Market Sizing: how to solve market estimation questions

Market sizing questions appear in almost every case interview. Sometimes as a complete case. Sometimes as part of a broader business case. And almost always, candidates don't know where to start.

The good news: market sizing has a pattern. The bad news: that pattern is not what you think.

In 300+ interviews at Bain & Company, I saw candidates who knew many formulas but failed at market sizing. And I saw candidates who barely knew numbers but solved it with structured thinking. The difference is not the accuracy of the final number. It's how you get to that number.

This article teaches you exactly how to think through a market sizing, how to structure your analysis, and how to answer with confidence when the interviewer asks: "How many taxis are there in Spain?" or "What is the market for coffee shops in Buenos Aires?"

What you'll find here are not formulas. They are patterns. They are practical exercises. And it's the methodology I used as an evaluator at Bain.

Mistake #1: Thinking Market Sizing is a Formula

The most common mistake I see is that candidates try to solve market sizing with 3 variables: population × penetration rate × per capita consumption = market.

That works in some cases. But if you apply it the same way to all market sizings, you'll arrive at completely wrong numbers.

Real example: "How many cups of coffee are sold daily in Paris?"

Candidate using the formula quickly:

• ●Population: 2.2 million
• ●Coffee consumers: 70% = 1.54 million
• ●Cups per consumer per day: 2
• ●Total: 3.08 million cups

Sounds good. But a critical fact got lost: where these cups are consumed. At home, in a coffee shop, in an office. Because the market we're looking for might be only coffee shops, not home consumption.

Candidate who thinks about structure:

Starts by asking: "Are we talking about the total coffee market or just the coffee shop market?" If it's coffee shops, then the analysis is different: I need to think about how many people buy coffee in shops, not at home.

• ●Paris population: 2.2M
• ●Average daily tourists: +0.5M (Paris is touristy)
• ●Total "potential consumers": 2.7M
• ●Segment buying at coffee shops regularly: 45% = 1.2M
• ●Frequency: 3 coffees per week per person (some buy daily, others don't)
• ●Per capita average: ~0.4 coffees daily at shops
• ●Total: 2.7M × 0.4 = ~1.08 million cups in shops daily

The number is lower. But the logic is clear. And it's defendable. Because the candidate explained each step. And each step has a reason.

The difference is crucial: the first candidate applied a formula. The second candidate built a structure. And that's what evaluators at McKinsey, BCG, and Bain look for.

And if you learn better by practicing, download our market sizing exercises for free and come back to this article with the method fresh.

Two Approaches: Top-Down and Bottom-Up

Before building any market sizing, you need to choose an approach. There are two: top-down and bottom-up. Almost all questions can be solved with both. And usually it's better to start with one, then validate with the other.

Top-Down: From General to Specific

You start with a large number and break it down.

Example: "How many taxis are there in New York?"

Step 1: New York population: 8.3 million.

Step 2: People who use taxis regularly: 40% = 3.3 million.

Step 3: How many people per taxi can be supported? If each taxi has capacity for 4 people, it needs to make trips all day to cover demand.

Step 4: Assuming 10 effective trips per day per taxi, and each trip has 2 people on average, one taxi covers 20 people per day.

Step 5: 3.3 million ÷ 20 people per taxi = 165,000 taxis.

This is a top-down analysis. You started with population and broke down to taxi demand.

Bottom-Up: From Specific to General

You start with a small number (one unit, one segment) and scale it up.

Example: "How many taxis are there in New York?"

Step 1: In a neighborhood like Manhattan, there are 1 million people.

Step 2: In a neighborhood of that density, I estimate there are 5,000 taxis.

Step 3: New York has multiple neighborhoods with density similar to Manhattan: ~33 neighborhoods.

Step 4: But not all have equal density. Manhattan is the densest. On average, I adjust the factor to 2.5x.

Step 5: 5,000 × 33 × 2.5 = 412,500 taxis.

This is a bottom-up analysis. You started with one neighborhood and scaled up to the entire city.

Which is correct? Both should give similar numbers if done well. If they differ significantly, something in your logic is wrong. That's good. It lets you review.

Good candidates do both analyses or at least announce they could. They tell the interviewer: "I'm going to use a top-down approach, but afterward I can validate with bottom-up if you'd like." That shows flexibility and rigor.

The 6-Step Methodology

There is a methodology that works in 95% of market sizings. I learned it at Bain and have used it with candidates. It works because it's systematic, it's defensible, and it leaves nothing to chance.

Step 1: Pause and Structure

When you hear the question, don't jump to numbers. Pause. Breathe. Think about what you need to know.

"How many gyms are active in Spain?"

