How to Size a Market Using Free US Census Data

The US Census Bureau publishes the most accurate market data in the world, and founders ignore it because the interface looks like a tax form from 1998. You do not need to pay thousands of dollars to a market research firm to size a market. You just need to know which three free datasets to query and how to read the output. We are going to walk through the exact data sources founders use to validate B2B and B2C markets. We will cover County Business Patterns, the American Community Survey, and Nonemployer Statistics. Then we will run the math on a specific local service market to show you exactly how this works in practice.

County Business Patterns for B2B Markets

County Business Patterns is your primary tool for B2B market sizing. It counts every business establishment in the United States with paid employees. It breaks these businesses down by industry, geography, and employee count.

Are there enough target businesses in this specific area to support my software or agency?

The government classifies businesses using the North American Industry Classification System, known as NAICS. Every industry has a two-digit to six-digit code. For example, 23 is Construction. 238 is Specialty Trade Contractors. 238220 is Plumbing, Heating, and Air-Conditioning Contractors.

If you are building dispatch software for HVAC companies, you do not need to guess your total addressable market. You go to data.census.gov, search for County Business Patterns, filter by NAICS 238220, and filter by your target counties. The output will show you exactly how many plumbing and HVAC businesses exist there. More importantly, it shows you their size. You will see how many have one to four employees, how many have five to nine, and how many have over fifty. If your software costs five hundred dollars a month, a shop with two employees cannot afford it. County Business Patterns tells you exactly how many shops hit your required threshold of ten employees.

The American Community Survey for B2C Demographics

The American Community Survey maps consumer demand. While the decennial census only counts heads every ten years, the American Community Survey samples three million households every single year. It covers income, commute times, housing age, internet access, and education.

Do the residents in a specific zip code or county match my ideal customer profile?

The American Community Survey comes in two versions. The one-year estimates are more current but only cover areas with populations over 65,000. The five-year estimates pool sixty months of data to cover every single geographic area in the country, right down to the neighborhood level. Always use the five-year estimates for local service businesses. The statistical margin of error on small geographies in the one-year data makes it useless for precise targeting.

You can use this data to find high-income households living in older homes that need repairs. You can find areas with extreme commute times where residents have no time to cook, making them perfect targets for a premium meal prep delivery service. You combine these variables to build a deterministic map of your buyers.

Nonemployer Statistics for the Shadow Market

County Business Patterns only counts businesses with payroll. It completely ignores solo founders, freelance designers, independent truck drivers, and one-man landscaping crews. Nonemployer Statistics fills this gap. It counts businesses that have no paid employees and are subject to federal income tax.

Is there a viable market of solo operators I can sell to?

This matters because the solo market requires a completely different product and pricing strategy than the employer market. If you build a tool for freelance graphic designers, County Business Patterns will tell you the market is tiny. Nonemployer Statistics will reveal millions of potential users. You look up NAICS 541430 for Graphic Design Services in Nonemployer Statistics to see the true count of solo designers. You can also view their aggregate receipts to calculate the average revenue per solo operator. If the average solo designer in a county only makes thirty thousand dollars a year, they are not buying a premium enterprise product.

Navigating the Census Bureau Interface

The data is free. Extracting it requires patience. The primary portal is data.census.gov. Do not use the basic search bar on the homepage. It attempts to guess what you want and fails almost every time. Always click on the Advanced Search link.

Your workflow should always follow the same three steps.

1. Select your geography. Click Geography, then choose County or Census Tract. Select your state, then select your specific counties.
2. Select your topics. If you want business data, click Codes, then Industry and NAICS. Browse the tree until you find your target industry.
3. Extract the data. Click Search and navigate to the Tables tab.

The tables will have cryptic names like CB2100CBP. Ignore the names. Look at the column headers. You will see columns for Number of Establishments and columns for Employment Size of Establishments. Download the table as a CSV file. The web interface is too clunky for serious analysis. Open the CSV in a spreadsheet, delete the margin of error columns, and you have a clean dataset ready for market sizing.

Worked Example: Sizing a Local Landscaping Market

Let us apply these three datasets to a real scenario. You want to launch a premium, tech-enabled residential landscaping and hardscaping service in Fairfax County, Virginia. You need to know if the market is large enough and how many competitors already exist.

Step One: Proving Consumer Demand

You go to data.census.gov and load the most recent five-year estimates for Fairfax County. You are looking for two variables. You need households earning over two hundred thousand dollars a year, and you need single-family detached homes. The data shows Fairfax County has roughly four hundred thousand households. About one hundred thousand of those households earn over two hundred thousand dollars annually. Further filtering shows that roughly seventy percent of these high-earning households live in detached single-family homes. You now have a hard ceiling of seventy thousand target households. If you capture just one percent of them, you have seven hundred recurring clients.

Step Two: Analyzing the Competition

You search for NAICS 561730, which covers Landscaping Services. The table for Fairfax County shows roughly four hundred and fifty employer establishments in this category. You look at the establishment size breakdown. Over three hundred of these companies have fewer than five employees. These are small crews, likely owner-operators running one or two trucks. There are only about twenty companies with more than fifty employees. The market is highly fragmented.

Step Three: Checking the Shadow Market

You run the same NAICS code for Fairfax County in the Nonemployer Statistics database. The data reveals over two thousand nonemployer landscaping businesses. These are the solo competitors.

The synthesis is clear.

• You have seventy thousand wealthy households living in detached homes.
• You have four hundred and fifty established competitors, but most are tiny.
• You have two thousand solo operators.

The market is massive, fragmented, and flooded with small players who likely do not answer their phones or use modern billing software. You have a validated market gap. You can build a premium brand that wins on reliability and software integration. If you were building an app to help these exact solo operators manage their routing, you would use Unthinkable to cross-reference this demographic data with real demand signals from landscapers posting about broken scheduling software on web forums.

The Reality of Publication Lag

Census data is never real-time. It is historical data disguised as current data. You must understand publication lag before you build a business model around these numbers.

County Business Patterns has a two-year lag. If you access the database in 2024, you are looking at data collected in 2022. The American Community Survey one-year estimates have a one-year lag. The five-year estimates are a rolling average of the past five years, meaning the data point for a specific neighborhood is essentially three years old. Nonemployer Statistics also carries a two-year lag.

When does this matter? It matters immensely if you are targeting a boomtown. If you look at American Community Survey data for an exurb of Austin or Boise today, the five-year estimates will severely undercount the population and income growth that occurred over the last twenty-four months. The data will tell you the market is too small, but your eyes will tell you they are building three new subdivisions. In high-growth areas, you must supplement Census data with local building permit data.

When does publication lag not matter? It is irrelevant for established markets and mature industries. The number of plumbing businesses in Cleveland, Ohio does not swing wildly from year to year. The median age of housing stock in Boston, Massachusetts is not going to change because of a two-year data delay. If you are sizing a stable B2B market or a traditional home services market in a developed metro area, the two-year lag is a rounding error. The structural realities of the market remain identical.

Stop guessing how big your market is. The federal government spends billions of dollars a year collecting this data. Learn the NAICS codes for your industry. Pull the American Community Survey tables for your target counties. Run the math on the actual number of buyers who can afford your product.