Nobody expects the exact number.
A guesstimate tests whether you can build a defensible number from visible assumptions, not whether you happen to know the real figure. Four steps, applied consistently, get you most of the way there — with three fully worked examples below showing every assumption out loud.
The method
Define → Decompose → Estimate → Sanity-check
3 fully worked examples, every assumption shown.
Before you start multiplying
An estimate is only as strong as its weakest visible assumption.
Which single assumption, if wrong, would change your answer by 10x rather than 10%?
Would a stranger be able to redo your arithmetic from what you said out loud?
01 / The method
Four steps, applied out loud.
The interviewer is not grading your general-knowledge recall. They are watching whether you can build structure under uncertainty, keep every assumption visible and notice when your own answer doesn't make sense.
- 01
Step 01
Define
State exactly what you are estimating — the metric, the geography, the time period and what is explicitly excluded. Most guesstimates go wrong here, before any arithmetic starts.
- 02
Step 02
Decompose
Choose a demand-side route (population × behaviour × rate) or a supply-side route (capacity × utilisation), whichever the case facts support better. Say out loud which one you're choosing and why.
- 03
Step 03
Estimate
Use round numbers you can defend, state every assumption as you make it, and keep the arithmetic simple enough to do in your head or on paper without a calculator.
- 04
Step 04
Sanity-check
Ask whether the final number feels plausible against something you already know. If possible, triangulate with a second, independent route to the same answer.
Where candidates lose marks
Skipping the definition step and estimating the wrong thing precisely.
Hiding assumptions inside the arithmetic instead of stating them out loud.
Using false precision — a guesstimate answered as '₹14,872 crore' signals you don't understand what an estimate is.
Never sanity-checking the final number against something you already know.
02 / Three fully worked examples
Watch the method run end to end.
The numbers below are illustrative — built to show the method clearly, not asserted as verified market data. Try the question yourself before revealing each step.
How many auto-rickshaws are on the road in Bengaluru on a typical day?
Demand-side
The decomposition
Auto-rickshaws needed
= Daily auto-trips ÷ Trips per rickshaw per day
Daily auto-trips
Population × Trip rate × Auto share
How many trips happen, and what share choose an auto.
Trips per rickshaw
Working hours × Trips per hour
How much work one vehicle can realistically absorb in a day.
- 1
Define
Scope it precisely: rickshaws actually operating on a typical weekday in Bengaluru city, not registered-but-idle vehicles and not the whole metro region.
- 2
Decompose
Go demand-side: total daily trips by auto, divided by how many trips one rickshaw can complete in a working day.
- 3
Estimate
City population ≈ 1.3 crore. Assume 30% take at least one motorised trip needing an auto/cab on a given day, and of those, autos capture roughly 40% of trips (versus buses, metro, cabs, two-wheelers). That's roughly 1.3cr × 0.3 × 0.4 ≈ 15.6 lakh auto-trips a day. A driver working a 10-hour day, spending half of it in actual paid trips averaging 20 minutes each, completes roughly 15 trips a day.
- 4
Sanity-check
15.6 lakh trips ÷ 15 trips per rickshaw ≈ 1.04 lakh rickshaws. That is the same order of magnitude as widely cited estimates for a city this size — the number should feel plausible, not exact.
Final answer
≈ 1 lakh (100,000) auto-rickshaws operating on a typical day
Sanity-check habit
If this were off by 5x in either direction, the streets would visibly look very different from what a resident of the city actually observes — that gut check matters as much as the arithmetic.
03 / Continue practising
Now try one live.
Estimation improves fastest with a partner and a timer, inside a full case.