Hello, I was wondering if someone could help me understand what the difference is between setting an override experiment on a feature and setting up an experiment is? They both seem to have separate traffic split controls and I was wondering what the over lap is with them?

Hi Peter. Yes it is basically two different ways to the same thing.

Why does New Experiment Analysis have a traffic split setting? Does this need to be the same as the feature flag split to analysis the data correctly?

It is mostly there to catch Sample Ratio mismatch errors. We realize it is a bit confusing and are reworking this bit of the UI currently.

No worries. So if understand correct. I need to setup the traffic split on the feature to capture the data and then setup a experiment on the same key to analysis the results. (Keeping the split the same)

Correct

Amazing, thanks again for your help 👍

Sure thing

We actually just changed this a few hours ago to be more intuitive. For new experiment rules on features, targeting and traffic splits are now controlled by the experiment, not the override rule, so there's a single source-of-truth for who's included in a test.