Is top-down or bottoms-up market sizing better?

Bottoms-up is more defensible because it builds from countable units you can source, while top-down starts from a headline figure and applies a share with no mechanism behind it. Use top-down only as a cross-check on a bottoms-up number, never as the primary estimate.

Should I present market size as a single number or a range?

A range, with its driving assumptions and confidence levels attached. A single precise figure implies a certainty you don't have, and it collapses the moment someone asks how you got it — a range tells the truth about what you actually know.

What's the fastest way to spot a fake market size?

Look for precision without provenance — an exact figure with no visible chain of assumptions behind it. If no one can explain what would have to be true to capture the claimed share, the number is a guess, not an estimate.

The honest market-sizing playbook: numbers you can defend

Guide · Market Growth · 5 min read · last verified 2026-07-19

Reviewed before publication Editorial board Independent commercial review

In shortAn honest market size is bottoms-up, with every assumption named, sourced, and marked with a confidence level — so a skeptic can trace it rather than trust it. The sequence, the checks, and how to present a range.

An honest market size is a bottoms-up estimate where every input is named, sourced, and marked with a confidence level — so a skeptic can trace exactly how you reached the number instead of being asked to trust it. The number itself matters less than whether it survives being questioned. Most don't.

Why most sizing collapses under one question

Take almost any top-down market size — "the category is worth X, and we only need one percent" — and ask a single question: where does that one percent come from? The answer is usually nowhere. Top-down sizing starts from a big published figure and multiplies by a share pulled from optimism. It collapses because the share has no mechanism behind it; nobody can explain what would have to be true to capture it.

The tell is precision without provenance. A market sized at a suspiciously exact figure, delivered with no visible chain of assumptions, is a guess wearing a suit. False precision is the failure mode to design against: a defensible range beats an indefensible point estimate every time, because the range tells the truth about how much you actually know.

The bottoms-up sequence, step by step

Bottoms-up sizing builds the number from units you can observe, not from a headline you can't verify. The sequence:

Define the buyer precisely. Not "businesses" — the specific segment with the specific problem your product solves. Vague denominators produce vague sizes.
Count the buyers. How many entities match that definition? Anchor to something countable — a census, a registry, a directory — not an estimate of an estimate.
Estimate what each buyer would pay per year. Use symbolic terms: price per buyer times purchase frequency. Ground the price in real willingness to pay, not aspirational pricing.
Multiply for total addressable revenue. Buyers times annual value. That's your addressable market — the ceiling if you won everyone.
Discount to what's reachable. Serviceable and obtainable markets are the addressable market cut down by geography, segment, and how many buyers you can actually reach and win.

Each step is a number you can defend or a number you can't. The discipline is refusing to move to the next step until the current one has a source. SAM and SOM, practically defined walks through the reachable-market cuts in detail.

Naming every assumption: the confidence column

The single practice that separates a defensible size from a guess is a confidence column. Build the estimate as a table where every input carries three things: the value, the source, and how confident you are in it.

High confidence — countable, sourced facts. The number of registered businesses in a segment; a published price.
Medium confidence — reasoned estimates from partial data. An adoption rate inferred from a comparable category.
Low confidence — assumptions you're making to proceed. Willingness to pay for a product that doesn't exist yet.

The confidence column does two things. It shows a reader exactly where the estimate is soft, which builds trust instead of eroding it. And it tells you where to spend research effort: the low-confidence inputs with the biggest effect on the total are the assumptions worth a week of validation. A number with no confidence column is asking to be believed; a number with one is asking to be checked. What is willingness to pay? is usually the softest input in the chain — treat it that way.

Sanity checks that catch nonsense

Before presenting any size, run it through checks that catch the errors bottoms-up math is prone to:

Top-down cross-check. Build the number the other way, from a published category figure, and see if the two land in the same neighborhood. If they're wildly apart, one has a broken assumption.
The share-of-wallet check. Does your per-buyer revenue imply buyers spending an unreasonable fraction of their budget on your category? If so, your price or frequency is off.
The saturation check. Does capturing your "obtainable" market imply signing more customers than your segment plausibly contains, or than your team could ever service? If yes, the discount was too shallow.
The comparable check. Do companies serving this exact market have revenue consistent with your size? A public comparable is the cheapest reality test available.

A size that survives all four isn't proven — but a size that fails any of them is disproven, which is more than most decks can say.

Presenting a size with its evidence attached

Present market size as a range with its reasoning attached, never as a lone number on a slide. The format that earns trust: a defensible range, the two or three assumptions that drive it, the confidence on each, and what would move the number. That last part matters — telling a reader "if willingness to pay is half our estimate, the market is this much smaller" shows you understand your own model.

This is the same principle Magrios applies to market observation generally: a claim is only as strong as the evidence trail a skeptic can follow. A number you can defend under questioning does real work — it guides where to build and where to stop. A number you can't defend gets exposed in the first serious conversation and takes your credibility with it. How market intelligence informs positioning is built on the same evidence-first stance.

What to do with this

Rebuild your headline market size from the bottom up. If it started as "big number times one percent," you don't have a size you can defend — you have a slide.
Add a confidence column to every input. Value, source, confidence. The soft inputs with the largest effect on the total are your research priorities.
Run the four sanity checks before anyone else does. Top-down cross-check, share-of-wallet, saturation, and comparable — disprove your own number before a skeptic disproves it for you.
Present a range, not a point. State the driving assumptions and what would move the number. A range that tells the truth about your uncertainty beats a precise figure that hides it.