Identify Booking Drivers
Company: Turo
Role: Data Scientist
Category: Analytics & Experimentation
Difficulty: hard
Interview Round: Onsite
You are a Data Scientist at a peer-to-peer car-sharing marketplace. The team wants to understand which product, listing, host, renter, and market features have the largest impact on whether an exposed listing is booked.
Assume you have one row per eligible listing impression or listing-detail-page view. Each row includes: `user_id`, `session_id`, `listing_id`, `host_id`, `impression_timestamp`, `market`, trip dates, listing price, fees, discount, distance from renter, vehicle class, vehicle age, photo count, host rating, review count, cancellation history, response time, instant-book eligibility, availability, renter history, device, traffic channel, and whether the impression converted into a booking within 24 hours.
Design an analysis and presentation that answers:
1. How would you define the target, denominator, and primary metric?
2. Which features appear most important for predicting booking conversion?
3. Which features are likely causal levers versus merely correlated signals?
4. How would you control for confounding, selection bias, seasonality, and marketplace effects?
5. What recommendations would you make to product, pricing, host-quality, or search-ranking teams?
6. How would you validate the recommendations with offline analysis and experiments?
Quick Answer: This question evaluates skills in product analytics, causal inference, feature importance modeling, metric definition, and experimental validation within a marketplace context.