How to Research Market Size Using Only Free Public Data

You don't need a paid research subscription to build a market size estimate that survives scrutiny. Government statistics agencies, public company filings, industry trade associations, and a bottom-up count of buyers are all free, and combined correctly they triangulate into a range that's more defensible than a single number pulled from one source. This guide walks through the method: define a segment, pull three independent free estimates, normalize them to the same unit, and document the assumptions behind the range you land on.

Quick Answer: Pull three independent estimates from free sources — a government/statistics source, an industry or trade-association or filings source, and a bottom-up count you build yourself — convert each to the same unit, currency, and time period, then triangulate the range where they overlap. Write down every assumption so someone else can check your work.

If you're earlier in the process and haven't validated the underlying idea yet, the complete guide to startup idea validation covers where market sizing fits relative to problem and demand validation — worth doing first, because a precise size estimate for the wrong segment isn't useful.

Prerequisites: Define Your Segment and Unit of Analysis First

Before opening a single data source, write down exactly who you're counting, what unit you're counting them in, and what time period the number covers — every downstream step depends on this being fixed before you start pulling numbers.

A market size figure is only as useful as the definition behind it. "The project management software market" is not a segment; it's a category umbrella that includes enterprise suites, freelancer tools, construction scheduling software, and dozens of adjacent product types with wildly different buyer counts and price points. Before you pull a single number, decide:

• Who counts as a buyer. A named job title, a company size band, a household income bracket, or a specific regulated entity type — whatever unambiguously includes or excludes a given prospect.
• What unit you're measuring. Number of buyers, number of transactions, dollars spent annually, or units sold. These don't convert into each other without an assumption (average price, purchase frequency), so pick one as primary and treat the others as derived.
• What geography and time period. A national estimate for last year and a five-country estimate projected three years out are not comparable, even when every other input matches.
• What currency and price basis, if you're sizing in dollars — nominal or inflation-adjusted, retail or wholesale, list price or realized price after discounts.

Skipping this step is the single most common reason two "market size" numbers for the same product disagree by an order of magnitude: they're not measuring the same thing. If you're doing this as a technical founder rather than someone with a research background, market research approaches built for technical founders covers how to scope a segment definition without a formal research background.

Where to Find Free Market Data: A Source Directory

No single free source will hand you a finished market size number for your specific segment. Instead, each source type is strong for a different piece of the puzzle, and the work is combining them correctly.

Use government and labor data to establish the size of the population you're selling into, use filings and trade associations to establish spend or revenue within that population, and use research summaries only to sanity-check the other two — never as your only source.

Step 1: Pull Three Independent Estimates

Get at least three numbers for the same segment from sources that didn't copy each other: a top-down population or spend figure, a supply-side figure from someone already selling into the space, and a bottom-up count you build yourself.

Three independent pulls, not three citations of the same underlying study, is what actually reduces your error. A common mistake is treating three articles that all cite the same trade-association report as three data points — they're one data point, cited three times.

A workable combination:

1. Top-down: a government or labor-statistics figure for the population or spending category your segment sits inside — for example, the number of establishments in a relevant industry code, or aggregate spending in an adjacent, broader category.
2. Supply-side: a public company's disclosed revenue in a comparable segment, or a trade association's membership count or published estimate, treated as one data point among several rather than the answer.
3. Bottom-up: a count you build yourself — the number of qualifying buyers you can enumerate or estimate, multiplied by a defensible price or spend assumption you can state explicitly.

If two of the three land in a similar range and the third is an outlier, investigate the outlier before discarding it. Sometimes it's using a broader or narrower segment definition than the others, which is informative in itself.

Step 2: Normalize Every Estimate to the Same Unit

Convert every pulled number to the same unit, currency, geography, and year before comparing them. An unconverted set of numbers isn't a comparison — it's three unrelated facts sitting next to each other.

This is where most of the real work happens, and it's mechanical rather than creative:

• Convert units. If one source reports buyer counts and another reports dollars, use your assumed average price or spend per buyer to convert one into the other, and note that conversion as an assumption, not a fact.
• Convert currency and adjust for inflation if sources span different years or countries.
• Convert geography. If a source covers a wider or narrower region than your segment, scale it using a defensible proxy — population share, GDP share, or establishment count share — and flag the result as an estimate, not a reported figure.
• Align time periods. A number from several years ago needs a stated growth-rate assumption to bring it forward; don't present a stale figure as current without saying so.

Every conversion you apply should stay visible in your working notes rather than getting folded invisibly into a single final number. Anyone reviewing your estimate should be able to see the raw source figure next to the adjustment applied to it.

Step 3: Triangulate Into a Single Defensible Range

Once your three estimates are normalized, look for where they overlap rather than averaging them blindly. The overlap is your range, and a wide spread between sources is itself useful information about how uncertain the segment really is.

Averaging three numbers that disagree by several multiples produces a figure that matches none of your sources and hides the disagreement between them. Triangulation instead asks where independent methods point to roughly the same territory.

