Your question is A/B Testing for Product Features. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You're evaluating a new product feature and need to decide whether an experiment will give a trustworthy read. You want a clear primary metric, a sensible minimum detectable effect, and guardrails so a local win does not hide broader harm.
How do you approach A/B testing for a product feature?