This message was deleted.

Looking closer at it now, it seems like I can discard the draft, but the updated traffic percent is still applied. This is pretty confusing as the UI suggest that all changes are first stored as drafts before publishing 🤔

We're working on some clarifications / redesigns to make this easier and more transparent. That said, all attributes (targeting, rollout percentage, etc) of a linked experiment are not managed by the parent feature and its draft/revision control. Instead, these properties are actually managed in the experiment itself, which is not under any sort of revision control.

Got it! As a first step, not creating the empty draft would result in a lot less confusion I think