TAM, SAM and SOM: how to size a market by counting the customers in it

TAM, SAM and SOM are three nested estimates of market size. TAM (total addressable market) is the revenue you'd make in a year if every possible customer bought from you. SAM (serviceable addressable market) is the part your product, price and channels can actually reach. SOM (serviceable obtainable market) is the part you can realistically win in the next few years, with the team you have.

The most reliable way to work them out is bottom-up: count the customers, multiply by what one of them pays in a year, and cut the count down at each step. Starting from an analyst's headline figure and taking a percentage of it is quicker, and it's the version investors, and your own team, should trust least.

TAM, SAM and SOM definitions

Two details matter more than they look. All three are annual figures, so use what a customer pays a year, not lifetime value or a one-off fee. And all three are counted in customers before they're counted in dollars. A figure that can't be turned back into a number of customers is a guess with a currency sign on it.

Top-down vs bottom-up market sizing

Top-down sizing starts with a big published number ("the accounting software market is worth $X billion") and takes a slice of it: "if we win 1%, that's $Y million". It's fast and sounds rigorous, and it tells you very little about your business. The headline figure usually includes products you don't sell and customers you'll never reach, and the 1% isn't connected to anything you can do.

Bottom-up sizing starts with the customers. How many companies look like the ones who'd buy? What does one of them pay a year? How many can your team actually sign? Every number in it can be checked, and every number can be wrong, which is exactly what makes it useful. When the market turns out smaller than you hoped, a bottom-up model tells you which assumption to go and test.

A bottom-up TAM, SAM and SOM example, with real counts

Say you sell software that handles client document collection and filing deadlines for accounting firms in the US, priced per office by size. The office counts below are real. The prices and the sales capacity are made up to show the method.

Step 1: count every possible customer

The Bureau of Labor Statistics' Quarterly Census of Employment and Wages (QCEW) counted 62,838 private establishments under NAICS 541211, Offices of Certified Public Accountants, in the first quarter of 2025. It also breaks them down by number of employees, which is what lets you price each band differently:

Step 2: cut it down to the customers you can serve

Say the product is built for firms that have outgrown spreadsheets but don't have an IT team of their own, and it sells through a free trial and a short demo. That suits offices of 5 to 49 people. Offices under five rarely pay for workflow software, and offices of 100 or more expect security reviews and custom contracts your team can't handle yet.

That leaves 16,230 offices, worth about $100.7M a year. That's the SAM.

Step 3: work out what your team can actually win

This is the step most pitch decks skip, and it's the one that should drive the plan. Say you have two salespeople, each running about 40 demos a month, and one demo in four turns into a customer. That's 20 new customers a month, 240 a year, and about 720 after three years if nobody leaves. At the SAM's average price of about $6,200, that's roughly $4.5M a year.

That's the SOM: about 4% of the SAM, from a model where you can see exactly what would have to change to make it bigger.

One caveat about the data. The QCEW counts establishments, not firms, so a practice with three offices shows up three times, and it only counts businesses with employees. For a product priced per office, that's close to what you want. For one priced per firm, you'd want a firm count instead, which the Census Bureau's Statistics of US Businesses publishes.

How do you calculate SOM?

From capacity, not from a percentage. A SOM of "3% of SAM" is a target dressed up as an estimate, and nothing in the business connects to it.

SOM = customers your team can sign in the period × annual price

The first half comes from how many sales conversations you can have and how many of them close. If the answer is far below the SOM you wanted, you've learned something more useful than any market report: the limit is sales capacity, not market size, and the plan should be about that. If you have no close-rate data yet, use a range and say so. In the early months, the founder's own sales numbers are usually the only honest input.

Where to find the counts

For US businesses, the free public sources are the Bureau of Labor Statistics' QCEW and the Census Bureau's County Business Patterns, both broken down by industry code and company size. Most countries' statistics offices publish something similar.

For narrower markets, counts usually come from a directory or a search: LinkedIn's company filters, a trade association's member list, an app marketplace's list of companies using a particular tool. These are less complete than a census, but often closer to your real customers than an industry code, which lumps together every business that ticked the same box on a form.

The test for any count is whether you can name a handful of the companies in it. Pick ten at random from your SAM. If most of them would plausibly buy, the count is good. If half are obviously wrong, the definition is too broad, and your ideal customer profile needs another pass before the market size does.

What investors look for in TAM, SAM and SOM

On a pitch deck slide, the three numbers matter less than the working behind them. A bottom-up model shows you know who your customers are and how you'll reach them. A top-down one mostly shows you found a report.

Expect the questions to land on the assumptions. Where does the count come from? Why that price, and do current customers really pay it? Why that slice for the SAM? And does the SOM match the hiring plan on the next slide? If the SOM says 720 customers in three years and the plan has one salesperson, someone will notice.

Mistakes that make the numbers meaningless

• Counting users instead of buyers. The same census counts about 546,000 people working in those accounting offices. None of them is your customer if the office buys the software rather than each person.
• Using industry revenue as the TAM. What accounting firms earn isn't what they spend on software. Your TAM is spend on the thing you sell.
• Mixing one-off and annual prices. All three figures are per year. A setup fee doesn't belong in them.
• A SAM that's just a smaller percentage. The SAM should be a definition you could hand someone and get a list back, like "accounting offices in the US with 5 to 49 employees".
• Never updating it. Your first SAM is a guess about who'll buy. Six months of sales tell you who actually did.

When TAM, SAM and SOM are the wrong question

When nobody spends on the category yet. If you're creating a new kind of product, there's no existing budget to count, and the honest version is a count of companies with the problem plus an estimate of what solving it is worth to them.

When you're very early. Before your first ten customers, a precise SOM is false precision. What matters more is whether you can name a hundred accounts that look like your best customer and reach them, which is a go-to-market question more than a market-sizing one.

And when the market is obviously big enough. If your SAM is thousands of companies and you have twelve customers, the next year's work is finding and winning the next fifty. That's about knowing which accounts have a reason to buy now, not about the size of the pie.