— GUIDE —
HOW TO ANSWER ESTIMATION QUESTIONS IN QUANT INTERVIEWS
You are not expected to know the exact number. The interviewer wants to see how you build and check a rough estimate.
UPDATED 2026-08-12 · 6 MIN READ · BY THE VARIANCE TEAM
The five-step framework
- Define the target: clarify the geography, time period, and unit.
- Build an equation: split the target into quantities you can estimate.
- State assumptions: choose round, defensible numbers and say them aloud.
- Calculate cleanly: keep units attached and use scientific notation when helpful.
- Sanity-check: test whether the result is plausible and name the most sensitive assumption.
Worked example: daily coffees sold in a city
Suppose the interviewer asks: “How many takeaway coffees are sold in a city of one million people each day?” First define the target as takeaway cups sold within the city on a typical weekday.
Build the estimate from demand: population × adult share × coffee-drinker share × takeaway purchases per drinker.
- 1,000,000 residents
- 75% adults
- 60% drink coffee
- About 1/3 buy a takeaway coffee on a typical day
That gives 1,000,000 × 0.75 × 0.60 × 1/3 ≈ 150,000takeaway coffees per day.
Cross-check from the supply side
Now estimate another way. If the city has roughly 1,000 coffee-selling locations and each sells 150 takeaway cups per day, the supply-side estimate is also 150,000. The agreement does not prove the answer, but it shows the assumptions are mutually plausible.
What the interviewer is listening for
- A clear equation before arithmetic begins.
- Assumptions that are explicit and easy to revise.
- Correct units and order of magnitude.
- A quick sensitivity check: which assumption moves the answer most?
- Comfort correcting course when challenged.
Common mistakes
Do not hunt for a secretly exact answer, bury the interviewer in tiny demographic segments, or use precise-looking assumptions you cannot defend. Two or three meaningful drivers with round numbers usually reveal more judgment than ten fragile assumptions. Finish with a range. Here, 100,000 to 250,000 is defensible. Then say which assumption could push the estimate toward either end.