Define ride success metric for Uber
Company: Uber
Role: Data Scientist
Category: Analytics & Experimentation
Difficulty: hard
Interview Round: Technical Screen
Define a single primary metric to measure 'Uber ride success' across markets and cohorts. Provide: (1) the exact formula and units; (2) inclusion/exclusion rules for canceled/no‑show/reassigned/shared and multi‑leg trips; (3) the aggregation window (trip vs daily vs rolling 28‑day) and how you handle right‑censored or incomplete trips; (4) normalization so cities with different trip lengths, surge, and traffic are comparable; (5) guardrails to prevent gaming (e.g., incentives to cancel to avoid late arrivals); and (6) a validation plan showing the metric’s sensitivity and directionality using historical backtests and an A/B test design with primary/secondary metrics, MDE, sample size, and power. Finally, list two plausible failure modes (e.g., Simpson’s paradox across neighborhoods, seasonality drift) and how you’d detect and mitigate them.
Quick Answer: This question evaluates skills in defining product-level KPIs, statistical validation, and experimental design for an on-demand mobility service, covering metric specification, normalization across markets, inclusion/exclusion rules, anti-gaming guardrails, and failure-mode analysis.