TAM, SAM and SOM for B2B SaaS

A broad market estimate can make a SaaS opportunity look convincing while saying little about the accounts a team can actually serve or acquire. Build three connected views instead: total demand for the product, the segment the current product and operation can serve, and a time-bound portion the team has a reason to believe it can win. The useful output is a documented account universe and a model whose assumptions a colleague can challenge.✓ 1

• Start with a customer count and an annual revenue assumption, rather than presenting a large industry spending figure as your plan.✓ 2
• Narrow the count by product fit, geography and operating capacity before calling it serviceable.✓ 1 2
• Put a period and acquisition capacity next to the obtainable portion. A percentage alone does not establish how many customers you can win.✓ 1 2

1. Define the market before counting it

Write down the product being sized, the buyer or account that pays, the geography, and the annual revenue unit. These choices decide what belongs in the denominator. A platform sold to manufacturing companies is not sizing the same population as a seat sold to individual employees, even if both products serve the same industry. The SBA frames demand, market size and location as separate research questions. Answer each rather than letting one broad industry number stand in for all three.✓ 3

For a B2B SaaS model, make the account the primary counting unit if contracts are sold to companies. Record whether an input source counts companies, locations, establishments or people. This matters when using public data: Census County Business Patterns reports establishments with paid employees, and its tables contain establishment and employment measures. A company with several sites can therefore appear as several establishments. That dataset can start a U.S. industry and geography screen, but the model must reconcile its unit before it claims a number of potential buying companies.✓ 4

2. Calculate TAM as demand, not a sales forecast

TAM describes the total revenue opportunity from the specified product market if all relevant demand were captured. HubSpot expresses a simple value model as total potential customers multiplied by average revenue per customer. For SaaS, decide whether that revenue input means an annual contract amount or a monthly amount annualized consistently. Do not multiply an annual account count by a monthly fee and label the result annual revenue.✓ 1✓ 2

Create a row for every input: eligible population, the reason for inclusion, data source, measurement year, counting unit and revenue assumption. If the population comes from an industry report, note the report's boundaries. If it comes from a prospecting database, note what the database actually covers. HubSpot describes top-down sizing as starting with broad market data and narrowing it; it describes bottom-up sizing as starting from observed customer counts, pricing and sales measures. Salesforce also points to customer insights and local calculations. A credible model can use both and investigate why their totals differ.✓ 1✓ 2

The first error to avoid is promoting TAM into a near-term revenue target. TAM is a ceiling for a defined market, not a claim that the company can acquire every potential customer. Keep the acquisition forecast on a different line of the model.✓ 2

3. Turn TAM into a serviceable account universe

SAM is the part of that market the current business can serve. Build it as a visible sequence of filters, with an account count after each filter: supported location, product requirements, implementation needs, buyer type and any capacity limit that genuinely prevents service. Salesforce explicitly names geography, production capacity, competition and demographics as factors in this narrowing. HubSpot's calculation also begins by counting customers that fit the business.✓ 1✓ 2

Keep each exclusion testable. “Not our ICP” gives a reviewer little to check; “requires an integration our current product does not support” gives a concrete reason. The same discipline applies to geography. The SBA asks where the business can reach customers, while Census CBP offers geography cuts down to ZIP code. A geography filter is useful only if it matches a real selling and service boundary, not merely because the table makes it easy to apply.✓ 3✓ 4

Salesforce's restaurant SaaS example illustrates the logic: its broad restaurant market becomes multi-location restaurants with enough volume for the product. That example is illustrative, and its populations are not a reusable B2B SaaS benchmark. In your model, use your own documented fit rules and the evidence behind them.✓ 1

4. Put a period and capacity behind SOM

SOM is the part of SAM the company can realistically convert within a specific timeframe. Label the period, then show the path from reachable accounts to potential customers. Inputs might include the number of accounts the team can work, how many can be served if they buy, and observed sales outcomes from the business itself. Salesforce lists pricing, competition, market trends and available resources among the considerations. A result without its assumptions is difficult to review or revise.✓ 1

Do not enter a convenient small percentage of SAM merely because it sounds conservative. HubSpot warns that claiming a share of a big market is meaningless without proving the ability to acquire and serve those customers. If a new product has no history, label the SOM scenario as a planning assumption, show what must happen for it to hold, and update it as direct customer evidence arrives. The SBA notes that research with customers gives a more specific view of the target audience than general existing sources alone.✓ 2✓ 3

Separate three questions in the spreadsheet: How many accounts meet the service criteria? How many can the team reach during the chosen period? How many of those could become customers under a stated, internally observed or explicitly assumed acquisition path? The first belongs in SAM. The final, time-bound result belongs in SOM. This separation makes it possible to challenge one assumption without rebuilding the entire market definition.✓ 1

5. Audit the model before using it

Review the units first. A company, establishment, employee and software seat are different units; Census CBP's unit is an establishment. Then inspect the overlap between source datasets so the same account does not enter twice. Check the revenue period against the count period. Finally, ask whether excluded geographies and product requirements reflect actual service limits. These checks follow from the source definitions and the published Census data unit; they are model hygiene, not a claim about a typical SaaS market.✓ 4✓ 1

Review the market side too. The SBA's competitive analysis guidance includes barriers to entering a market. A SAM can look large on paper while access to that segment is constrained. Record the barrier, its source, and whether it limits serviceability or only the portion obtainable in the modeled period. Do not silently subtract an arbitrary share.✓ 3

Finally, compare the top-down view with the bottom-up account count. HubSpot recommends refining broad estimates with actual business metrics. Where the figures diverge, trace the difference to population definitions, revenue units, data age or exclusions rather than averaging two incompatible numbers. The defensible result is a range of assumptions and a clear reason for the chosen operating case.✓ 2

FAQ

How we researched this

We read Salesforce and HubSpot's market-sizing guides, the SBA's market research guidance, and the Census Bureau's County Business Patterns description on October 12, 2026. We used those pages for definitions, calculation methods, research questions and dataset scope. This article does not assert a universal SaaS market total or acquisition rate; a company must supply its own market boundaries, account counts and sales evidence.✓ 1✓ 2✓ 3✓ 4