Assess Adding Bicycle Dashers

DoorDash is weighing whether to let bicycle couriers (bike dashers) make deliveries in a city where car and scooter dashers currently handle most fulfillment. How would you assess if introducing bike dashers is beneficial? Address each of the following:

1. Historical analysis: Which historical marketplace data would you examine to judge whether this city or particular zones are a good fit for bike dashers?
2. Success criteria: Which primary, secondary, and guardrail metrics would you follow? Cover consumer experience, merchant experience, dasher experience, and unit economics.
3. Experiment design: How would you set up an A/B test or marketplace experiment to estimate the causal effect of allowing bike dashers?
4. Network effects / interference: Introducing bike dashers may influence dispatch and matching for every order, not only the treated group. How would you account for these spillover effects in the experiment design and analysis?
5. Decision-making: Which results would persuade you to launch, limit, or roll back the change? Assume you can access order-level, merchant-level, dasher-level, zone-level, dispatch, weather, and geography data. Overview: The question assesses skill in experimental design, causal inference, marketplace analytics, metric definition, and handling network effects on a two-sided delivery platform.