Fermi estimate of annual new-car production volume
Company: Google
Role: Product Manager
Category: Product / Decision Making
Difficulty: medium
Interview Round: Other
Estimate how many brand-new cars were built over the most recent full year. Explain how you frame the estimate, label each assumption, and finish with a number and how confident you are in it.
```hint Pick a side to build from
Decide whether you will reason from why buyers need new cars each year or from how much the industry can turn out, and settle the geography first, because it changes which side is easier to defend.
```
### Clarifying Questions
- Is this worldwide or for one country? If one country, do cars assembled there for export count, and are imported cars sold there left out?
- Which vehicles count as cars: are SUVs, pickups and minivans included, and are heavy trucks, buses and motorcycles excluded?
- Was the most recent year a normal one for the industry, or one disrupted by a parts shortage or a sudden drop in demand?
- Is an order of magnitude enough, or should the answer land within about 20%?
### What a Strong Answer Covers
- A settled scope for geography and vehicle type, and how cars built differ from cars sold under that scope
- A top-line equation linking a year's output to quantities that can be reasoned about
- Labeled, round assumptions with a rationale each, split wherever markets behave very differently
- Correct arithmetic with consistent units
- A cross-check from a second angle, a final range, and the inputs that move it most
### Follow-up Questions
- How would your answer change for a single country that imports a large share of the cars it buys?
- Which input would you verify first, and what data would you use to do it?
- How would a lasting shift toward car sharing or ride-hailing change this number over the next decade?
Overview: A product manager estimation prompt asking for the number of new cars built over the most recent full year. It tests how a candidate scopes geography and vehicle types, separates production from sales, builds a defensible equation from labeled assumptions, and sanity-checks the result with a range.