Case interviews · 8 min read
Guesstimate Questions: a Four-Step Method, a Worked Example, and How Interviewers Score Them
Solve a guesstimate in four steps: clarify what is being counted, break it into a formula of parts you can reason about, put a stated assumption on each part and do the arithmetic aloud, then sanity-check the result. Interviewers are not scoring the final number. They are scoring the structure, the quality of your assumptions, and what you do when they challenge one.
The four steps
| Step | What you do | What the interviewer is looking for |
|---|---|---|
| 1 · Clarify | Pin down the unit, geography and time period | That you do not start calculating an ambiguous question |
| 2 · Structure | Write the estimate as a formula of 3–5 parts | A breakdown with no overlaps and no gaps |
| 3 · Assume and calculate | State each assumption with a reason; round; compute aloud | Reasoned inputs, clean arithmetic, no silent maths |
| 4 · Sanity-check | Compare with something you know; name the weakest assumption | Judgement — and honesty about uncertainty |
Worked example: cups of tea sold by roadside stalls in Pune per day
Clarify. Roadside stalls only — not homes, offices or cafés. Pune city, an ordinary working day.
Structure. Cups per day = population × share who buy stall tea on a given day × cups per buyer.
Assume. Every figure below is an assumption made for the exercise, stated so that it can be challenged:
| Segment | People (assumed) | Buy stall tea on a given day | Cups each | Cups per day |
|---|---|---|---|---|
| Working adults who commute | 25 lakh | 40% | 2 | 20 lakh |
| College students | 8 lakh | 50% | 1.5 | 6 lakh |
| Everyone else (children, elderly, at home) | 37 lakh | 5% | 1 | ~1.9 lakh |
| Total (population assumed ~70 lakh) | 70 lakh | ~28 lakh cups |
Sanity-check from the supply side. If a stall sells about 250 cups a day, 28 lakh cups implies roughly 11,000 stalls — about one for every 640 residents. For an Indian city that feels plausible rather than absurd, so the estimate stands. The weakest input is the 40% for working adults; at 30% the total falls to about 23 lakh.
Notice what earned the marks: the segments behave differently so they were split; every number was announced as an assumption; and the answer was checked from a second direction, with its most fragile input named.
The follow-ups — where the real scoring happens
| The interviewer says | What they are testing | A good response |
|---|---|---|
| “Why 40%?” | Whether assumptions have reasons | Give the reasoning, offer a range, show the effect on the total |
| “Your answer is ten times too high. Where is the error?” | Composure, and knowing your own model | Go to the most uncertain input first, not the arithmetic |
| “Now do it for the monsoon.” | Whether the structure is reusable | Change one or two inputs; do not rebuild |
| “So should we open a tea-stall chain?” | The “so what” | Turn cups into revenue, mention margins and competition, give a view |
Common mistakes
- Starting to calculate before clarifying what is being counted.
- One giant average instead of two or three segments that behave differently.
- Silent arithmetic, followed by a number. The interviewer cannot score what they cannot hear.
- False precision — “27.9 lakh” from inputs that were guesses. Round.
- No sanity check, and no idea which assumption matters most.
- Defending a challenged assumption instead of testing it.
How to practise
Do one a day, aloud, with a timer set to eight minutes, and write the formula before any number. Then practise being challenged, because that is the half you cannot do alone. Mentari’s consulting, product manager and MBA placement mock interviews pose estimation questions by voice or text and push on the assumption that carries the most weight; the scorecard then quotes where your structure held.
Keep reading
Last updated · Written by the Mentari team