12/05/2023, 3:31 PM
Hey Katarina šŸ‘‹šŸ»
Welcome to GrowthBook!
ā¤ļø 1
> if John A once gets access, will he always be included in that 10% of affected users, or will the access randomly change according to the traffic? Yes, he will alwys be in the same 10% of users. This is because of how we do assignments, which is with a unique hash of both the user_id and the experiment_id. As long as John A always has the same user_id and we're talking about one experiment (so, a consistent experiment_id), John A will always be shown the same variation for the duration of that experiment.
is there a way for me to keep track of which users are affected by the feature flag?
You could make use of a Saved Group, perhaps: https://docs.growthbook.io/features/targeting#saved-groups You'd need to query your data source for all the user_ids that were included in the feature flag. You could save those user_ids into a Saved Group, and use that Saved Group for future testing.
n

narrow-lion-95606

12/05/2023, 3:46 PM
Hey August! Thanks for the quick answer and clarification! I have a few follow-up questions 1. Just as a sanity check, the
experiment_id
is referring to the feature flag identifier, unrelated to the applied override rules? 2. Are there some limits for Saved Groups? If I'm working with large data sets of e.g. 100k+ users, can I save them in one group or multiple ones? 3. Can I combine Saved Groups and percentage rollout and, still see what users are affected?
b

brief-honey-45610

12/05/2023, 3:53 PM
Just as a sanity check, the
experiment_id
is referring to the feature flag identifier, unrelated to the applied override rules?
In GrowthBook, the
feature id
and
experiment id
serve different purposes and are not necessarily the same. ā€¢ Feature ID: This is the unique identifier for a feature flag within GrowthBook. It is used to control and roll out features to users. When you define a feature flag, you give it a unique key that you use in your code to check if the feature should be enabled for a particular user. ā€¢ Experiment ID: This is the unique identifier for an experiment, which is used to track and analyze A/B tests. An experiment can be associated with a feature flag when you are testing different variations of that feature. When you set up an experiment as a rule for a feature flag, the
key
of the experiment acts as the
experiment_id
. This
key
is used to track the experiment's performance and to identify which variation a user has been assigned to.
šŸ‘ 1
The experiment key is also referred to as the "tracking key" and it's used to identify experiments in the SDK. It is part of the configuration when setting up feature flag experiments. The tracking key can be set when you are creating or configuring an experiment rule in the GrowthBook UI. By default, the tracking key is the same as the feature name, but it can be any string you choose.
Are there some limits for Saved Groups?
Yes, currently they can support up to 1,000 users. This is the "legacy" style of Saved Groups which we now call Inline Saved Groups. We recently released Runtime Saved Groups which have no group size limit. https://docs.growthbook.io/features/targeting#saved-groups
Can I combine Saved Groups and percentage rollout?
Yes. If you specify both, they will be ANDed together to produce the final condition that is sent down to the SDKs. The final targeting condition will ensure that only users who meet both the Saved Group criteria and the percentage rollout criteria will be included in the experiment or feature flag rollout.
n

narrow-lion-95606

12/05/2023, 4:06 PM
Thank you so much for the clarification! Have a nice day! šŸ™Œ
b

brief-honey-45610

12/05/2023, 4:27 PM
You're welcome, Katarina šŸ˜„ Let us know if you have any others questions!
4 Views