You are interviewing for an Amazon non-technical product or operations role. The interviewer asks: How would you evaluate delivery driver performance? Assume that looking only at the number of delivered packages and total delivery time is not enough. Explain what additional information you would need, how you would build a fair performance framework, and how external factors such as weather should affect the evaluation. ### Constraints & Assumptions - The goal is to evaluate controllable driver performance while separating it from route difficulty, dispatch quality, traffic, weather, and package mix. - The framework should support coaching and operational improvement, not only ranking or punishment. - Safety and customer trust should act as guardrails; speed alone should not define performance. - Use normalized metrics where raw counts would be unfair. ### Clarifying Questions to Ask - Is this score used for coaching, compensation, routing decisions, staffing, or quality investigations? - Are drivers employees, contractors, or third-party delivery partners? - What route, package, safety, customer-feedback, and weather data are available? - How frequently is the score calculated, and who will see it? ### Part 1 - Define The Evaluation Goal What should a driver performance framework optimize for? #### What This Part Should Cover - Balanced goals: reliability, customer experience, safety, efficiency, and compliance. - Separation of driver-controlled actions from system, route, or external factors. - A clear stakeholder view for drivers, operations managers, customers, and the business. ### Part 2 - Identify The Data Needed What information do you need beyond package count and total delivery time? #### What This Part Should Cover - Route complexity, package characteristics, time windows, building/access difficulty, route distance, and stop density. - Delivery quality, proof-of-delivery accuracy, misdelivery, damage, complaints, and instruction adherence. - Safety signals, compliance, traffic, weather, road closures, van loading issues, and dispatch quality. ### Part 3 - Build A Fair Scorecard How would you construct the performance framework? #### What This Part Should Cover - A scorecard with reliability, customer experience, safety, and normalized efficiency. - Cohort comparison against similar routes and conditions rather than all drivers globally. - Guardrails that prevent unsafe speed from looking like high performance. ### Part 4 - Handle Weather And External Factors How should external factors such as weather affect the evaluation? #### What This Part Should Cover - Weather and traffic adjustment, exclusion rules for extreme conditions, and similar-route comparisons. - Treatment of external factors as context, not a blanket excuse for every defect. - A feedback loop that identifies route-planning or operations issues separately from driver behavior. ### What a Strong Answer Covers - Explains why raw packages and time are incomplete and potentially unfair. - Builds a normalized, actionable, and safety-aware framework. - Shows good judgment about incentives and unintended consequences. - Defines how to measure whether the evaluation system itself is fair and useful. ### Follow-up Questions - How would you prevent the scorecard from encouraging unsafe driving? - What would you do if drivers distrust the score? - How would you treat new drivers versus experienced drivers? - Which metric would you use as a guardrail? - How would you detect whether route assignment, not driver behavior, is the root problem?