Lead cross-functional decision without RCT evidence

Tell me about a time you had to recommend a ship/rollback decision when an important feature had already launched globally without a holdout and stakeholders wanted a fast read. What was the conflict, what alternatives (designing a retro holdback, natural experiment, metrics redefinition) did you propose, and how did you influence skeptical partners (PM/Eng/legal/marketing) to align on a path? Walk through how you set decision criteria up front, communicated uncertainty and risks to non-PhD stakeholders, managed timelines, and handled a postmortem. What would you do differently if the team culture were highly academic and passive?