## Evaluate Courier-Selected Delivery Distance Limits A delivery marketplace is considering allowing each courier to set a maximum distance for delivery offers. Explain how you would decide whether to launch the feature and how you would evaluate its marketplace impact. ### Constraints & Assumptions - A courier's selected limit can change which offers they see and may affect customers, merchants, and other couriers. - Outcomes are interconnected: one courier declining a long trip changes the offers available to others. - Stated distance preferences may differ from observed acceptance behavior. - A successful launch should improve courier control without making delivery availability materially worse. ### Clarifying Questions to Ask - Does the limit apply to pickup distance, delivery distance, total route distance, or all three? - Is the feature intended to hide offers, rank them differently, or let couriers express a preference without a hard cutoff? - What problem is the feature meant to solve: low acceptance, cancellations, safety, satisfaction, or retention? - Which markets and delivery modes are in scope? - Can the experiment be randomized by market, courier, or time block? ### What a Strong Answer Covers - Defines the feature, exposure, analysis unit, target population, and primary success metric. - Builds a baseline using the relationship among offer distance, acceptance, completion, earnings, and supply coverage. - Recognizes interference and proposes an experimental unit that protects marketplace validity. - Measures courier choice and satisfaction alongside customer wait, unassigned orders, delivery time, merchant outcomes, and distributional effects. - Examines heterogeneity by market density, time, courier behavior, and selected limit without using post-treatment segments carelessly. - Provides launch, iteration, and rollback criteria tied to both average and tail outcomes. ### Follow-up Questions 1. How would you evaluate the feature in a market where cluster randomization is too expensive? 2. What if courier satisfaction improves while unassigned orders also rise? 3. How would you distinguish genuine distance preferences from reactions to low pay? 4. What evidence would support a soft preference instead of a hard cutoff?