You work on a growth team testing a change to Instagram Reels that is expected to increase engagement near the middle of the AARRR funnel, such as more IG Save actions per viewer. Early reads look encouraging, but you know short-term lifts on Reels can reflect a novelty effect and that leadership may ask for results before the experiment has stabilized.
How long would you run that experiment, and what would you consider before deciding? Walk through how you would choose the runtime rather than relying on an arbitrary fixed number of days.
Primary metric should be explicit and tied to the Reels growth goalRuntime should be driven by power and MDE, not by convenienceGuardrails should protect retention and user experienceYou should account for novelty effect, peeking, and SRM