TAM, SAM, and SOM: How to Size a Startup Market

TAM vs. SAM vs. SOM at a glance#

TAM is not "everyone who could benefit." SAM is not "the segment we like." SOM is not "1% because 1% sounds conservative." Each number should be reconstructable from a buyer definition, a revenue unit, filters, a time period, and named evidence.

Why does market sizing matter to founders?#

It tests whether the business model can support the desired outcome. A large user population can still produce a small revenue opportunity: a marketplace may move billions in gross merchandise value but retain only a small take rate; a vertical SaaS product may reach many locations but earn too little per location to fund an expensive sales motion. For fundraising, the question is not only "Is the market large?" but "Can this company capture enough gross profit and enterprise value to fit the fund's return requirements?" Sequoia's business-plan framework places market potential alongside the customer, competition, business model, and long-term vision—the size claim has to fit the rest of the story.

It forces a precise customer definition and links pricing to strategy. A defensible estimate requires naming the buying unit: a company, establishment, team, licensed professional, household, device, or transaction. That choice drives both the count and the price. In a bottom-up model, price is a primary lever—but use realized annual revenue per account by segment, net of ordinary discounts and actual usage, not list price. If the resulting market is too small, the answer is to change packaging, add a revenue stream, or move upmarket—not to inflate the number of buyers.

It makes the go-to-market plan auditable. SOM should reconcile to the revenue plan. If the deck claims $20M of revenue in three years, the model should show enough reachable accounts, qualified opportunities, wins, onboarding capacity, and retained customers to produce it. A focused SAM and a capacity-based SOM often build more credibility with investors than a giant TAM, because they show the founder understands the buyer and the path to revenue.

How do you calculate TAM, SAM, and SOM?#

Step 1 — Define the market boundary. Write one sentence: Customer + problem/use case + offer category + geography + base year + revenue basis. For example: "Annual subscription revenue for compliance-documentation workflow software sold to U.S. multi-site outpatient clinic groups, measured in 2026 dollars." "U.S. healthcare" is an industry, not a market definition. The SBA recommends examining demand, number of interested buyers, customer location, saturation, and the prices customers pay for alternatives, using both existing data and direct research.

Step 2 — Choose the buying unit and revenue unit. Use the unit that signs the contract and drives billing.

Step 3 — Build TAM from the bottom up. For one segment, TAM = total eligible buying units × realized annual revenue per unit; for several, sum the segments. Bottom-up does not mean "small"—it means the estimate is assembled from the economic units that produce revenue. Y Combinator's pitching guidance recommends this over a purely top-down estimate: identify who buys comparable solutions, estimate the unit price, and multiply. Use top-down data (government statistics, filings, trade data) as a boundary check, not as a source of unsupported percentages.

Step 4 — Derive SAM with explicit serviceability filters. Start from the TAM buyer table and apply only filters that affect whether the current offer can serve an account: required integrations, customer size, launch geography and language, regulatory approval, procurement route, current (not roadmap) capabilities, and an economically reachable channel. A useful segment formula:

SAM_i = TAM accounts_i × product-eligible rate_i × geography-eligible rate_i × other non-overlapping eligibility rates_i × annual revenue per account_i

Avoid multiplying correlated percentages as if independent. Where possible, enumerate the jointly eligible accounts directly and deduplicate.

Step 5 — Build SOM from the acquisition and retention system. SOM needs a horizon (for example, exit ARR in 36 months). The simplest model:

New customers_t = reachable accounts_t × qualified-opportunity rate_t × win rate_t

Active customers_t = active customers_(t-1) × (1 − logo churn_t) + new customers_t

SOM revenue_t = active customers_t × realized annual revenue per active customer_t

Then impose operational caps: new customers cannot exceed sales capacity or onboarding/production capacity. The resulting market share is a diagnostic (SOM share = SOM revenue / SAM revenue at the same horizon)—do not start with the share. No universal 1%, 3%, or 5% rule can substitute for a distribution model.

