Estimation question

One day, all of New York's Google Maps Street View is lost. What is the cost of rebuilding it?

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Estimation with a real operational process behind it, costing out a physical data collection effort at city scale.

Break the problem into inputs, road miles to cover, capture method, e.g. camera-equipped cars, and processing cost per image.
Estimate New York's drivable road miles, using borough count and average road density as a reasoning anchor.
Estimate capture speed and cost, e.g. miles covered per car per day and cost per car-day including driver, vehicle, and equipment.
Estimate processing cost, cloud storage and image stitching or blurring compute per mile of footage captured.
Combine the pieces, total road miles divided by miles per car-day gives car-days needed, multiplied by cost per car-day plus processing cost.
State what you would validate with real data before trusting the number, actual road mileage and Google's real per-mile capture cost.

Breaks the estimate into capture cost and processing cost as two distinct components, not one lump number.
Anchors road mileage to a real reference point, e.g. NYC's roughly 6,000 centerline miles, marked as an assumption if recalled imprecisely.
Shows the full calculation chain explicitly, miles, car-days, cost per car-day, to arrive at a total dollar estimate.
Sanity checks the final number against plausibility, e.g. comparing it to what a single car-day realistically costs a company like Google.

Chapters of the AI PM course that teach what this question tests.