# How to Estimate Market Size: A Guide to TAM, SAM and SOM

- URL: https://girisimcikafasi.com/en/blog/how-to-estimate-market-size/
- Published: 2026-10-17
- Updated: 2026-10-17
- Category: Starting and growing a business
- Publisher: Girişimci Kafası

John Graunt was a London haberdasher. By his own account, whenever the city's men of experience talked about how many people lived in London, the figure rarely dropped below millions.

One day a respected alderman went further. He claimed that in 1661 there were two million more people in London than in 1625, before the great plague. Graunt couldn't let the remark go, and he turned to the city's weekly bills of mortality.

The first check was simple. No more than 15,000 people a year were buried in London. If six million people lived there, only one in four hundred would die each year. For that age, it was an unbelievable level of health.

Graunt's own first quick guess, meanwhile, was 100,000. He then passed the same verdict on both: hasty thoughts, and both far from the truth.

So he rebuilt the calculation three separate ways. First he started from births: about 12,000 a year suggested 24,000 women of childbearing age and 48,000 families. At eight people a family, the result was 384,000. Next he counted families one by one in a few parishes; each year three families in eleven saw a death, which again pointed to 48,000 families. Last, he opened Richard Newcourt's map of 1658, divided the area inside the walls into squares a hundred yards across and assumed 54 families to each. That gave 11,880 families inside the walls; since a quarter of the deaths happened there, the whole city came to 47,520 families.

The three routes landed close together. In his book published in 1662, Graunt wrote that even the worst of his calculations showed well enough that there were no millions of people in London.

## Why estimate market size?

Startup pitches have the same dinner-table talk. "This is a multi-billion-dollar market" is an easy sentence to say, because nobody checks it.

According to Failory's guide, estimating market size is how you work out whether an idea is worth your time and money, whether to raise investment or grow on your own revenue, and how ambitious your goals can be. So the point isn't to find an impressive number, but a number realistic enough to make the right decision.

## Three rings: TAM, SAM and SOM

### TAM: the total market

TAM is the total revenue you'd reach by selling your product to everyone who needs it. It means taking a hundred percent of the market, and almost no company gets there. Still, it shows the ceiling for growth.

### SAM: the market you can reach

SAM is the slice you can actually reach with today's distribution, geography and channels. In Failory's example, a drink brand that wants to sell across the US but can only supply three West Coast states today calculates its SAM from the stores in those three states.

### SOM: the share you can win

SOM is the part you can realistically capture. When you have past sales data, Failory suggests a simple route: divide last year's revenue by last year's SAM, then multiply that ratio by this year's SAM. If you have no sales yet, write SOM down not as a target but as an assumption to test.

## Top-down

This method starts, like the alderman, with a big number and narrows it. You take the whole market from industry reports, then remove the slices that don't fit you.

Failory's example is a startup selling HR software worldwide. If 40 percent of the companies that could use the software already use a competitor's, the remaining 60 percent is your market. It's fast, but only as accurate as your data is current.

## Bottom-up

This method resembles Graunt's. You count real units and multiply: how many potential customers are there, and how much does each spend a year?

In Failory's example, someone selling wine in New York finds 200 shops that could stock it. If each shop buys an average of 5,000 dollars of wine a year, the market is 1,000,000 dollars. The number looks smaller but it can be defended, because behind every multiplier is something you counted.

## Value-based

For a product with nothing like it on the market, there's no past data to count. Then you estimate how much value customers will get from your product and what they'd be willing to pay for it. In Failory's example, someone selling a new kind of bicycle brake estimates how much more bike shops would pay for it than for the brakes they stock, and multiplies by the number of shops.

## Check your sums like Graunt

The real lesson isn't to pick one method, but to put at least two side by side. If the top-down number and the bottom-up number are far apart, one of your assumptions is wrong.

- **Ask it backwards:** like Graunt's "one in four hundred" question, look at what the world would have to be like for the number to be true.
- **Write down the multipliers:** keep each assumption on its own line and see how the result shifts when one changes.
- **Start with units you can count:** customers, shops, companies, things whose door you can knock on.

## The root of a guess, and its anchor

The Turkish word for estimate, "tahmin", comes from the Arabic taḫmīn; its root, ḫ-m-n, carries the sense of supposing and surmising. The same root gives "muhammen"; the "muhammen bedel" in Turkish tender notices means the estimated price. So from the start the word promises not certainty but a reasoned guess.

The sneakiest thing that spoils such a guess has a name: the anchoring effect. In Amos Tversky and Daniel Kahneman's study published in Science in 1974, participants were asked what percentage of the countries in the United Nations were African. Before answering, they watched a wheel spin. Those whose wheel stopped at 10 said around 25 percent; those whose wheel stopped at 65 said around 45 percent. A wheel with nothing to do with the question dragged the estimate along behind it.

The alderman's "millions" was an anchor of this kind. What saved Graunt was that instead of trying to correct that number, he started from scratch, from things he could count.

## Exactly wrong, or roughly right?

The English philosopher Carveth Read, in Logic: Deductive and Inductive, first published in 1898, says that on some subjects it is a mistake to aim at an unattainable precision, and adds:

> "It is better to be vaguely right than exactly wrong."

Graunt's 384,000 wasn't right to the last digit, and he admitted it might be somewhat at random. But it was far closer than millions. The market figure in your pitch: how many customers whose door you could knock on is it built on?

*Once you've sized the market, read [What Is Product-Market Fit?](https://girisimcikafasi.com/en/blog/what-is-product-market-fit/) to find out whether you're really gaining a foothold in it.*