Market Sizing and Guesstimates: The Complete Guide (2026)
"How many cars are sold in the US each year?" "How many golf balls fit in a Boeing 747?" If you have started preparing for consulting interviews, you have met the market sizing question, and you have probably been told to drill hundreds of them.
So let us answer the obvious question directly. Do you need to be good at market sizing to get into McKinsey, BCG, or Bain? Yes. A rough one, done cleanly, is a genuine skill the firms test. Do you need to grind a hundred standalone guesstimates until you can estimate anything on this earth? No. At MBB, a pure market sizing question is rarely asked on its own. And it is not an essential skill on which you are assessed on. The skill matters, but it almost always shows up woven inside a normal case, not as a separate puzzle.
We spent 6+ years each at McKinsey, as consultants and later as interviewers scoring cases. This guide covers what market sizing actually is, where it really appears in interviews, the method for sizing any market, a full worked example, and the handful of facts you genuinely need to know.
What Market Sizing Actually Is
A market sizing question asks you to estimate a number you cannot look up: the size of a market, the volume sold in a year, the quantity of something in a place. The question tests one ability: Can you derive a reasonable estimate for a number (e.g., units sold) using a logical approach, breaking the problem into smaller pieces you can reason about.
That is the whole game. Nobody expects the right answer to three decimal places. As we tell every candidate, the process matters more than the end result. A clean structure with sensible assumptions beats a lucky number every time.
Where Market Sizing Really Shows Up In The Case Interview
Here is the thing most prep material gets wrong. It treats market sizing as its own exam, worth hundreds of dedicated reps. In reality, a standalone "size this market" question is one of the least common formats at the top firms.
Market sizing appears in roughly a quarter of first round MBB interviews, but usually inside a bigger case, not as a separate stand-alone question. Most often, it is a sub-question when the interviewer wants to push you. Or it turns up as the maths part of a market entry case ("before we decide whether to enter, how big is this market?"), or as one calculation inside a broader problem. McKinsey, running an interviewer-led format, will rarely ever hand you a pure market sizing question and nothing else. Bain leans on explicit market sizing a little more than the others, but even there it is often one step in a longer case.
And most importantly, it’s always in the context of a case. A question like “How many golf balls fit in a school bus?” will never come up.
Think back to a normal case. Our own renewable energy case asks the candidate to estimate the wind and solar potential of an island from a handful of data points. That is market sizing. It is just not labelled as such, and it does not arrive as a cute brain teaser. It is the same muscle (i.e., structure the number and break it down into pieces, estimate the inputs, sanity check the result), used in service of an actual business decision.
The takeaway: build the skill, but keep it in proportion. The candidate who has done 200 golf-ball guesstimates and 5 full cases is much worse prepared than the one who has done 30 full cases, because the second one has practised market sizing every time a case needed it implicitly, alongside everything else that actually gets tested.
Why Market Sizing Still Matters
If it is rarely a standalone question, why bother at all? Because the skills underneath it are exactly the skills a case tests everywhere else.
A good market sizing forces you to structure a problem logically (the same skill as building an issue tree), do quick conceptual maths without a calculator, make sensible business assumptions, and land on a clear answer. Those are four of the core things interviewers score. So practising market sizing is really practising the fundamentals in a compact form. That is why it is worth doing, and also why you do not need to prepare separately from your case prep.
How to Size Any Market: The Method
Here is the repeatable approach. It works for a market, a guesstimate, or a sizing step buried inside a case.
1. Clarify what you are estimating.
Pin down the exact number, the unit, the geography, and the timeframe before you touch any maths. "Cars sold in the US" means new cars, per year, in the United States. Getting this wrong is the fastest way to waste four minutes on the wrong number.
2. Pick top-down or bottom-up.
Top-down starts from a big anchor (usually a population figure) and filters down by who actually buys and how often. It is the most common approach and the right default for consumer markets. Bottom-up starts from a single unit (one store, one customer, one machine) and scales up. A strong move is to name your choice out loud: "I will go top-down, and if useful I can sanity check bottom-up afterwards." That signals flexibility and rigour.
3. Build the equation.
Break the number into a short chain of drivers you can actually estimate, and segment only where it changes the answer. Keep it MECE, but keep it simple. Two or three levels is usually plenty.
4. Plug in the few numbers you must know, estimate the rest.
Some inputs you can reason out (what share of people own a car). One or two you simply have to know (the population of the US). For everything you estimate, sanity check it against your own life. You know roughly how many people around you own a car, so your assumption should not be wildly off.