Your mind should process it like this: "OK, I need to know how many people there are potentially, how many go to gyms, and of those, what's the penetration of active memberships."

It's not complex. It's structured.

Step 2: Define the Approach

Top-down or bottom-up. Or both. Communicate which one you'll use.

"I'm going to use a top-down approach. I start with population, segment by age, and estimate what percentage has an active gym membership."

This takes 20 seconds. But it tells the interviewer you have a plan. That builds confidence.

Step 3: Build the Issue Tree (MECE)

The issue tree for a market sizing is simpler than in other cases, but it's still MECE.

For "How many gym memberships are there in Spain?" your tree could be:

```

Active memberships in Spain

├── By age segment

│ ├── Young (15-30 years)

│ ├── Adults (30-50 years)

│ └── Older (50-65 years)

├── By gym type

│ ├── Premium (chains)

│ ├── Budget (small gyms)

│ └── Community

└── By activity rate (active vs inactive membership)

```

Each branch is different. Together they cover the entire market.

Step 4: Make Assumptions

This is where many candidates get lost. They think making assumptions is "guessing badly." Wrong. Assumptions are explicit statements of what you're assuming because you don't have exact data. They are defensible.

Example of a good assumption: "I assume Spain's population is 47 million because it's data I should know."

Example of a bad assumption: "I assume 80% of the population goes to the gym because it sounds right."

Assumptions must be reasonable. They must be based on logic or on data you do know.

Step 5: Do the Math

This is where you write the numbers and multiply them. But do it step by step. Don't skip steps. If one step has an important number, express it. That way the interviewer can question you on that specific point, not on the entire conclusion.

Step 6: Sense-Check

At the end, ask yourself: "Does this number make sense?"

If you calculate there are 5 million gym memberships in Spain, but Spain has 47 million people, that means 10.6% of the population has a membership. Sounds high but possible. It's defensible. If you calculated 30 million, it's indefensible. That's 64% of the population. Nobody believes that.

The sense-check is your safety net. It's where intuition helps you avoid big mistakes.

Practical Exercise: Gym Market Sizing in Spain

I'm going to give you a complete market sizing solved step by step. This is exactly the type of question you hear in interviews. Afterward, you'll solve a similar one.

The question: "How many gym memberships are active in Spain today?"

Step 1: Pause and Structure

OK. Spain. Gym. Active memberships. I need to know: population, age, gym penetration rate, active vs inactive membership rate.

Step 2: Define the Approach

I'm going to use top-down. I start with population, segment by age (because different ages have different gym rates), and for each segment I calculate the penetration of active memberships.

Step 3: Issue Tree (MECE)

```

Active Gym Memberships in Spain

├── Adult population (15-65 years)

├── Segmented by age

│ ├── Young (15-30): penetration rate X%

│ ├── Adults (30-50): penetration rate Y%

│ └── Older (50-65): penetration rate Z%

└── Adjustment for inactive memberships (subtract)

```

Step 4: Assumptions and Data

• ●Spain population: 47 million (this is data I should know)
• ●Adult population (15-65): ~65% of total = 30.55 million
• ●Age segmentation:
• ●Young (15-30): 20% of adult population = 6.11 million
• ●Adults (30-50): 35% of adult population = 10.69 million
• ●Older (50-65): 10% of adult population = 3.06 million

(Total: 65% of 47M = ~30.55M, check)

• ●Gym penetration rate (active membership):
• ●Young (15-30): 18% (they go most, but many only in summer)
• ●Adults (30-50): 12% (they work more, less time)
• ●Older (50-65): 5% (less interest, more injuries)

Step 5: The Math

| Segment | Population | Rate | Memberships |

|----------|-----------|------|-----------|

| Young (15-30) | 6.11M | 18% | 1.10M |

| Adults (30-50) | 10.69M | 12% | 1.28M |

| Older (50-65) | 3.06M | 5% | 0.15M |

| TOTAL | 20M | 13% | 2.53M |

(Note: the 20M is the gym-age population I considered. The 13% is the weighted average.)

But wait. I need to consider that some people with memberships pay but don't use them. "Active membership" could mean "valid membership with usage in the last month."

Adjustment for inactive memberships: I assume 15% of people with a membership pay but don't use it regularly.

• ●Active memberships: 2.53M × 85% = 2.15 million

Step 6: Sense-Check

Spain: 47 million people. Active gym memberships: 2.15 million. That's 4.57% of total population, or 10.75% of the adult population I considered (20M).

Does it make sense? Yes. In Spain, gym culture is important in big cities, but rural areas have less penetration. In major cities like Madrid or Barcelona, the rate is higher. In small towns, almost nothing. An average of 10.75% of active adults in gyms is reasonable.

