Cracking any estimation prompt
The five steps, the two directions, regional assumptions, and four fully worked examples, one of them interviewer-led with data.
Key takeaways
- Every guesstimate follows five steps: clarify what is being estimated, turn it into a short formula, fill each part with a defensible assumption, calculate, then sanity-check against something you know and say what the number means.
- For smartphones sold per year, splitting the population by age helps, because young adults and older people own and replace phones at very different rates.
- People per household: about 2 in Germany, about 2.4 to 2.5 in the UK, Spain, and the US, about 3 in Singapore, about 4.5 in India, and often 5 or more among Saudi nationals.
- Passenger cars per 1,000 people: roughly 500 to 700 in Western Europe; in the US about 800 once SUVs and pickups are counted; about 110 in Singapore; and well under 100 in India.
- Big, dense cities usually have fewer cars per person than their country average, because of public transport and the cost of parking.
Guesstimates, also called market-sizing or estimation cases, ask you to estimate a number you cannot look up: how many cars are in a city, how much revenue a cafe makes in a year, how many smartphones a country buys. They come first because they are self-contained and build the three habits every later case relies on: structuring, math, and defending assumptions out loud. The interviewer is not checking whether your number is right; nobody knows the true number. They are checking whether you can build a sensible answer from a clean structure and clear assumptions, calmly.
Key idea
Every guesstimate follows five steps: clarify what is being estimated, turn it into a short formula, fill each part with a defensible assumption, calculate, then sanity-check against something you know and say what the number means. The structure is the skill; the arithmetic is the easy part.
Clarify before you estimate
- Volume or value: number of units, or money spent?
- Stock or flow: how many exist today, or how many are sold each year?
- Scope: which country or city, which customers, which products?
- Time and units: per day, per month, or per year, and in which currency?
You can build the formula in two directions. Top-down starts from a big known number, usually a population, and narrows it with percentages. Bottom-up starts from a small unit you can picture, such as one store on one day, and scales it up. A third view, supply-side, asks how much the sellers can deliver (seats, counters, machines) rather than how much buyers want. When you have time, estimate one way and check it roughly another way. If the two land close together, your answer is much stronger.
Assumptions depend on where you are (rough anchors; check current figures)
- People per household: about 2 in Germany, about 2.4 to 2.5 in the UK, Spain, and the US, about 3 in Singapore, about 4.5 in India, and often 5 or more among Saudi nationals.
- Passenger cars per 1,000 people: roughly 500 to 700 in Western Europe; in the US about 800 once SUVs and pickups are counted; about 110 in Singapore; and well under 100 in India.
- Big, dense cities usually have fewer cars per person than their country average, because of public transport and the cost of parking.
- Say which region your assumption is for. An assumption that is fine in Madrid can be wrong by five times in Mumbai.
Worked case
How many cars are in a Spanish city (top-down)
The prompt
Estimate the number of passenger cars registered in a Spanish city of 2,000,000 people.
Clarifying questions, with the interviewer's answers
- Do you mean cars registered in the city, or cars on its roads at one moment?Answer: Cars registered to people and businesses in the city.
- Only private cars, or also taxis, company cars, and rental cars?Answer: All passenger cars.
A hypothesis to say out loud: City households own fewer cars than rural ones, and many own none. My hypothesis is about one car per household, plus a small number of business cars.
The structure
- Cars = household cars + business cars
- Households = population / people per household
- Household cars = households x cars per household
- Business cars (taxis, company, rental) as an add-on
Working it through
1. Households
Cars are owned by households more than by individuals. Spain averages about 2.5 people per household.
Households:2,000,000 ÷ 2.5 = 800,0002. Household cars
Assume about 1.0 car per household on average, blending households with none, one, and two cars.
Household cars:800,000 × 1 = 800,0003. Add business cars
Taxis, company cars, and rental cars add roughly 5 percent.
Total cars:800,000 × 1.05 = 840,0004. Sanity-check against an outside anchor
Convert to cars per 1,000 people and compare with Spain's national figure of roughly 500. Our answer is a little lower, which fits a large city with good public transport.