5. Round hard, then give the "so what."
Use round numbers throughout so the mental maths stays clean and easy for you, then finish with a one-sentence read of the result. Never recite the number and stop. Say what it means, or whether it looks sensible, the way a consultant would.
Worked Example: Cars Sold in the US Each Year
Let us run the classic. The goal is new cars sold in the US per year, and we will go top-down.
Start with the population of the US, which is about 340 million. This is a number you simply have to know, at least roughly. If you say 10 million you are hopelessly low, if you say a billion you are hopelessly high, but 340 million or around 300 million is fine.
Now, what share of people own a car? Children under 16 do not. Many people in big cities do not need one. Some households share. Some elderly people no longer drive. Call it half. That gives roughly 170 million cars on the road.
But those are not all bought in a single year. They get replaced over the life of a car, which is maybe 10 years. So the cars sold each year are roughly the total on the road divided by the average lifetime:
170 million ÷ 10 ≈ about 17 million new cars a year.
Now the "so what": that looks sensible, and it is. Actual US new vehicle sales run around 15 to 16 million a year. Every input we used was a little rough, and the answer still landed in the right ballpark. That is the whole point. A clean structure with honest, rounded assumptions gets you a defensible number, even when none of the inputs is exact.
The Handful of Facts You Actually Need to Know
Logic only takes you so far. At some point you have to plug in a real number, and a few of these you cannot deduce. The good news is the list is short.
The essentials: world population (about 8 billion), the US (about 340 million), China and India (about 1.4 billion each), the country you are interviewing in, and roughly the other big economies near you. On top of that, keep a rough sense of a few exchange rates, so you know whether a price sounds high or low, and a basic feel for metric versus imperial (what a mile is, what a kilogram is).
You do not need the population of every nation, and you do not need to memorise Wikipedia. If a figure is genuinely obscure, it is completely fair to ask the interviewer. The only real risk is being clueless about the basics, because that reads as a gap in general education rather than a gap in case technique. Stay observant, read a bit of business news, and this takes care of itself.
Common Mistakes
Stopping at the number.
The most common soft fail. You do the maths, recite the figure, and go quiet, leaving the interviewer to supply the meaning. Always close with a one-sentence read of the result.
False precision.
Carrying 337.2 million through the maths in your head. Round to 340 million and move faster with fewer errors. The small rounding difference won’t make a huge difference in the final number.
Not segmenting where it matters, or segmenting where it does not.
If splitting by age or geography changes the answer, do it. If it does not, do not clutter the structure to look clever.
Skipping the sanity check.
If your method says 500 million cars are sold in the US each year, that is more than one per person per year (incl. children), which cannot be right. Catching that in the room is a strong signal.
Panicking on a fact you do not know.
If you blank on an industry or a figure, do not blush and apologise. Stay calm, offer what you do know, and ask a smart, specific question. Handling a knowledge gap with composure is itself a skill the job demands.
How to Practise (In the Right Proportion)
Do a handful of standalone market sizings first, enough to get comfortable with the five steps and to feel the maths under a little pressure. Practise them out loud, talking an imaginary interviewer through your structure. In the actual interview you would also walk the interviewer through your logic. This allows the interviewer to intervene and assist you and it also helps you to practice explaining your logic. You don’t want to fail an interview because your explanations are clumsy.
Then stop drilling them in isolation and fold the skill into full case practice, where it actually lives. Every time a case needs a number estimated, you get another rep, in context, alongside structuring, data reading, and communication. Learn your handful of key facts cold. That is a far better use of your prep hours than a hundredth golf-ball question.
The Bottom Line
Market sizing is a real skill, and you should be able to do a rough one cleanly: clarify the number, break it into drivers, plug in the few facts you must know, estimate the rest, round hard, and finish with a "so what." But keep it in proportion. At MBB it is rarely a standalone question. It shows up inside normal cases, so the best way to get good at it is to practise full cases, not to grind guesstimates in a vacuum.
The strongest way to build this is on realistic cases where market sizing appears the way it does in a real interview, embedded and in context, with feedback on what a top candidate would have done. Our Case Interview Mastery course on Udemy gives you 7 full McKinsey-style cases, each with a do-it-yourself version, a model solution, and a feedback version where we explain every move. Taught by us, two former McKinsey interviewers.
