I started with population and tried to work down from there, something like residents per household, meals eaten out per week, average restaurant capacity and turnover.
Break the problem into a structured estimation using a top-down approach: start with San Francisco's population, estimate the number of people who eat out regularly, and then derive the number of restaurants based on typical demand and capacity. State your assumptions clearly and show your calculations step by step, focusing on the logic rather than the exact number.
Pro tip: Acknowledge that this is a Fermi problem and that the process matters more than the final number; explicitly state your assumptions and sanity-check your answer against known benchmarks (e.g., restaurants per capita in similar cities).
Define what counts as a restaurant (e.g., exclude fast food chains, include only sit-down? or all food service?) and confirm the geographic boundary (San Francisco city proper).
Start with the population of San Francisco (approximately 900,000) and consider any relevant subpopulations (e.g., tourists, commuters) that affect demand.
Calculate how many meals are eaten out per day: assume a percentage of people eat out or order takeout, and how many meals per week they do so.
Determine the average number of meals a restaurant serves per day (e.g., based on seating capacity, turnover, and hours) to derive the number of restaurants needed to meet demand.
Compare your estimate to known data (e.g., San Francisco has about 3,000-4,000 restaurants) and adjust assumptions if necessary, explaining any discrepancies.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.