Cars per 1,000 people:840,000 ÷ 2,000,000 × 1,000 = 420
The recommendation
About 840,000 cars, or about 420 per 1,000 people. First, 2,000,000 people at about 2.5 per household make about 800,000 households. Second, about one car per household plus about 5 percent business cars gives about 840,000. Third, that is about 420 cars per 1,000 people, a little below Spain's national figure of roughly 500, as expected for a big city with good public transport. The same method for a city of 2,000,000 in India would need very different inputs (about 4.5 people per household and far fewer cars per household), which is why you always say which region an assumption is for.
Risks: Cars per household is the swing assumption; 0.8 instead of 1.0 would lower the answer by a fifth.
Next steps: Check the national figure for cars per 1,000 people and adjust the city discount.
A strong candidate
Clarified what counts as a car, split households from businesses, and checked the answer against a national anchor rather than against its own assumptions.
A weak candidate
Assumed "one car for every two people" without saying for which country, and called that a sanity check.
Worked case
A Bengaluru cafe's yearly revenue (bottom-up, with a supply-side check)
The prompt
Estimate the yearly revenue of a single busy cafe in central Bengaluru.
Candidate-led: you drive the estimate and ask for the facts you need; the interviewer answers or asks you to make an assumption.
Clarifying questions, with the interviewer's answers
- One cafe, or the whole chain?Answer: One cafe.
- In-store sales only, or delivery orders too?Answer: In-store only.
A hypothesis to say out loud: A busy central cafe probably serves a few hundred customers a day, so I expect yearly revenue of a few crore rupees (1 crore is 10 million).
The structure
- Yearly revenue = customers per day x average spend x days open
- Customers per day (demand side), checked against counter capacity (supply side)
- Average spend per customer
- Days open: about 360
Working it through
1. Supply-side check on customers
Candidate: "I will assume about 400 customers a day, then check it from the supply side. How fast can the counter serve?" Interviewer: "About one customer a minute." Candidate: "And how long is it open, and how busy?" Interviewer: "14 hours a day, and busy about half the time." Candidate: "Then the counter can serve about 420 a day, close to my 400, so 400 is reasonable."
Customers the counter can serve per day:60 × 14 × 0.5 = 4202. Daily revenue
Candidate: "What does a typical customer spend?" Interviewer: "What would you assume?" Candidate: "About INR 300: a coffee and sometimes a snack."
Daily revenue (INR):400 × 300 = 120,0003. Yearly revenue
Open about 360 days a year.
Yearly revenue (INR):120,000 × 360 = 43,200,0004. In crore
Indian business figures are often given in crore: 1 crore is 10 million.
Yearly revenue (INR crore):43,200,000 ÷ 10,000,000 = 4.32
The recommendation
About INR 43 million (about INR 4.3 crore) a year. First, about 400 customers a day, a number the counter's capacity of about 420 a day supports. Second, an average spend of about INR 300 gives about INR 120,000 a day. Third, about 360 open days a year gives about INR 43 million. Checking customers two ways makes the answer much stronger than either estimate alone. If the interviewer pushes on any input, flex it and show the new answer.
Risks: Average spend could be higher if many customers buy food as well as coffee.
Next steps: Count customers for one busy hour and one quiet hour to test the 400 a day.
A strong candidate
Estimated customers two ways, used local prices and local units, and stated each assumption so it could be challenged.
A weak candidate
Picked a customer number with no check and gave the answer with no currency.
For smartphones sold per year, splitting the population by age helps, because young adults and older people own and replace phones at very different rates. For cars per household, fine age groups would add work without much accuracy. Segment where the groups behave differently, and not everywhere.
Worked case
Smartphones sold per year in a country (segmented, with replacement and first-time buyers)
The prompt
Estimate how many new smartphones are sold per year in a middle-income country of 100,000,000 people.
Assume the population splits into about 18 million children under 12 (not counted as buyers), 70 million people aged 12 to 59, and 12 million aged 60 and over.
Clarifying questions, with the interviewer's answers
- New phones only, or second-hand phones too?Answer: New phones only.
- Is this a mature market where most adults already own a smartphone?Answer: Yes, ownership is high and still growing slowly.
