Market sizing is the work of estimating how much demand exists for an offering and, more importantly, building that estimate so it holds up under scrutiny. The three standard terms order the question by scope. TAM (total addressable market) is the entire universe of demand for a category: everyone who could conceivably buy. SAM (serviceable addressable market) is the slice your specific offering, pricing, and channels can actually reach. SOM (sometimes written out as serviceable obtainable market, but we define it as Share of Market) is the portion of that reachable market you actually capture — your realized share. The acronyms are the easy part. Frankly, each organization may use slightly different definitions, but the intent is consistent — paring down from the possible to the practical. Building a market forecast you can defend means something more: building the number from the bottom up out of explicit, segment-level assumptions, accounting for the customer's real decision (including their option to do nothing), and making visible which assumptions, if wrong, would change the answer. A market-size number you can't defend is worse than no number at all: it gets punctured in the first investor meeting, and it takes your credibility with it.
When a founder presents a market size and an investor pushes back, the conversation almost never turns on whether the figure is large. It turns on where the figure came from. A nine-figure TAM that resolves, under one question, to "we took the industry's total revenue and assumed we'd get one percent" tells the investor exactly one thing: the founder hasn't done the work. One percent of a huge number is a fantasy dressed as a forecast, and it survives exactly as long as nobody asks the second question.
So the real deliverable is not the headline figure. It's the characterization of the opportunity — a chain of stated assumptions, each one visible and each one challengeable, that produces the figure as an output rather than asserting it as an input. The goal of new-business financial modeling is not a precise forecast; precision is unavailable this early, and pretending otherwise is its own kind of weakness. The goal is a defensible characterization: sound logic, plus insight into which assumptions are consequential enough to be worth gathering evidence on. When you can show an investor the logic and point to the two or three numbers that actually move the result, you've shown them how you think — something a polished slide can't fake.
There are two ways to arrive at a market size — but the difference that actually matters isn't top-down versus bottom-up. It's whether the logic underneath is rigorous or simplistic.
The top-down approach starts from a big, published number — total industry revenue, an analyst's category forecast — and carves it down with a series of percentages. Done with care, a top-down estimate is a useful cross-check on the order of magnitude. The trouble is how it's usually done: each percentage is a single point value, asserted without support, and the errors compound. A top-down number built that way tells you what's theoretically out there; it tells you almost nothing about whether you can capture any of it.
The bottom-up approach builds the number from the unit the business actually runs on. How many customers are there in a defined segment? What share of them have the unmet need your offering addresses? What does each one pay, how often, and how many of them will plausibly adopt over what period? Bottom-up is slower and it often produces a smaller, less flattering figure — and that is exactly why it's more defensible. Every input is something you can name, source, and revise as you learn. When an investor asks "where does this come from," you have an answer for each link in the chain instead of a single embarrassing percentage.
Built this way, the three terms stop being labels stacked on a slide and become a chain of filters you can inspect. TAM is the total annualized demand across the segments you're modeling. SAM applies reach and constraint filters, the portion of that demand your offering and channels can actually serve. Share of Market applies share filters — the portion of that reachable demand you might actually capture. Multiply Share of Market by your pricing and you have revenue. That's what bottom-up means in practice: building from segments, weighted by reach, weighted by share, multiplied by price, each filter a named, sourceable assumption rather than a percentage pulled from the air.
A bottom-up build wins for the same reason a structured strategy beats a narrative pitch: vague, single-point concepts can't be evaluated. A market sized by hand-waving can only be argued about; a market sized from explicit, segment-level assumptions can be examined — and a number that can be examined is a number you can stand behind. Two things make that build real rather than cosmetic: the quantities have to be in a defined unit (customers, transactions, or units sold), and they carry the customer's purchase and repurchase logic, how many buy, how often, and whether they renew. That's where the deeper modeling lives. Building bottom-up also forces you to be honest about the realism factors inflated models gloss over: adoption rate, switching cost, segment reachability, pricing realism — and the customer's option to do nothing at all, which in many markets is the most likely outcome by default and the real baseline a new offering has to beat.
A market size you can defend has two properties a market size you can't defend lacks.
First, the assumptions are explicit. Every consequential input is named and classified, not buried in a cell. It helps to separate three kinds of inputs: the choices within your control (which segments you'll serve, how you'll price); the assertions about what others in the market will do (whether a channel will carry you, whether customers will switch); and the genuine uncertainties — the values you can't pin down yet, carried as ranges rather than false-precision point estimates. Each kind calls for a different response: a choice needs a decision, an assertion needs validation, an uncertainty needs a range and then evidence to tighten it.