Quick comparison? In the US, penetration is higher, around 19% of population. In Spain, less penetration, 10-12% is believable. My estimate of 10.75% is consistent. Good.

Conclusion: There are approximately 2.15 million active gym memberships in Spain.

Now your turn. Solve this:

Your market sizing: "How many hotel nights are sold annually in Madrid?"

Tips:

• ●Think about segments: tourists vs business travelers vs others
• ●Consider that Madrid is an important tourist city
• ●One hotel night = one room × one night
• ●Sense-check: How many hotels are there approximately in Madrid? If I have X hotels with Y average rooms, and Z% occupancy, does that give me a similar number?

Take 15 minutes. Do it step by step. Then compare with your intuition.

Structure, Not Exact Number

Here comes an insight I learned evaluating candidates at Bain:

Structure always beats the exact number.

I've seen candidates who arrived at incredibly exact numbers, but when I asked "why?", they had no justification. "I read an analysis yesterday that said..." That doesn't count.

I've seen candidates who arrived at a number that was 40% below the real one, but whose structure was flawless. "I assume X because Y. I calculate Z. Sense-check: A." Those candidates passed. Because they showed thinking. Because the evaluator could see where to reason differently if needed.

The real example that convinces me:

A candidate arrives at a market sizing: "Imagine market sizing, how many taxis are there in Spain?" The candidate gives me the exact number. And that, by itself, is not a reason to pass. They say: "I looked at an analysis yesterday, let's say 60,000 taxis." That doesn't count.

But if the candidate tells me: "Spain has 47 million people. The main cities (Madrid, Barcelona, Valencia, Seville, Bilbao) have 30% of the population. In those cities, taxi density is high. In towns, almost nothing. In cities, I assume 1 taxi per 500 people. That gives me..." Then I'm interested. Because I'm seeing structure. Because I see the person is thinking.

What we care about is how you think, not the exact number.

Data Every Candidate Should Know

There are some numbers you're expected to know. They are general business data. It's not memorizing an encyclopedia. It's having a foundation of basic knowledge.

Populations of major countries:

• ●World: ~8 billion
• ●Latin America: ~650 million
• ●Mexico: ~130 million
• ●Brazil: ~215 million
• ●Colombia: ~52 million
• ●Argentina: ~46 million
• ●Chile: ~19 million
• ●Peru: ~34 million
• ●Spain: ~47 million
• ●USA: ~340 million
• ●China: ~1.4 billion
• ●India: ~1.4 billion

Typical age segmentation:

• ●0-14 years: ~15-20% of population
• ●15-64 years: ~65-70% of population (active population)
• ●65+ years: ~15-20% of population

Common consumption rates:

• ●Internet penetration in Latin America: 60-70%
• ●Smartphone penetration: 70-80%
• ●Coffee shop consumption in major cities: 3-5 coffees per person per week

Do you need to memorize this? No. But having it handy speeds up your analysis. In a high-pressure interview, having these numbers is the difference between "it takes me 2 minutes to structure" and "it takes me 10."

Most Common Mistakes (and How to Avoid Them)

Mistake #1: Forgetting segmentation.

Candidate: "Spain has 47 million people. 15% have gym memberships. Total: 7 million."

Problem: Doesn't consider that penetration is different by age. In young people it's 25%, in older it's 2%. The average is not 15%.

How to avoid: Always ask yourself: "Is this number the same for all groups? Or does it vary?" If it varies, segment.

Mistake #2: Confusing demand with supply.

Candidate: "There are 1,000 pizzerias in Madrid. Each sells 100 pizzas a day. Total: 100,000 pizzas daily."

Problem: That's supply (how many pizzerias and their capacity). It's not demand (how many people want pizza).

How to avoid: Define whether you're calculating demand (how much do people want?) or supply (how many businesses are there?). They're different directions.

Mistake #3: Not adjusting for saturation or seasonality.

Candidate: "1 million people in Madrid. 20% eat at restaurants each week. 52 weeks a year. Total: 10 million restaurant meals per year."

Problem: Didn't consider that the same person can go many times. You're counting the same people multiple times. Or if you did it correctly, you didn't make it explicit.

How to avoid: Be explicit. "Each person who eats at restaurants does so 20 times per year on average. Population: 1M. Penetration rate: 50%. Total: 1M × 50% × 20 = 10 million occasions per year." Now it's clear.

Mistake #4: Exact number without structure.

Candidate: "There are 150,000 hotels in the world."

Problem: Where did you get that? How did you get there? If I can't validate your logic, I don't trust the number.

How to avoid: Always justify. "I assume the world has 200 countries. On average, each country has 500 cities relevant for tourism. Each city has 100-150 hotels. That gives me approximately..."

