Design DoorDash Marketplace Experiments

You are interviewing for a Data Scientist role at a food-delivery marketplace such as DoorDash. The marketplace has three sides: customers, restaurants, and advertisers/sponsored listings. For each scenario below, describe:

• the product objective and the decision you are trying to make,
• the most appropriate experiment or quasi-experiment design,
• the unit of randomization,
• the primary metric, secondary metrics, and guardrails,
• how you would handle interference, seasonality, and selection bias,
• and what results would make you launch or reject the change.

Scenarios:

1. Auto-apply credits at checkout The app can automatically use a customer's existing credits during checkout. The team wants to change the current experience so credits are applied by default more aggressively. How would you test whether this is a good change?
2. Duplicate restaurant listings in search Sometimes the top organic restaurant result and the sponsored restaurant are the same merchant, so the customer sees two nearly identical listings in the same search results page. Is this good or bad for the marketplace? If you want to change the experience, how would you test it?
3. Restaurant promotions Restaurants can currently opt into a fixed promotion such as " $5 off$ 30".
   • How would you increase restaurant adoption of this promotion?
   • Suppose you introduce a self-serve tool that lets restaurants customize promotions, for example " $3 off$ 15" or " $10 off$ 50". How would you design and evaluate this feature?
   • If you had to compare the existing fixed " $5 off$ 30" option against the new customizable-promotion option, how would you run the test and what would success look like?

Your answer should reflect common marketplace issues such as customer trust, ad revenue tradeoffs, restaurant ROI, cannibalization, and cross-unit spillovers.