A hypothesis to say out loud: In a mature market most sales are replacements, so my hypothesis is that sales are roughly owners divided by the years between replacements, plus a smaller number of first-time buyers.
The structure
- Yearly sales = replacement purchases + first-time buyers
- Replacements, ages 12 to 59: owners / 3 years
- Replacements, ages 60 and over: owners / 4 years
- First-time buyers: growth in ownership
Working it through
1. Owners aged 12 to 59
Assume about 90 percent of this group owns a smartphone.
Owners aged 12 to 59:70,000,000 × 0.9 = 63,000,0002. Owners aged 60 and over
Assume about 50 percent of this group owns one.
Owners aged 60 and over:12,000,000 × 0.5 = 6,000,0003. Replacement purchases
The younger group replaces about every 3 years, the older group about every 4 years.
Replacement phones per year:63,000,000 ÷ 3 + 6,000,000 ÷ 4 = 22,500,0004. Add first-time buyers
Ownership is still growing slowly: assume about 1 million people buy their first smartphone each year.
Total phones per year:22,500,000 + 1,000,000 = 23,500,0005. Sanity-check against outside anchors
Our answer is about 0.24 phones per person per year. India buys roughly 150 million smartphones a year for about 1.4 billion people (about 0.1 per person), and the US roughly 130 million for about 340 million (about 0.4). A middle-income country with high ownership should sit between them, and it does.
Phones per person per year:23,500,000 ÷ 100,000,000 = 0.235
The recommendation
About 23.5 million phones a year. First, about 63 million owners aged 12 to 59 who replace every 3 years buy about 21 million phones. Second, about 6 million older owners who replace every 4 years add about 1.5 million. Third, about 1 million first-time buyers complete the total, which is about 0.24 phones per person, between India and the US as expected. Candidates who size only first-time buyers badly underestimate, and candidates who size only replacements miss the growth.
Risks: The replacement cycle is the swing assumption; a 3.5-year cycle for the younger group would lower sales by about 3 million.
Next steps: Test the answer with a 3.5-year cycle and state the range.
A strong candidate
Segmented by age where behavior differs, included both replacements and first-time buyers, and checked the result against two real countries.
A weak candidate
Multiplied the whole population, babies included, by one ownership rate and never checked the answer against anything.
If the interviewer says "older people replace their phones every three years too," do not defend your number. Say: "Fair, let me see how much that moves the answer." Older owners then add 2 million replacements a year instead of 1.5 million, and the total rises from 23.5 million to 24 million, about 2 percent. That also tells you which assumptions matter most: those for the large 12 to 59 group, its 90 percent ownership and above all its 3-year replacement cycle. Showing calmly how much an assumption moves the answer is the composure being tested. The drill in the next lesson checks this number.
An interviewer-led estimate with data
In interviewer-led formats, a sizing question often comes with a small table. The skill is the same, but you must use the data you are given, ask only for what is missing, and still sanity-check the answer against something outside the table.
Worked case
Home EV chargers sold per year in Australia (interviewer-led, with data)
The prompt
An Australian maker of home electric-vehicle (EV) chargers asks: in five years, how many home chargers will be sold in Australia each year, and what will that be worth? Use the household table below. A charger costs the household about AUD 1,800 installed.
Interviewer-led: the interviewer gives the household table, answers your clarifying questions with the ownership assumptions, and asks for the estimate step by step.
Clarifying questions, with the interviewer's answers
- New installations only, or replacements too?Answer: Both. A home charger lasts about 10 years.
- Can people in apartments install one?Answer: Only if they have a private parking space. The table shows the share of each dwelling type with off-street parking.
- What EV ownership should I assume?Answer: In five years, about 12 percent of households with off-street parking own an EV, and that share is rising by about 2.5 points a year. About 80 percent of those owners install a home charger.
A hypothesis to say out loud: A home charger needs off-street parking, so my hypothesis is that detached houses drive almost all demand, and that yearly sales in five years are a few hundred thousand units, worth a few hundred million Australian dollars.
