DoorDash — Merchant, Ads & Sales — Product Sense & Metrics

Role context: Data Scientist, Merchant/Ads & Sales Analytics · Est. study time: 30 min · 3 practice questions

How to prepare for this role

The day-to-day is measuring merchant acquisition/retention, the Ads business (sponsored listings), merchant ROAS, take rate, and GMV. The instinct that separates strong candidates: knowing that ad-attributed orders over-credit ads (a popular merchant would get orders anyway), so the value is incremental orders, and that ads are an auction where one merchant's win is another's loss.

Where to spend your prep time:

• Merchant/ads product sense (this article) — how ads work, why incremental orders and merchant ROAS beat ad revenue, and the ad-load-vs-consumer-experience tension.
• Incrementality and causal inference — ad incrementality (holdouts, geo), auction interference. See the A/B Test & Causal Inference section.
• SQL and communication — be fluent in SQL, and turn an incrementality or merchant analysis into a clear recommendation.

The through-line: anchor on incremental orders and merchant ROAS, respect the auction's interference, and don't degrade the consumer experience or churn merchants.

What DoorDash Merchant, Ads & Sales actually is

DoorDash's merchant side is the supply of restaurants/stores consumers order from; DoorDash earns a commission (take rate) per order plus Ads, a value-added service where merchants pay to be promoted. The identity to carry into every answer: a marketplace where merchant selection and retention drive consumer choice, and where ads monetize merchant demand for visibility — but only create value if that visibility is incremental, not cannibalized.

Two things make this different from a normal analytics job:

• Ad value is about incrementality, not revenue or clicks. A sponsored placement can be credited with orders a popular merchant would have received organically anyway. The DS's job is to measure the incremental orders an ad causes (and the merchant's true ROAS), not gross ad-attributed orders — the "influenced vs incremental" trap.
• Ads are an auction, so they interfere. Merchants compete for limited promoted slots and consumer attention; boosting one merchant's ad takes orders from others (and from organic). So ad experiments have interference, and the platform trades ad load against consumer experience and organic relevance.

Merchant health is selection (enough good merchants), retention (merchants staying active), and a take rate that funds the platform without driving merchants away.