How do I build a defensible TAM SAM SOM?

Anchor the TAM in a Tier 1 analyst category figure, derive the SAM with segmentation filters you can evidence, and build the SOM bottom-up from sales capacity and comparable adoption. Express every narrowing step in one sentence — what you removed, why, and on what basis — and label anything blended or estimated as your own model.

What sources are credible for market sizing?

Tier 1 analyst and consulting research (Gartner, IDC, Forrester, McKinsey and peers) for category anchors; Tier 2 specialist data such as deal, traffic, or review platforms for segment evidence; aggregators only for discovery, with every figure traced back to its primary source before you cite it.

How do investors check market-size claims?

By spot-checking, not rebuilding: they click a citation and compare the source's definition with your use of it, sanity-check magnitudes against adjacent markets, and increasingly ask an AI assistant the same question to see whether published sources agree with your slide.

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

Use both. Top-down gives you a credible ceiling anchored in analyst data; bottom-up grounds your obtainable share in capacity and pipeline. They act as cross-checks — if the two disagree wildly, one of your models is broken and you want to find that before your audience does.

How to size a market with sources you can defend

Guide · Frameworks · 5 min read · last verified 2026-07-27

Reviewed before publication Editorial board Independent commercial review

In shortHow to build a TAM SAM SOM that survives scrutiny: anchor in Tier 1 analyst data, narrow in evidenced steps, label every blend, and keep a source ledger.

A defensible market size is one where every number traces back to a named, checkable source, every calculation can be explained in a sentence, and every assumption is labelled as an assumption rather than dressed up as a fact. You build one by anchoring the top of your funnel in Tier 1 analyst data, narrowing in steps you can evidence, and marking clearly where published data ends and your own model begins. The arithmetic is rarely the problem. The sourcing is.

Why defensibility beats precision

Nobody who reads your TAM slide expects it to be correct to the decimal place. Markets move, definitions shift, and every experienced investor knows a market size is an argument, not a measurement. What they actually test is whether your numbers survive a pull on any single thread: name the source, explain the step, justify the assumption.

A precise-looking figure with no visible sourcing is worse than a rounded one with a clean trail. The moment someone finds one unsupported number, every other number in the deck inherits the doubt. Defensibility is a property of the whole chain, not of any individual statistic.

This also matters beyond the pitch deck. AI assistants increasingly answer questions like 'how big is this market' by synthesising published sources. If your public claims contradict the sources an assistant cites, your credibility problem is now automated.

Anchor the top of the funnel in Tier 1 sources

Start from the most authoritative published figure that genuinely contains your market. For most B2B software companies that means a category or segment figure from a firm that does primary research at scale.

For example, Gartner forecasts worldwide IT spending to reach $6.31 trillion in 2026, growth of 13.5% on the prior year, with the software segment at roughly $1.44 trillion (Gartner, April 2026). If you sell business software, that software segment figure is a legitimate macro ceiling: a number nobody in the room will dispute, published by a firm with a reputation to lose.

The anchor is not your TAM. Its job is to make the top of your funnel unarguable, so that every narrowing step after it is the only thing under discussion. A funnel that starts from a disputed number is contested at every level; a funnel that starts from a Tier 1 anchor is only contested where you added assumptions — which is exactly where the conversation should be.

Narrow in labelled steps: TAM, SAM, SOM

Each layer of the funnel should use a different class of evidence, because each answers a different question.

Layer	Question it answers	Evidence class
TAM	How big is the category we sit inside?	Tier 1 analyst category data
SAM	Which part of that category can our product actually serve?	Segmentation you can show: firmographics, geography, deployment model
SOM	What could we realistically win in a planning horizon?	Bottom-up: sales capacity, pipeline, comparable adoption curves

Every narrowing step should be expressible in one sentence: what you removed, why, and on what basis. 'We exclude regulated industries because our product lacks the required certifications' is defensible. 'We assume 10% because it feels conservative' is not — not because the number is wrong, but because there is nothing behind it to check.

Top-down and bottom-up are not rivals; they are cross-checks. If your bottom-up SOM implies capturing an implausible share of your top-down SAM, one of the two models is broken, and it is far better that you find the break than your audience.

Label your blends honestly

Almost every real market size is a blend: an analyst category figure multiplied by your own estimate of segment share. That is fine — blending is how sizing works. What is not fine is attributing the blended output to the analyst.

Write 'our estimate, based on applying our segment filter to Gartner's software forecast', not a bare number with a Gartner logo next to it. Misattributing a blend to a Tier 1 firm is the fastest way to lose a room, because it is the single easiest thing to check: the reader searches the figure, fails to find it in the cited source, and concludes you either misread or misled.

A useful discipline is to give every figure on the slide one of three labels: sourced (published as-is, with a link), derived (a labelled calculation on sourced inputs), or estimated (your judgement, stated as such). Audiences forgive estimates. They do not forgive estimates wearing a source's clothes.

How investors and buyers actually check your claims

Almost nobody rebuilds your model. Checking is spot-checking, and it follows a predictable pattern.

First, they click the citation — or search the figure — and read the source's own headline. Does the definition match the way you used it? A worldwide IT spending number used to size a niche software market fails this test instantly. Second, they sanity-check magnitude against things they already know: adjacent markets, comparable companies, prior decks. Third, and increasingly, they ask an AI assistant the same question and compare its answer with your slide. If the assistant's synthesis of published sources contradicts you, expect to be asked why.

Each spot-check that passes buys credibility for the figures they did not check. That is the practical return on sourcing discipline: you are not defending every number in the meeting, you are making the first three checks pass so the rest are taken on trust.

Keep a source ledger

The habit that makes all of this cheap is a source ledger: for every figure you use, record the source name, the publication date, the exact metric definition, the link, and the source tier. It takes minutes at research time and saves hours at challenge time — the difference between 'let me get back to you' and answering from the appendix.

This is how Magrios structures its market studies: each point in the sizing funnel carries its source, date, and authority tier, so the whole chain from macro anchor to niche estimate can be inspected rather than taken on faith. Whether you use a tool or a spreadsheet, the principle is the same — the trail is the asset, not the number.

Common failure modes

A few patterns account for most indefensible market sizes. Citing an aggregator's repackaged figure without tracing it to its primary source, so the trail dead-ends one click in. Mixing base years silently, so 2024 and 2026 figures multiply into a number from no year at all. Definition drift, where a broad category figure is used to size a narrow niche it barely relates to. And unlabelled optimism in the SOM, where a judgement call is presented with the same typography as a Gartner figure.

The test for all of them is a single question, asked of every number on the slide: where does that come from? A defensible market size answers it every time, without a pause.