Should Uber double member discounts?

Uber is considering increasing the member discount on rides from 5 percent to 10 percent. This can affect rider demand, driver supply, marketplace balance, and overall unit economics in a two-sided marketplace. Answer the following: 1. What are the main potential benefits and costs of this policy change? 2. Which primary success metrics, guardrail metrics, and diagnostic metrics would you track? Consider rider conversion, trips per member, gross bookings, take rate, contribution margin, wait time, cancellation, driver earnings, and non-member cannibalization. 3. How would you design an experiment to evaluate this change in a marketplace with interference? 4. If you propose a switchback design, what assumptions must hold for unbiased inference? In what real ride-hailing scenarios could those assumptions fail? 5. Which parameters determine sample size and minimum detectable effect? How does clustering or switchback randomization change the calculation? 6. How would you decide how long the experiment should run? 7. Suppose the experiment is planned for two months, but halfway through the p-value for the primary metric is 0.04. Should the company stop early and launch? Why or why not?