What the interviewer is actually testing
A market sizing question, like 'how many coffee shops are there in your city' or 'what's the annual market size for pet food in this country', isn't testing whether you know the real answer. It's testing whether you can build a logical structure from scratch, make reasonable assumptions out loud, and do clean math under pressure. The number you land on matters far less than how defensibly you got there.
Interviewers are watching for a clear structure stated before you start calculating, sensible round-number assumptions, and a sanity check on the final answer against something you'd actually expect to be true in the real world. They're also watching whether you stay calm and organized as the numbers get more complex, since that composure under pressure is a reasonable proxy for how you'd handle ambiguous client work.
Top-down vs bottom-up approaches
A top-down approach starts from a large known or estimated number and narrows down through a series of percentages. For example, starting from total population, then narrowing to the relevant age group, then to the percentage who buy the product, then to average spend.
A bottom-up approach builds up from a smaller unit. For example, estimating the number of coffee shops per neighborhood, then multiplying by the number of neighborhoods in the city. Bottom-up tends to be more defensible when you have a concrete, countable unit to start from, while top-down works better when you're estimating something diffuse across a whole population. Either is acceptable, choose whichever gives you more solid assumptions to build on for the specific question, and it's fine to mention briefly why you chose one over the other before diving in.
Structuring your assumptions before you calculate
Before doing any math, lay out loud the full chain of numbers you'll need and where each one comes from. This matters because it lets the interviewer catch a flawed assumption early rather than watching you calculate confidently down a wrong path for five minutes without any way to intervene.
- State each assumption as a specific number, not a vague range, so the math stays clean and easy to follow.
- Round to numbers that are easy to work with, such as 300 million instead of 331 million for a population estimate.
- Explain briefly why each assumption is reasonable, even a single sentence like 'assuming roughly half the population drinks coffee regularly' shows your thinking.
- If you're unsure about an assumption, say so explicitly and pick a reasonable middle-ground number anyway, rather than stalling on a number you can't be certain of.
A worked example: sizing the coffee shop market in a city
Say the question is: how much annual revenue do coffee shops generate in a city of 5 million people? A reasonable bottom-up structure: assume 40 percent of the population are regular coffee shop customers, that's 2 million people. Assume they visit an average of twice a week, spending 5 dollars per visit. That's 2 million times 2 times 5 times 52 weeks, which comes to roughly 1 billion dollars a year.
Then sanity check it. Does a billion dollars across a 5 million person city sound plausible for an entire category of retail? Compare it mentally to something you know, like total city retail spend, to see if the order of magnitude feels right before presenting it as your final answer. If it feels off by an order of magnitude, walk back through your assumptions and identify which one is most likely too aggressive or too conservative.
Common errors that lower your score
- Jumping straight into numbers without stating the overall approach first.
- Using overly precise inputs (like 8.3 percent) that suggest you're recalling a memorized fact rather than reasoning live.
- Doing the math silently and only announcing the final number, which gives the interviewer no way to follow your logic.
- Skipping the sanity check at the end, which is often what separates a good answer from a great one.
- Losing track of units partway through, for example mixing up weekly and annual figures, which throws off the entire final number even when the structure was sound.
How to get better at this specific skill
Market sizing is one of the more trainable parts of case interview prep, since the pattern repeats across most questions of this type. Practice on everyday objects and categories, how many gas stations in your metro area, how big is the market for umbrellas in a rainy city, building the same habit of stating structure, listing assumptions, calculating out loud, and sanity checking every time. After a dozen or so practice rounds, the format starts to feel routine rather than intimidating, which frees up mental space to focus on the specifics of whatever question you're actually given in a real interview.
Handling follow-up questions on your estimate
Interviewers frequently push on a market sizing answer after you've given it, asking things like how your number would change if a key assumption were different, or what would happen if a new competitor entered the space. Treat these as an extension of the same exercise rather than a signal that your original answer was wrong. Walk back to the specific assumption in question, explain how changing it would ripple through your calculation, and give a revised estimate. Handling this smoothly, without getting flustered or defensive about your original number, is often what separates a strong performance from an average one.
When it makes sense to present a range instead of a single figure
For some market sizing questions, especially ones with a wide amount of genuine uncertainty in the key assumptions, presenting a range rather than one precise number can be a stronger answer than forcing false precision. Stating 'depending on how many people actually buy this category regularly, I'd estimate somewhere between 700 million and 1.2 billion dollars, with my best single estimate around 900 million' shows awareness of where the real uncertainty in your assumptions lies, without abandoning the discipline of giving a concrete answer the interviewer can evaluate.