Experimentally evaluate jogging-route recommendations

Google Maps is considering suggesting optimal jogging routes. Design the evaluation as follows: choose one primary metric (for example, jog completion rate) and specify secondary metrics such as route adherence, repeat jog within 7 days, safety incidents, time variance versus the baseline route, and battery drain. Define eligibility and triggering rules—show the feature only when the user has opted into fitness mode, location accuracy is under 30 m, speed is between 1 and 4 m/s, and a route suggestion is shown. Specify the experimental unit and assignment method, address contamination risks (multiple devices, social sharing), and list guardrails including crash rate, navigation reroutes, and ETA accuracy. Calculate the MDE and required sample size for a baseline completion rate of 40% with an expected +3 percentage point lift, 80% power, and $\alpha=0.05$, including formulas and any adjustments for clustering or novelty effects. Propose a ramp plan, geographic stratification, novelty decay checks, and a rollback criterion.

Overview: This question tests experimental design, metric selection and definition, statistical power and MDE calculation, contamination prevention, and operational monitoring for route recommendation features, aimed at an Analytics & Experimentation Data Scientist role.