A market sizing case asks you to estimate how large a market is — in revenue, units sold or number of customers — using only facts you could reasonably be expected to know, plus arithmetic. There is no figure to recall and no data will be handed to you. The interviewer is assessing how you break the problem into estimable parts, how you justify each assumption, and whether you know which assumption the answer actually depends on.
It is the most common opening format in consulting interviews and the one most exposed to practice, because the structure is transferable even when the subject is not.
Define the market before you estimate it
More market sizing answers fail here than anywhere else, and they fail invisibly: the candidate estimates a real market competently, and it is not the market that was asked about.
Four things need settling before any arithmetic:
| Dimension | Question to ask | Why it changes the answer |
|---|---|---|
| Geography | Which country or region? | Population and income assumptions rest on it |
| Unit | Revenue, units, or customers? | Revenue needs a price; units do not |
| Buyer | Consumer, business, or public sector? | Different populations and purchase logic |
| Period | Annual, or a stock at a point in time? | Annual sales and installed base differ by an order of magnitude |
The stock-versus-flow distinction in the last row catches candidates repeatedly. “The market for dishwashers” can mean the number of dishwashers in use or the number sold this year. Those differ by roughly the product’s lifetime — perhaps tenfold. Ask which is meant.
State your definition back before starting: “I will estimate annual retail revenue from dishwasher sales to households in Canada. Is that the right frame?” Fifteen seconds, and it removes the failure mode entirely.
Choose top-down or bottom-up, and say why
Top-down starts from a large known quantity — usually a population — and narrows it with successive filters until you reach the buyers, then multiplies by frequency and price. It suits consumer markets, where population is a solid anchor.
Bottom-up starts from a single unit of supply — one store, one machine, one route — estimates its throughput, and multiplies by the number of such units. It suits questions about a specific location, channel or asset, and it is usually the better choice when the client owns something you can count.
Say which you are using and why:
“I will work top-down from the Canadian population, because coffee is bought by nearly everyone and population is the most reliable number I have. If you would rather I sized it from the number of outlets, I can do that instead.”
That sentence signals that the choice was deliberate. Announcing a structure without justifying it is the difference between a candidate who has practised the format and one who has memorised it.
A worked top-down estimate
Prompt: estimate the annual retail market for coffee sold in cafés in Canada.
Canada’s population is roughly 40 million. I will hold that as the anchor and filter down.
- Population: 40 million.
- Age filter. Café coffee is bought mainly by adults. Take those aged 18 to 70 as roughly 65 per cent of the population — Canada’s age distribution is broad with a substantial older cohort, so this is a moderate estimate. That gives 26 million.
- Participation. Not every adult buys café coffee. I will assume 60 per cent do at least occasionally, which reflects both non-drinkers and people who drink coffee only at home. That gives 15.6 million, call it 16 million.
- Frequency. This is the assumption the answer is most sensitive to, so I will split it rather than take an average. Say a quarter are daily buyers at roughly 250 cups a year, and three-quarters are occasional at roughly 40 cups a year.
- Daily: 4 million × 250 = 1.0 billion cups
- Occasional: 12 million × 40 = 0.48 billion cups
- Total: roughly 1.5 billion cups a year
- Price. An average café coffee, blending a basic drip coffee with more expensive espresso drinks, at around $4.
- Market size: 1.5 billion × $4 ≈ $6 billion a year.
Sanity check. That is about $150 per Canadian per year, or roughly $3 a week averaged across the whole population including children and non-drinkers. For a country where café coffee is an everyday purchase, that is plausible. If it had come out at $1,500 per person I would have gone back and checked step 4.
What would move it most. Frequency, by a wide margin. If daily buyers average 300 cups rather than 250, the total rises by about 13 per cent. Price is the second lever, and it is the one I would most want real data on, since the mix between $2 drip coffee and $6 speciality drinks drives it entirely.
That last paragraph is the part candidates most often omit and interviewers most reliably reward.
A worked bottom-up estimate
Prompt: estimate annual revenue for a single urban grocery store.
Top-down would be wrong here — there is no population filter that lands on one store.
- Opening hours: 14 hours a day, 360 days a year.
- Customers per hour: varies by daypart, so split it. Roughly 4 peak hours at 200 customers, and 10 off-peak hours at 60. That is 800 + 600 = 1,400 customers a day.
- Average basket: $45, blending large weekly shops with small top-up visits.
- Daily revenue: 1,400 × $45 = $63,000.
- Annual revenue: $63,000 × 360 ≈ $22.7 million.
Sanity check. Roughly 1,400 customers through 8 to 10 checkouts over 14 hours is about 10 to 12 customers per checkout per hour, or five to six minutes each. That is realistic. Had the arithmetic implied one customer per checkout per minute, the customer estimate would need revisiting.
Structuring the assumptions so errors do not compound
Two techniques materially improve accuracy, and both are visible to the interviewer as technique.
Segment where behaviour genuinely differs. The coffee estimate split daily from occasional buyers rather than applying one average frequency. Averages across genuinely different groups are where estimates go wrong, because the mean of a bimodal distribution describes nobody.
Prefer fewer, larger filters. Each filter carries an error. Five filters at ten per cent uncertainty each compound into something much wider than one filter at twenty. If two steps can be combined into one defensible assumption, combine them and say so.
Common mistakes
| Mistake | What it looks like | Fix |
|---|---|---|
| Sizing an undefined market | Answer describes a different market from the question | Define geography, unit, buyer and period first |
| Stock and flow confused | Annual sales estimated as installed base, or the reverse | Ask which is wanted; divide by product lifetime to convert |
| Unjustified assumptions | “I will assume 30 per cent” with no reason | Give the number and its justification in one breath |
| Averages across different groups | One frequency for daily and occasional buyers | Segment where behaviour differs |
| Arithmetic without labels | Zeros lost between steps | Fix the unit up front and label every figure |
| Stopping at the number | “So, six billion dollars.” | Sanity-check, then say what would move it |
What good sounds like
The strongest market sizing answers share a shape. The candidate defines the market and confirms it. They name the approach and say why it fits. They build in clearly numbered steps, each with a stated assumption and a one-clause justification. They flag the sensitive assumption while passing it, not afterwards. They check the result against something known. And they close by saying what they would want data on first.
None of that requires the number to be right. It requires the reasoning to be visible, which is a different and far more practisable skill.
Practising the format
Market sizing rewards repetition more than any other case type, because the structures recur. Estimate something every day: the annual market for bicycles in your city, the number of commercial flights over your country right now, the revenue of the café you are sitting in. Spend two minutes structuring and one minute calculating, out loud, on paper.
Then check the two things that matter. Was every assumption justified as you stated it? Did you say what would most change the answer? Those are the two habits that transfer to the interview.
The arithmetic underneath is covered in the case maths techniques guide, and market sizing frequently opens a larger case — most often a market entry case, where the size of the opportunity determines whether entry is worth analysing at all. For where this sits in a full preparation plan, see the case interview hub.