• If all three estimates cluster within a reasonably tight band, that band is your working range.
• If one estimate is a clear outlier, revisit its assumptions before including or excluding it — don't drop it silently.
• If all three disagree substantially, report the range honestly instead of picking the middle. A stated range of "roughly X to Y, driven by differing segment definitions across sources" is more credible, and more useful to a reader making a decision, than a single false-precision number.

Once you have a defensible market size, the next step is usually splitting it into addressable and obtainable slices. How to calculate TAM, SAM, and SOM walks through that breakdown once you have a top-line figure to work from.

Step 4: Document Every Assumption You Made

Write down, next to your final number, every conversion, every proxy, and every judgment call you made to get there. An estimate without its assumptions attached can't be updated, challenged, or trusted by anyone but you.

At minimum, record:

• The exact segment definition — buyer type, geography, time period, unit.
• Each source you pulled, with a link or citation and the date you accessed it.
• Every conversion applied — currency, unit, geography, time — and the proxy or rate used.
• Which figures were reported directly by a source versus derived by you.
• The final range, and why you triangulated to that range rather than a single point.

This isn't bureaucratic overhead. It's what lets you, an investor, or a co-founder revisit the estimate in six months and update just the piece that changed, instead of redoing the whole exercise from scratch.

Worked Example: Sizing a Market From Public Data

The method holds whether you're sizing a broad category or a narrow niche. The figures below are hypothetical, included only to show how the pieces fit together — not a real market, and not real data.

Say you're sizing the market for a scheduling tool sold to independent physical therapy clinics in one country.

• Top-down: a labor-statistics source reports a total headcount for practicing physical therapists nationally. That's a population figure, not a customer count — most of those people work inside clinics rather than owning one, so you apply an illustrative ratio, such as roughly one clinic-owner per some number of practicing therapists, to estimate the number of clinics.
• Supply-side: a public scheduling-software company's investor materials disclose revenue attributable to healthcare-vertical customers, plus roughly how many clinics they serve, giving you an illustrative average revenue-per-clinic figure you can apply elsewhere.
• Bottom-up: you build a list of clinics from a public business directory or licensing registry, count them directly for a sample region, and extrapolate nationally using population share as a proxy.

None of the specific figures in that walkthrough are real data — they stand in for the shape of the calculation. In practice, each of the three legs above produces one estimate of total addressable spend. You normalize all three to "dollars per year, national, current," then triangulate the range where they land close together, flagging any leg that diverges sharply so a reader can see which assumption is doing the work.

Common Mistakes That Undermine a Free-Data Market Size

Treating a Stale Report as Current

A trade association report from several years ago, cited without a date, reads as current-year data to anyone who doesn't check the publication date. In a market that grows or shrinks quickly, an unstated staleness is one of the fastest ways to lose credibility with a reader who checks your source. Always cite the report's publication year next to the figure, and apply an explicit, stated growth assumption if you're bringing an older number forward.

Mismatching Segment Definitions Across Sources

The second recurring failure is combining sources that define the segment differently without noticing. A government industry code broader than your actual product category, a trade association's membership count that undercounts non-members, and a competitor's addressable-market claim that's aspirational rather than measured — none of these are wrong on their own, but averaging them as though they measure the same thing produces a number that isn't real. Before combining any two sources, write one sentence describing exactly what each one counts, and check that the sentences actually match.

Key Takeaways

• Define the segment before searching. Buyer type, unit, geography, time period, and currency all need to be fixed first, or your sources won't be comparable.
• Use several source types together: government and statistics agencies for population data, filings and trade associations for spend data, labor statistics for workforce proxies, and research summaries only as a sanity check.
• Pull three independent estimates, not three citations of the same underlying report — top-down, supply-side, and bottom-up.
• Normalize before comparing. Convert unit, currency, geography, and time period explicitly, and keep the raw figures visible alongside your adjustments.
• Triangulate to a range, not a false-precision point. A wide spread across sources is informative, not a failure.
• Document every assumption next to the final number so it can be checked and updated later.
• Watch for stale reports and mismatched segment definitions — the two most common ways a free-data estimate quietly goes wrong.

Frequently Asked Questions

Is free market data as reliable as paid market research reports?

Free public data can be just as reliable for triangulation purposes, though it takes more manual work to assemble than a single paid report hands you. Paid reports mainly save you the normalization and combination work — they don't have access to fundamentally better raw data than what's publicly available for most segments.

How many sources do I need before a market size estimate is credible?

Three independent sources is a practical minimum: enough to spot an outlier and triangulate a range. A single source, however authoritative, gives you no way to sanity-check whether its methodology or segment definition matches yours.

Where do I find data for a very narrow or new market niche?

Narrow niches often lack a direct government category or trade association, so lean harder on the bottom-up method — build a list of qualifying buyers from public directories or registries and count them directly, then use a nearby broader category's data only to sanity-check that bottom-up number, not to replace it.

How often should I update a market size estimate?

Revisit it whenever a key input changes materially, or at minimum once a year for a fast-moving segment, since even a well-documented estimate is only current as of the sources it was built from. Because you documented your assumptions and conversions in step four, an update usually means refreshing one or two inputs rather than redoing the whole exercise.