Every business plan I’ve ever written has a line that says something like “the market for X is worth $Y billion.” I used to pull that number from the first Google result and move on. It felt like a formality, not a real input into any decision.
Then I started actually using market size numbers to decide things: whether a product idea was worth six months of my time, whether a niche was too small to bother with, whether I was pricing myself out of the buyers who actually existed. Once the number had to hold weight, “first Google result” stopped being good enough.
This is how I actually estimate market size now, using public data, without a market research subscription and without pretending I know more than the data supports.
What market size estimation is actually for
Market size estimation answers one practical question: if this thing works, how big can it get. Not “how big could it theoretically get in a perfect world,” but “how many people or businesses could plausibly buy this, and how much money is realistically in that pool.”
You’ll see this broken into three layers, usually called TAM, SAM, and SOM:
- TAM (total addressable market): everyone who could ever want this category of product, worldwide, no constraints.
- SAM (serviceable addressable market): the slice of TAM you could actually reach given your product, geography, language, and pricing.
- SOM (serviceable obtainable market): the slice of SAM you could realistically capture given competition, your resources, and time.
For a solo operator, TAM is mostly useless. Knowing that the global project management software market is enormous tells you nothing about whether a niche tool for wedding photographers has a viable audience. SAM is where the real thinking happens. SOM is where you have to be honest about the fact that you are one person with limited hours, not a funded team.
The mistake I made for years was quoting TAM and calling it market research. It made pitches sound good and told me nothing about whether to build the thing.
Where the public data actually lives
You don’t need a paid report to do this well. You need to know where to look and how to read what you find.
Government data. In the US, the Census Bureau publishes business counts by industry code (NAICS), including how many establishments exist in a given category and roughly how many people they employ. The Bureau of Labor Statistics publishes wage and employment data by occupation, which is useful if your buyer is a type of professional rather than a type of company. Most countries have an equivalent statistics agency. This data is slow to update and coarse, but it’s a real count, not a guess.
Industry associations and trade publications. Almost every niche has a trade group or trade magazine that publishes some version of “state of the industry” content, often free or behind a light registration wall. These numbers come with their own biases (associations want their industry to look healthy) but they usually cite a source, which means you can trace where the number came from and judge it yourself.
Search and keyword data. Tools built for SEO, including free tiers, show you search volume for terms related to your product. This isn’t market size, but it’s a proxy for demand intent. If almost nobody searches for the problem you solve, that’s worth knowing before you build anything.
App stores and marketplaces. If your competitors sell through an app store, a marketplace like Etsy, or a plugin directory, you can often see review counts, install estimates, or sales badges. Review counts aren’t sales counts, but they give you a floor. A competitor with 4,000 reviews has sold at minimum many multiples of that.
Job postings. If you sell to businesses, job posting sites tell you which companies are hiring for the role your product supports. A spike in “growth marketer” postings at small ecommerce brands tells you something real about where budget is flowing, without anyone publishing a market report on it.
None of these sources is complete on its own. That’s the point of the next step.
Building a bottom-up estimate
Top-down estimation starts with a huge global number and slices it down by percentage. Bottom-up estimation starts with a real, countable unit and multiplies up. Bottom-up is slower but much harder to fool yourself with, because every input is something you can point to.
Here’s the shape of it, walked through with round, illustrative numbers rather than a real dataset, just to show the mechanics.
Say you’re evaluating a tool for independent bookkeepers who serve small businesses. Your bottom-up chain might look like this:
- Find a count of the target unit. Census or BLS data might show a certain number of people employed under the “bookkeeping, accounting, and auditing clerks” occupation code, plus a separate count of self-employed bookkeepers if that’s tracked separately.
- Narrow it to your actual buyer. Not everyone in that occupation code is self-employed or running their own client base. You might apply a reasonable percentage, based on what industry sources say about self-employment rates in that field, to get to “independent bookkeepers” specifically.
- Estimate what they’d pay. Look at what comparable tools charge (a public pricing page is public data too), and estimate an average annual spend per user.
- Multiply the buyer count by the estimated spend to get your SAM.
- Apply a realistic capture rate for SOM. A capture rate of 1 to 3 percent of SAM in a first few years is a common planning assumption in a lot of go-to-market writing, not a guarantee, and you should treat it as a stress test, not a forecast.
The number that comes out the other end is not a prediction. It’s a boundary. It tells you the ceiling is probably not high enough to matter, or that the ceiling is high enough that the real constraint is something else, like distribution or trust, not market existence.
Why top-down alone misleads you
Top-down estimation (start with “the global SaaS market is worth $X” and take 0.1 percent of it) feels rigorous because the starting number is big and citable. But the slicing step is almost always a guess dressed up as math. There’s no defensible reason 0.1 percent is the right share instead of 0.01 percent or 1 percent. The precision of the final number hides the fact that the one input that actually matters was invented.
I still use top-down, but only as a sanity check against the bottom-up number, never as the primary estimate. If the two methods land in wildly different places, that’s a signal to dig into why, not a reason to average them and move on.
What a market size number can’t tell you
A market size estimate tells you how many potential buyers exist and roughly how much money moves through that category. It does not tell you:
- Whether those buyers will trust a solo operator over an established brand
- Whether you can reach them without a marketing budget
- Whether the market is already served well enough that switching costs kill your chances
- Whether the number will still be true in three years
I’ve sized markets that looked plenty big on paper and then discovered the real constraint was that buyers in that niche only trust referrals from people they already know, which is a distribution problem no market size number will ever surface. Market size answers “does this room have enough people in it.” It says nothing about whether you can get in the door.
A simple way to keep yourself honest
Every time I run this exercise now, I write down where each number came from, in a plain list next to the estimate itself. Census table number, competitor pricing page URL, the exact search term and volume. If I can’t point to the source six months later, I don’t trust the number six months later either, and I redo it before making a decision based on it.
That habit alone has stopped me from repeating the mistake that started this whole approach: quoting a number I found once, liked, and never checked again.
If you want more of this kind of walkthrough on the metrics and methods behind running a one-person business, check out the rest of what we cover here.
Get new guides and videos first — join the Telegram channel.