The structure
- Yearly chargers = new installs + replacements
- Households that can install: households x share with off-street parking, by dwelling type
- New installs: rise in EV ownership that year x install rate
- Replacements: installed chargers / 10-year life
- Value: units x AUD 1,800
The exhibit
| Dwelling type | Households (millions) | With off-street parking (%) |
|---|---|---|
| Detached house | 7 | 90 |
| Townhouse or terrace | 1.4 | 60 |
| Apartment | 1.6 | 20 |
Working it through
1. Households that can install
Read the table: 90 percent of 7.0 million detached houses, 60 percent of 1.4 million townhouses, and 20 percent of 1.6 million apartments have off-street parking. Detached houses are about 85 percent of the total, which supports the hypothesis.
Households with off-street parking (millions):7 × 0.9 + 1.4 × 0.6 + 1.6 × 0.2 = 7.462. Installed chargers in five years
12 percent of those households own an EV, and 80 percent of them install a charger.
Installed home chargers (millions):7.46 × 0.12 × 0.8 = 0.71623. New installs that year
Ownership rises about 2.5 points a year, and 80 percent of the new EV households install.
New installs per year (millions):7.46 × 0.025 × 0.8 = 0.14924. Replacements
The installed base divided by a 10-year life. This is a simplification: the base is growing fast, so in five years most chargers will still be young and fewer will need replacing. Treat about 70,000 as an upper figure.
Replacements per year (millions):0.71616 ÷ 10 = 0.071625. Total units
New installs plus replacements: about 220,000 chargers a year.
Chargers per year (millions):0.1492 + 0.071616 = 0.22086. Value
About 220,800 chargers at AUD 1,800 each is about AUD 400 million a year.
Market value (AUD a year):220,816 × 1,800 = 397,468,8007. Sanity-check against an outside anchor
Australia sells roughly 1.2 million new cars a year. The new EV households implied here are about 187,000 a year, or about 16 percent of new-car sales. That is a plausible EV share five years out, so the estimate is not far off.
Implied new EV households as a share of new cars (%):7.46 × 0.025 ÷ 1.2 × 100 = 15.54
What the exhibit shows
Detached houses hold about 6.3 million of the 7.46 million households that can install a charger, so they drive the market.
The recommendation
About 220,000 home chargers a year in five years, worth about AUD 400 million a year at AUD 1,800 installed. First, about 7.5 million households can install a charger, most of them in detached houses. Second, new EV households add about 150,000 installs a year. Third, replacements add up to about 70,000 a year; because most chargers will still be young, the true number is likely lower, so the total sits between about 150,000 and 220,000 units. The biggest uncertainty is the 12 percent ownership figure. If it is 10 percent, the installed base and replacements fall by a sixth, so the next step is to test the answer with a low and a high ownership case.
Risks: EV ownership may grow faster or slower than 2.5 points a year; Apartment buildings may add shared chargers, which changes who buys; Replacements may be well below 70,000 a year while most chargers are still young.
Next steps: Rerun the estimate at 10 and 15 percent ownership; Check the install rate with two electricians who fit chargers.
A strong candidate
Used the table instead of inventing numbers, split new installs from replacements, and checked the implied EV sales against new-car sales, an anchor outside the table.
A weak candidate
Multiplied all 10 million households by the EV share, ignoring that apartments without parking cannot install, and forgot replacements.
Common mistakes
- Starting to multiply before clarifying what is being estimated.
- Using one assumption, such as household size, for every country.
- A "sanity check" that only repeats your own assumptions.
- Giving an answer with no unit or currency.
- False precision: say "about 23.5 million," not "23,456,789."
- Mixing up stock (how many exist) and flow (how many are bought each year).
- Ignoring data the interviewer gave you and inventing your own numbers instead.
You are asked to estimate dishwashers in UK homes. Which clarifying question matters most before you build anything?
Your top-down estimate is 2 million units and your bottom-up check gives 9 million. What is the best next move?
An estimate for a city in India uses 2.5 people per household. What is the most likely problem?
Sources for this lesson (1)
- Recognized public explanations of case-interview concepts and frameworks
My notes on this lesson
0 of 5,000 characters. Saves automatically.
Try the 3 remaining checks and drills above to complete this lesson (0 of 3 done).
Spotted something wrong or out of date? Report a mistake. We check every report and correct the page.