With Experiments, you can create several variations of your features and test them using the Experiment rule to determine which one yields the best metrics or delivers the best UX. You can use variations and experiments together to quickly and continuously test and optimize your features.
To create a feature experiment, add a new rule and select the Experiment type to start.
Give your experiment a name and select an audience segment that users will be selected from. Once done, you’ll need to define the exposure rate for the selected segment, followed by the traffic split for variations you are testing.
In a feature experiment, the control variant serves as the baseline or reference, representing the default experience users receive without any changes. The treatment variant(s) introduce the new feature or changes you want to test. The goal here is to compare user behavior and metric outcomes between the control and treatment groups to determine if the changes have a positive impact or otherwise.
You may be prompted to associate at least one goal with your feature flag if you haven’t already so that you can ensure your feature experiments provide clear, actionable insights. Always try to align your experiment setup with your product or business goals to get the most out of experiment results.