Mistake #5: Assumptions that aren't credible.

Candidate: "I assume 80% of the population buys online."

Problem: In 2026, that's too high in many regions. In some yes. But if you don't specify where, it's not defensible.

How to avoid: Reasonable assumptions. "I assume that in major cities, ecommerce penetration is 60-70%. In mid-size cities, 30-40%. In rural, 10-15%. Spain is 80% urban, so..." Now the assumption is credible.

Frequently Asked Questions

Q: What if my calculated number is very wrong compared to reality?

A: It depends on what "very wrong" means. If you calculated 1 million and reality is 1.2 million, that's excellent (20% error is a good range). If you calculated 1 million and reality is 10 million, you have a structure problem. The evaluator probably gives you a hint: "Are you sure?" That's where you review your logic. Be flexible. Adjust. Don't die on the hill.

Q: Is top-down or bottom-up better?

A: Both are valid. Best is to do one, announce that you could validate with the other, and if time permits, do it. Good candidates use both. But if you have to choose one in 15 minutes, use whichever feels more natural.

Q: Can I use a calculator?

A: Some interviews allow it, others don't. Ask before starting. If not, that's fine. Simple math (multiplication, division) is expected to be done in your head or on paper. If the number is very complex, the evaluator expects you to say: "This calculation is complex. In a real situation, I'd use Excel. But the reasoning is X."

Q: What data can I assume vs what should I ask for?

A: Populations of major countries: assume. Specific industry penetration rates: ask or explicitly assume. Data about a specific client: ask. Macroeconomic data (inflation, growth rates): you can assume or ask, but declare it.

Q: Does market sizing appear in strategy or finance cases?

A: Both. In strategy it's more common. "What's the potential market?" In finance, it can appear to estimate future revenue. In both cases, the methodology is the same.

Q: How much time should I spend on market sizing?

A: If it's a complete case, 20-30 minutes. If it's part of a larger case, 10-15 minutes. It depends on how deep the evaluator wants to go.

Q: If I make a mistake in multiplication, is it automatic failure?

A: No. Arithmetic errors happen. What matters is that you catch it quickly. If the interviewer says "check that number," check it. If you see it yourself, fix it. The evaluator focuses on thinking, not on whether you did the math perfectly.

Market Sizing Patterns by Industry

Different industries have different patterns. Learning to recognize them speeds you up.

Restaurants/Coffee shops:

Typical structure: Population × penetration (how many visit) × frequency (how many times per year) = consumption. Then divide by market capacity (how many people can a restaurant serve per year).

Retail:

Typical structure: Number of stores × average ticket × transactions per store per day × 365 days = annual revenue. Or: population × penetration × average annual spending = total market.

Technology/Apps:

Typical structure: Potential users × adoption rate × revenue per user = market size. (Watch out: adoption rate is commonly low in year 1, increases over time.)

Hospitals/Healthcare:

Typical structure: Population × disease rate × visit frequency = demand. Then divide by capacity (how many patients a hospital serves per year).

Learn these patterns. When you hear the question, your brain will recognize the pattern and activate the right structure.

Next Step: Structured Practice

Market sizing is practiced. Not studied. Here's the plan:

1. 1Week 1: Do 5 simple market sizings (population × rate = result). Time yourself. Target: 10 minutes per market sizing. Check if your number makes logical sense.

1. 2Week 2: Do 5 more complex market sizings (multiple segments, top-down AND bottom-up). Target: 15 minutes.

1. 3Week 3: Have someone ask you follow-up questions. "What's your hypothesis?" "Why that number?" Listen to the feedback.

1. 4Week 4: Record yourself doing market sizing. Listen to yourself. Is the structure clear? Do you state the assumptions? Or just give numbers?

This is the training that works. It's not memorization. It's deliberate practice.

Related Resources

Dive deeper into related topics:

• ●Types of cases you'll see in MBB interviews
• ●MECE Framework: the tool behind every case
• ●Issue trees: how to build the tree that solves the case
• ●Math in cases: quick calculations without a calculator
• ●How to practice cases alone: step-by-step guide

And always check the complete preparation plan to understand where market sizing fits.

Books and Resources

For exhaustive practice, the book Crack The Frameworks has 30+ market sizing exercises. And Crack The Case Interview covers specific estimation techniques with cases designed with MBB methodology.

30 exercises: coffee shops in Paris, taxis in New York, gym memberships in Madrid, hotels in Bangkok, electric cars in California, weddings in Mexico, and much more.

Each exercise includes step-by-step solution. Use one as a guide, but practice first without seeing the answer.

Download the books: nextepmbb.com/en/resources. Also available in digital format on the Prep Platform.