Key takeaways
- TAM counts every firm that could buy the product. SAM counts the firms the product and sales region can serve. SOM counts the firms a sales team can win in the next 12 months.
- Count the firms that match the buyer definition, not all businesses. In the US, 6.4 million of 36.2 million firms have paid employees, which makes the starting count about 5.6 times smaller for products that need staff.
- SOM follows sales capacity and the share of SAM that is free to switch this year. A flat percentage of SAM ignores both.
TAM (total addressable market) is simply a way to answer the most basic question – how many companies may purchase your product and how much will those purchases cost you over one year. Analysts often use their own version of the total by using the entire size of the analysts’ “industry” as their version of TAM – which leaves out both the identity and quantity of the potential customers. By counting actual firms, there is a number that can be verified and a list that a sales team can work.
This beginner’s guide describes TAM, SAM and SOM, explains how to calculate each from data regarding publicly traded firm counts; concludes with an example and provides a chart to assist in correcting for an underachieving SOM. The firm-count data comes from the U.S. Small Business Administration Report dated Feb. 2026.
What Do TAM, SAM and SOM Mean?
Six terms appear throughout this guide:
- Firm: one legal business, counted once however many locations it runs.
- Employer firm: a firm with paid employees. A nonemployer firm has an owner and no staff.
- Annual contract value (ACV): the yearly revenue from one customer.
- Opportunity and win rate: an opportunity is an account a rep is actively selling to. Win rate is the share of opportunities that end in a signed deal.
- NAICS code: the number the US government assigns to each type of industry, with two digits for a broad sector and six for a narrow one.
- CRM: software where a sales team records accounts and deals.
The three numbers describe the same buyers at three levels of filtering. TAM applies no limits, SAM applies the limits of the product and region, and SOM applies the limits of the sales team. Some sources write SAM as serviceable available market, which means the same thing.
Also Read: A Beginner’s Guide to Conversion Optimisation in B2B Marketing
What Is the Difference Between Top-Down and Bottom-Up TAM?
- Top-down: start from a published industry total, such as yearly global spend on a software category, and take a percentage. It is fast, and the total often covers buyers the product cannot serve.
- Bottom-up: count the firms that fit the buyer definition and multiply by ACV. Each input has a source and a date.
Market-intelligence firm Infomineo reports that the two methods often land three to ten times apart. Bottom-up works better as the base, and top-down works as a cross-check. A large gap means one input is wrong, so each input should be checked until the gap closes.
How Is TAM Calculated? Step-by-Step Formula
- Define the buyer in one sentence. Include industry, employee range, region and the job title that approves the purchase, for example finance teams at US firms with 50 to 499 employees. A NAICS and SIC code search turns an industry name into the NAICS codes that official statistics use.
- Count the firms. Pull counts for those codes, size bands and regions from public statistics (sources follow below).
- Decide which firms stay in the count. Drop firms with no employees when the product needs staff to use it or is priced per seat. Keep them when the product serves sole proprietors.
- Multiply by ACV. Use the median of recent signed deals or the list-price average, and leave out the most expensive plan.
TAM = firms that fit the buyer definition × ACV
How Many US Businesses Belong in a TAM? 2026 Firm Counts
The SBA Office of Advocacy counted 36,207,130 US small businesses (fewer than 500 employees) in February 2026, plus 21,041 large businesses, for 36,228,171 firms in all. Most small businesses have no staff, and many B2B products have no user at a firm with no staff.
Source: SBA Office of Advocacy, Frequently Asked Questions About Small Business, February 2026, based on 2022 Census data. The last row adds small employer firms and large businesses.
The 36,228,171 total is about 5.6 times the employer count. The figures reflect 2022 data released in 2025, so a newer Census release should replace them once published.
The unit also matters. Census statistics report firms and establishments separately, and an establishment is one physical location. A retail chain with 200 stores is one firm and 200 establishments. Software bought once for the whole company counts the firm, a product priced per site counts the establishments, and a product priced per user counts employees.
Also Read: Tracking Attendee Overlap Across Events: Benefits, Limitations, and a Decision Rule for 2026
Where Can Free Business Counts for TAM Be Found?
Public counts show how many firms exist by industry, size and region. Turning a count into a list of target accounts takes firm-level fields such as installed software, job titles and funding history, which commercial B2B data providers supply.
How Is SAM Calculated From TAM?
SAM shrinks TAM with filters. Each filter needs evidence from a count, a price list or a record of past deals. Four filters cover most B2B products:
- Industry. Keep the industries where the product solves a real problem. Company data by industry helps build the account list for each one.
- Region. Keep the countries and states where the team can sell, support and invoice. Company data by geography shows which firms sit in each area.
- Technology. Keep firms that run the software the product connects to. Company technology stack data lists firms by the software they run.
- Size. Keep firms large enough to have the person who approves the purchase. Employee range from the public counts handles this filter.
SAM = firms that meet all filters together × ACV
The filters should be applied on one list. Multiplying separate percentages treats the filters as independent, and industry and installed software often overlap, so the result can misstate SAM.
