Measuring Instacart Shopper Supply Against Order Demand by Zone and Hour

Instacart fills grocery orders through shoppers: independent contractors who choose when to work, pick the customer's items in the store and deliver them. You are a product manager on the marketplace. How would you know whether there are enough shoppers to handle the orders in a given area at a given time? Describe what you would measure, at what level of geography and time, and how you would spot a shortfall both while it is happening and before it happens. ```hint Follow a shortage through the system Picture an hour when a neighborhood has too few shoppers. What changes first for orders waiting to be claimed, for the shoppers themselves, and for customers trying to book? Which of those changes can you see early, and which only after customers have been let down? ``` ### Clarifying Questions - What does "enough" mean here: scheduled orders arriving inside the window the customer picked, fast orders meeting the time quoted at checkout, customers finding an open delivery slot, or all of these? - What counts as an area: a single store, a delivery zone or a whole city? And should time be tracked by the hour or by delivery window? - Are both scheduled and on-demand orders in scope, given how differently they use shopper time? - Is the task only to measure and forecast supply, or also to choose the levers, such as shopper incentives, delivery fees or closing slots, once a gap appears? ### What a Strong Answer Covers - A definition of "enough" in terms of customer and shopper outcomes, rather than a raw count of shoppers online. - Capacity expressed as shopper time against the work the orders require, accounting for order size, travel, batching and shopper acceptance. - Real-time supply signals per zone and time bucket, ordered from leading to lagging, and when each one becomes too late to act on. - Whether the observed order count is a trustworthy measure of demand in a zone where supply is tight. - A forward-looking forecast of demand and supply, validated against late and lost orders, with oversupply treated as a failure as well. ### Follow-up Questions - A zone shows orders being claimed within seconds, yet customers there complain they cannot find a delivery slot. What could explain it, and how would you change your metrics? - A shortfall is forecast for the day before a major holiday. How would you choose between paying shoppers more, raising delivery fees and closing slots early? - How would you show that your supply measure actually predicts late deliveries and lost orders?