Your question is Mobile Flag Rollout Experiment Analysis. 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 work on a mobile product where a new feature is being released behind a feature flag and rolled out in stages. Your team wants to use the rollout as an experiment, but there is concern that mobile app behavior, staged exposure, and logging issues could distort the results.
How do you manage the feature flag, staged rollout, and experiment analysis on mobile? Explain how you would define the test, choose the unit of randomization, and decide whether the result is trustworthy enough to ship.