How Is SOM Calculated?
SOM starts with sales capacity, which is the number of opportunities a team can work in a year. New customers won per year equal the number of reps, times the opportunities each rep works per year, times the win rate. Multiplying by ACV gives SOM revenue.
SOM = reps × opportunities per rep × win rate × ACV
The win rate comes from the CRM: signed deals divided by all opportunities worked in the same period. Published benchmarks count deals in different ways, and some include open or abandoned deals in the denominator, so a team’s own figure is more reliable.
Two checks confirm that enough accounts exist to create those opportunities:
- Free to switch. Some SAM firms hold a multi-year contract with a competitor and will not buy this year. Only the share of SAM free to buy in the next 12 months can supply opportunities.
- Reachable buyer. A firm counts toward SOM only when someone with the buying title can be contacted. Decision-maker data by job title shows whether a named buyer exists for each SAM firm. Firms without one can be researched by hand, and each costs rep time.
Account order also affects results. Company job posting data shows firms adding roles in the function the product serves, and company funding data shows firms that recently raised money to spend. Neither signal proves a purchase, and each only moves likelier accounts to the top of the list.
TAM, SAM and SOM Example: A Worked B2B Calculation
The example uses a seller of accounts payable software to US businesses. Firm counts come from the SBA figures above. Rows marked assumed hold placeholder inputs for the arithmetic, and the seller’s own price, filter shares and win rate should replace them.
Counting all 36,228,171 firms at the same price would give a hypothetical $217.4 billion, 5.6 times the TAM above, because most of those firms do not fit the buyer definition. The team works 2,000 opportunities a year (8 × 250), about 0.4% of SAM firms, so capacity sets SOM long before market size does. A flat 1% of SAM would mean 5,133 wins a year, about 13 times what this team can close.
What Mistakes Make a TAM Calculation Wrong?
- Old counts. In 2023, about 1.3 million US establishments opened and about 1.2 million closed. Of new employer establishments from 1994 to 2022, 49.2% survived five years. A firm list loses accuracy every year.
- The same firm counted twice. Commercial lists can place one firm under two industries or two locations. Remove duplicates before applying a filter.
- Working backward from a target. Picking the TAM first and adjusting SAM and SOM to fit leaves numbers with no source behind them.
What Should Change When SOM Falls Short of the Target?
Suppose the revenue target is $3.6 million and the example above delivers $2.4 million, a $1.2 million gap (the target is assumed). Four inputs can close the gap. Each row below changes one input and holds the others fixed.
Widening SAM does not appear in the table because it adds nothing when the team already works under 1% of the firms available. Wider SAM matters only after the pool of reachable, free-to-switch accounts runs low. Win rate and opportunities per rep need no new hires, so the last column should be checked for those two before headcount is added. If the CRM shows deals lost at one stage, win rate is the first lever to test. If reps have no spare hours, headcount is the lever left.
Also Read: Data Quality Checks to Test Your Business Data Today
How Can a SAM Count Become a List a Sales Team Can Work?
A SAM count says how many firms fit. A sales team needs one record per firm, and each record needs six fields: firm name and website, industry code, location and employee range, the condition used as a filter (such as a software system), a named buyer with title, and a working email or phone number. Public statistics supply only the second and third fields, so the rest must come from research or a data provider.
The share of SAM firms that have all six fields is the reachable SAM. SOM should be rebuilt from that number, because opportunities cannot come from firms with no contact. Contacts also go out of date as people change jobs, so the reachable count needs a yearly refresh.
Any list should be tested before the plan relies on it. Draw a random sample of records and check three rates: how many match the filters, how many are duplicates, and how many contacts bounce or hold a new job. A sample file from a provider of verified B2B company data allows that test before a purchase. The numbers from the test replace the guesses in the SAM and SOM rows, and the plan then rests on measured rates.
FAQs
Does TAM include customers who already buy from competitors?
Yes. TAM and SAM include firms that use a competing product. SOM accounts for competition because it depends on how many of those firms the team can win.
How big does a TAM need to be?
No universal threshold exists. A TAM is large enough when SAM and SOM support the revenue target within the plan period, and the shortfall table above shows how to test that.
Is SOM the same as market share?
No. SOM is a number of accounts or dollars won in a set period. Market share is the portion of all buyers or all revenue a company already holds. They use different bases, so one cannot be read as the other.
How can a niche segment be counted when public statistics do not break it out?
Use public counts as the upper limit for the industry and size band. Then take a random sample of firms from that band, check what share meets the extra condition (for example, runs a given accounting system), and apply that share to the full count. The result is an estimate, and a larger sample narrows the error.
How is TAM calculated for a product category that does not exist yet?
Count the firms that do the task by hand today, then price the product against the cost of that manual work. The result is a value-based TAM and should be labeled as an estimate.
How should TAM, SAM and SOM be presented in a pitch deck or business plan?
Show one line per number with its formula, its source and the date of each count. Mark every assumed input as assumed. A reader can then check each step, which carries more weight than a larger total with no source.