Bottom-up or top-down, choosing the right market sizing approach

Bottom-up or top-down? If you're defaulting to the same approach every time, you're leaving points on the table.

I see this constantly in mock cases. Someone learns the top-down method first because it's what most prep books open with, gets comfortable with it, and then applies it to every sizing question they get for the rest of their preparation, even though each approach has situations where it works well and situations where it falls apart, so choosing the wrong one means your assumptions become harder to defend at exactly the moment when defensibility matters most.

The real question in any sizing problem isn't whether you can do the arithmetic, since the interviewer assumes that, but whether you can pick a starting point and a chain of assumptions that someone else would actually agree with. If the assumptions sound reasonable on the way in, the answer at the end is credible whether or not it's exactly right.

When top-down is the right tool

Top-down works when you have a reliable total that you can narrow down through reasonable segmentation, where the starting number is something nearly anyone in the room would broadly agree on and the steps that follow are about choosing the right percentages.

Sizing the annual revenue of a major coffee chain in a country is a natural fit, since you can start with the country's adult population, work through the share that drinks coffee regularly, and move through purchase frequency and average spend per visit, with each step using a number you can sanity-check against your own behaviour or a published statistic. The interviewer doesn't have to take your word for any of it, which is what makes top-down feel safe when it's the right tool: the conversation moves smoothly because every assumption can be challenged and adjusted without breaking the structure underneath.

Where top-down breaks

Top-down stops being defensible the moment the percentages you need to apply aren't grounded in anything anyone could check.

If someone asks you to estimate the annual revenue of a single major tourist attraction, top-down gets difficult fast, because you'd have to start with national or international tourist counts and then guess what share of those tourists visit that specific place, and there's no published number for that share, no obvious anchor, so the percentages you end up using are essentially invented. You can do the multiplication, but the answer is sitting on assumptions the interviewer can't follow you into.

The same problem shows up when sizing niche products, premium segments, or anything where the relevant population is small relative to the obvious starting total, because top-down forces you to apply percentages on percentages and each one quietly compounds the uncertainty.

Where bottom-up earns its place

Bottom-up shines when you can build the answer from a unit, then scale it up.

For the tourist attraction, you'd start with the attraction's daily capacity, work through how many visitors it can handle in a day given operating hours and throughput, factor in seasonality across the year, and multiply by a sensible ticket price to get to annual revenue, with each of those numbers grounded in something physical that the interviewer can picture and check.

The same approach works in plenty of other situations: for a hospital, you'd build from beds and occupancy through average stay length to revenue per stay; for a factory, from production lines through hourly output and shift count; and for most service businesses where capacity or supply is the binding constraint, the unit-economics frame applies the same way. The assumptions are auditable because they describe a real thing rather than an abstract slice of a larger total, and the chain of logic mirrors how the business actually generates revenue, which is why bottom-up answers usually feel more concrete to interviewers.

The decision happens in the first ten seconds

Before you start any sizing question, spend ten seconds asking which direction gives you the most defensible assumptions. Two questions are usually enough.

First, is there a reliable total I can start from? If yes, top-down might work, but if the total exists and the segmentation requires inventing percentages that nobody has data on, top-down probably won't.

Second, is there a unit I can build from, like a store, a customer, a hospital bed, or a machine that produces something countable? If you can describe the unit and its key economics in one sentence, bottom-up is usually cleaner.

The mistake I see most is candidates skipping this question entirely and starting with whatever method they used in their last mock, which the interviewer can almost always tell apart from a choice made by judgement.

Common traps once you've picked

Switching mid-sizing. A candidate starts top-down, hits the first percentage they can't justify, and quietly pivots to bottom-up halfway through, ending up with two incomplete sizing trees stitched together while the interviewer loses the thread. If your initial choice isn't working, say so, reset, and start over, which looks far better than trying to disguise the switch.

Sanity-checking only at the end. Both methods need pressure-testing as you go, not just on the final number. After you've laid out your top assumptions, take a moment to ask whether the intermediate numbers feel right, because a coffee chain serving 40 million daily customers in a country of 60 million should be telling you something well before you multiply.

Treating company sizing and market sizing as the same problem. Sizing a company is almost always bottom-up because the unit (a store, a contract, a customer) is something you can describe, while sizing a market often allows top-down because the relevant total is a population or spend pool, so don't reach for the same approach automatically across both situations.

Over-precising the assumptions. If you've started with a population of 60 million and need to get to coffee drinkers, "around half" is fine, since the risk is in picking the wrong order of magnitude rather than in whether you used 48 percent versus 52 percent. Use round numbers that the interviewer can follow without writing down.

How to practise the choice

Most candidates drill case math by working through a sizing problem to the final number, which trains the wrong muscle. What you actually want to practise is the first thirty seconds: take any sizing prompt you've seen before, decide which approach you'd use, write down the first three assumptions for each direction, and ask yourself which set is more defensible. Do this with ten or fifteen prompts back-to-back, and the choice starts becoming automatic in real interviews.

The arithmetic is rarely what costs candidates the case; the real moment of failure is usually two minutes in, when they realise the assumptions they've committed to are the kind nobody can check, and there's no clean way to backtrack.

The point of the choice

The interviewer is testing whether you can pick the right method for what's in front of you and then walk a sceptical client through the assumptions without anyone having to take you at your word; knowing two methods is the easy part.

A bottom-up answer with grounded assumptions almost always survives a first round even when the arithmetic is shaky, because the structure is something the interviewer can repair with you in real time. Top-down rarely survives the same situation: when the percentages aren't defensible, the only fix is to start over.

Spend ten seconds on the choice. The rest of the case gets easier from there.