Step 6 — Triangulate and show sensitivity. Use at least two approaches when the decision matters: a bottom-up buyer model, a top-down boundary check, and (for a new category) a value-based model of customers × economic value created × plausible value-capture rate. A customer saving $1M does not imply $1M of willingness to pay—validate with paid pilots and contracts. Show ranges for uncertain drivers rather than one falsely precise total.

Which data sources should you use?#

For U.S. B2B markets, the Census Bureau's County Business Patterns gives annual establishment counts, employment, and payroll by geography and detailed NAICS industry for employers; Statistics of U.S. Businesses adds firm-level views by enterprise size; and Nonemployer Statistics covers businesses without paid employees. Match the NAICS code to the actual buyer, distinguish a firm from an establishment, note the data year and lag, and avoid adding overlapping datasets.

For industry-level checks, the Bureau of Economic Analysis defines gross output as an industry's sales or receipts (including business-to-business intermediate inputs), whereas GDP is a value-added measure—so the two are not interchangeable with addressable vendor revenue. Published market reports can be useful inputs, but record each report's category definition, base year, geography, currency, and whether the figure is vendor revenue, customer spend, unit volume, or economic value.

Secondary statistics tell you how many units exist; they rarely tell you what fraction has the problem, can buy, will switch, or will pay your price. Combine them with primary evidence: customer interviews, paid pilots, win/loss data, CRM conversion by segment, realized price data, and channel-partner capacity. The SBA recommends pairing existing quantitative sources with direct research because public data is often not specific enough to the target audience.

Worked example: B2B compliance software#

A startup sells compliance-workflow software to U.S. multi-site outpatient clinic groups. (Figures are hypothetical.)

1. Boundary and unit. Buying unit: one clinic group that signs one contract (not each location). Revenue unit: realized annual contract value (ACV) of $12,000. Base year: 2026. Account-level research identifies 20,000 U.S. clinic groups with the relevant workflow and enough scale to buy.

TAM = 20,000 accounts × $12,000 ACV = $240 million

2. SAM. The product initially works only for groups that operate in the five launch states, use a supported clinical system, have a centralized compliance buyer, and can adopt without a certification the startup lacks. After deduplication, 5,000 groups meet all four conditions.

SAM = 5,000 accounts × $12,000 ACV = $60 million

Do not get 5,000 by multiplying four rough percentages unless the filters are independent and validated—a direct account count is stronger.

3. 36-month SOM. The plan: Year 1, founders close 20 accounts; Year 2, two sellers close 2/month for 10 productive months (2 × 2 × 10 = 40); Year 3, four sellers at the same productivity (4 × 2 × 10 = 80); annual logo retention 90%; ACV flat at $12,000.

Year 1 cohort = 20 × 0.9 × 0.9 = 16.2 · Year 2 cohort = 40 × 0.9 = 36 · Year 3 cohort = 80

Active accounts ≈ 16.2 + 36 + 80 = 132

36-month SOM = 132 × $12,000 = $1.58 million exit ARR

SOM share of SAM = $1.58M / $60M = 2.6%

The 2.6% share is the output of the hiring, productivity, and retention assumptions—it was not chosen first. If sellers close 1.5/month and retention is 85%, the total falls materially; present the two or three drivers that matter most rather than hiding them in a single "conservative" case. The financial model must use the same seller count, productivity, churn, and ACV.

Sources#

1. GOV.UK, Track 1: Investor Readiness Essentials.
2. U.S. Small Business Administration, Market Research and Competitive Analysis.
3. U.S. Census Bureau, County Business Patterns API, Statistics of U.S. Businesses, and Nonemployer Statistics.
4. U.S. Bureau of Economic Analysis, Gross Output by Industry.
5. Y Combinator, Practical Design: Pitching.
6. Sequoia Capital, Writing a Business Plan.

Research note: TAM, SAM, and SOM are planning estimates, not audited facts. Preserve the assumptions, source definitions, and calculation model so the page can be updated as the product and evidence change.