Analyze DoorDash marketplace product decisions

Assume you are a data scientist focused on product work at DoorDash. Walk through how you would tackle the three product analytics and experimentation questions below.

1. Top Dasher program DoorDash is evaluating modifications to the Top Dasher program, which grants selected dashers extra benefits and could influence fulfillment quality, dasher motivation, and overall marketplace equilibrium.

• What are the primary advantages and disadvantages of the program for consumers, dashers, merchants, and DoorDash itself?
• Which success, secondary, and guardrail metrics would you specify?
• What randomization unit should an experiment use, and what is the reasoning?
• If the treatment group's primary metric comes in below the control group's, how would you investigate before choosing to launch, adjust, or revert?

2. Order cancellation rate is increasing Imagine the overall order cancellation rate has increased substantially over the past few weeks.

• How would you diagnose what is happening?
• Which areas of the organization or stages of the product funnel might be driving the increase?
• How would you distinguish probable root causes from mere correlations?
• How would you test those hypotheses and rank the actions to take?

3. Merchant-created promotions vs. automatically generated promotions DoorDash is choosing between two merchant promotion systems:

• merchants build and set up promotions manually, or
• DoorDash automatically suggests or starts promotions for them. Compare the benefits and drawbacks of each approach, covering the trade-offs around merchant control, adoption, incremental demand, profitability, and the health of the marketplace. Then outline an experiment to determine which approach is better:
• specify the key product and business metrics,
• select the appropriate randomization unit,
• address spillover effects and selection bias,
• and describe how you would interpret outcomes when different stakeholders gain in different ways.

Overview: This question assesses product analytics, experiment design, causal inference, metric selection, and marketplace economics in the context of a delivery platform's product choices, and it falls under Analytics & Experimentation.