Second, the sensitivity is visible. Once the uncertain inputs are carried as ranges, you can see which ones the answer actually depends on. Most inputs in a market model don't matter much; a handful swing the result dramatically, and knowing which is which is the whole game. It tells you where to spend your limited evidence-gathering effort — tighten the assumptions that move the number, stop fussing over the ones that don't. It also makes the model honest with an investor: you can show the range the answer lives in, and name the two or three things you'd need to learn to narrow it. Match the depth of your analysis to what's actually at stake on each assumption. That's not hedging — it's the opposite of hand-waving. It's knowing precisely how confident you're entitled to be.
A number presented as a single point with no ranges and no stated assumptions is asking to be disbelieved. A number presented with its assumptions on the table and its sensitive inputs flagged invites a different conversation — one about evidence, not about credibility.
This is what it means to treat market sizing as a probabilistic forecast rather than a single guess. When the uncertain inputs are carried as ranges, the right output isn't one number — it's a distribution: the band of outcomes your assumptions actually imply, together with a reading of which assumptions drive the spread. Run the ranges through enough simulated scenarios and you get both at once: the shape of the answer, and the short list of inputs worth gathering evidence on. The point estimate was never the honest deliverable; the distribution and its sensitivities are.
TAM (total addressable market) is the entire universe of demand for a category — everyone who could conceivably buy if there were no constraints. SAM (serviceable addressable market) is the portion of that TAM your specific offering, pricing, and channels can actually serve. TAM answers "how big is the whole space"; SAM answers "how big is the part I can realistically reach." Share of Market (our definition of SOM in TAM/SAM/SOM) narrows one step further: the portion of that reachable market you actually capture — your realized share, after competition and the customer's option to stick with the status quo.
The divide that actually matters isn't top-down versus bottom-up — it's rigorous versus simplistic logic. Bottom-up tends to be more defensible because it forces explicit, segment-level assumptions you can name and source: how many customers, what share have the unmet need, what each pays, how many adopt and how often. A top-down estimate from a credible source, built on sound logic rather than single-point guesses, is a legitimate cross-check on the order of magnitude. The real failure mode is a top-down number carved down by unsupported point-value percentages — fast, impressive, and impossible to defend. Best practice: build bottom-up, and use a sound top-down estimate to sanity-check the magnitude.
Make the assumptions explicit and the sensitivity visible. Name every consequential input rather than burying it; separate the choices you control from the assertions you're betting on and the uncertainties you can't yet pin down; carry the uncertain inputs as ranges, not single points; and identify which inputs the answer actually depends on, so you can show what you'd need to learn to tighten it. Account for the realism factors inflated models skip — adoption rate, switching cost, segment reachability, pricing realism, and the customer's option to do nothing. The deliverable isn't a precise forecast; it's a sound, examinable characterization of the opportunity.
The acronyms are the easy part. Thousands of articles walk you through TAM, SAM, and SOM. Far fewer tell you the uncomfortable truth: a market size is only as good as the model behind it, and most models fall apart under a single honest question because their assumptions were never on the table to begin with.
This is the gap Growth Forge® Software is built to close. Growth Forge helps founders, startup teams, and innovation project leads build new-business strategy the way it holds up under scrutiny, exposing the choices and assumptions that make or break a venture, with financial modeling tools that don't require Excel mastery or strategic finance expertise. Its Market Sizing tool is a segment-based, bottom-up revenue-forecast model: it builds the number along the TAM-to-SAM-to-Share-of-Market funnel (total demand in your segments, filtered by what your offering and channels can reach, filtered again by what you actually capture) and turns Share of Market into a revenue forecast through your pricing. Uncertain inputs are carried as ranges and run through Monte Carlo simulation, so the output is a distribution of revenue outcomes plus a reading of which assumptions actually drive it, not a single point estimate pretending to a precision you don't have. It's built on BRI Associates' decades of practitioner experience in corporate innovation and new business development: you bring the logic and the assumptions about your business; Growth Forge handles the modeling and the math.
If you'd like to build a market forecast you can actually defend, with assumptions explicit and sensitivity visible, you can model it in Growth Forge. A free trial is available, with a pre-loaded worked example to model alongside your own.
