How to write a market analysis: from public data to a year-one customer count
BusinessPlanHub Editorial Team · Published
A market analysis is the section of a business plan that proves enough customers will pay your price. Its job is to produce one number: how many customers you will win in year one. Every other paragraph in the section exists to make that number believable, because the revenue forecast is built on it.
Most guides to this section explain TAM, SAM and SOM and then stop at the definitions. This one works an example from a public government dataset down to a year-one customer count and a revenue figure, so you can see where each filter comes from and which ones you have to research yourself. It ends with a fill-in box you can copy straight into the market analysis section of our template.
Why “1% of a big market” earns no credit
“The US market is worth $X billion, and we only need 1% of it” is the most common sentence in a weak plan. It fails for a simple reason. Nothing in your business produces that 1%. Not your sales team, not your location, not your marketing budget. The number is chosen, not derived, and a reader can tell.
A lender reading for repayment wants the opposite shape. The SBA’s lending rules ask for the assumptions behind projected cash flow, including the justification for revenue growth and a comparison with current industry trends, which our guide to what SBA lenders look for sets out in full. An investor wants to see a market large enough to matter and a credible first slice of it. Both readers are asking the same question: show me the customers, one filter at a time.
That is a bottom-up market size. It has three properties a top-down share never has:
- Every step names its source, so it can be checked.
- Every step can be challenged on its own, so one weak assumption does not sink the whole section.
- The last step is a count of customers, which converts directly into the revenue line of your financial projections.
TAM, SAM and SOM for a local business
TAM, SAM and SOM are three nested measures of the same market. TAM (total addressable market) is what every possible customer spends on what you sell in a year. SAM (serviceable available market) is the part your offer, price and location can actually serve. SOM (serviceable obtainable market) is the part you expect to win in a stated period, usually year one.
For a business that serves one town, county or metro, the TAM is local too. A dental-office cleaner in one county has no use for the national janitorial figure, and quoting it invites the “1%” problem by the back door. Build all three from a count of customers in the area you serve, multiplied by what one customer is worth per year.
Where the local numbers come from
Government statistics offices publish free counts of businesses and households by area. Two kinds matter. Business counts tell you how many potential customers there are if you sell to businesses, and how many competitors there are either way. Population and household data tell you how many potential customers there are if you sell to people.
| Country | Dataset | What it counts | Smallest area |
|---|---|---|---|
| US | County Business Patterns (CBP) | Establishments with paid employees, employment and payroll, by industry (NAICS) and employment size | County and ZIP code |
| US | Nonemployer Statistics | Businesses with no paid employees (mostly sole proprietors) and their receipts, by industry | County |
| US | American Community Survey (ACS) | People and households: income, age, housing, commuting and more | 1-year: areas of 65,000+ people. 5-year: all areas, including census tracts |
| UK | UK business: activity, size and location (ONS) | VAT- and/or PAYE-registered businesses by industry, employment size and turnover | Local authority district (broad industry groups) |
| UK | Census 2021 on Nomis | Population and households in England and Wales | Down to output areas |
| Australia | Counts of Australian Businesses (ABS) | Actively trading businesses by industry (ANZSIC) and employment size, with entries and exits | State and territory in the main release; further data cubes follow in December |
| Australia | Census QuickStats (ABS) | Summary population and household statistics for an area | Search by area name or address |
| Canada | Canadian Business Counts (Statistics Canada) | Business locations with employees by NAICS and size; locations without employees by NAICS | With employees: census subdivision. Without employees: province |
| Canada | Census Profile, 2021 Census | Population, households, income, housing, commuting | Down to census tracts |
Sources: each agency’s own page, linked above, checked 9 October 2026. The latest CBP reference year is 2023, released 26 June 2025. The ONS 2026 edition was released 24 September 2026 from a business register snapshot of 13 March 2026. The ABS release of 18 August 2026 counts businesses at 30 June 2026, with the next release due 18 December 2026. Canadian Business Counts are produced every June and December and released in August and February.
In the US, the simplest way in is data.census.gov: search for “County Business Patterns”, filter to your county and industry code, and the establishment counts by size band come back as a table. The SBA’s own market research guide lists further federal sources by topic, including the Bureau of Labor Statistics for income and employment.
Worked example: from a Census table to a year-one customer count
Suppose you are planning a commercial cleaning business that serves dental offices in one US county. Dental offices suit the method: they are a distinct industry code (NAICS 621210, Offices of Dentists), they need cleaning every evening, and the decision-maker is easy to reach. The figures below are illustrative, not real county data. Pull your own county’s numbers and replace each one.
Step 1: count every possible customer
CBP for your county, NAICS 621210, returns the number of dental establishments with paid employees. Say it is 420. That is the base of your TAM.
Step 2: filter by size
A two-person evening crew suits an office of roughly 5 to 49 staff. Larger practices tend to tender to big contractors. CBP publishes establishment counts by employment size band, so add up the bands that fit. Say 260 offices remain.
Step 3: filter by reach
You will only service offices within about a 20-minute drive of your base. CBP is also published at ZIP code level, so sum the ZIP codes inside that radius rather than guessing a percentage. Say 143 offices fall inside it. That is your SAM, counted in customers.
Step 4: filter by willingness to switch, with your own research
Here’s the thing: no government table tells you how many offices would consider a new cleaner. This is the step you research yourself. Phone 30 offices in the radius and ask whether they would take a quote at your price. If 9 say yes, that is 30%, which leaves about 43 offices open to a quote. Put the call log in your appendix. A small, real sample beats a large, invented one.
Step 5: convert, then check against capacity
If you expect to close about one quote in three, you win roughly 14 contracts in year one. Now test that against what you can deliver. If one crew can clean eight offices a night, two crews cap you at 16. The count fits. If the conversion step had produced 30, capacity would be the binding constraint, and your plan would have to say so.
Step 6: turn customers into the revenue line
Contracts do not all start on day one. If you win them evenly through the year, the average contract is billed for about six and a half months. At an illustrative $900 a month, year-one revenue is 14 × $900 × 6.5 = $81,900, with the business leaving year one at a run rate of about $151,200 a year.
| Step | Source | Offices | Yearly value at $10,800 each |
|---|---|---|---|
| All dental offices in the county (TAM) | CBP, NAICS 621210, county | 420 | $4,536,000 |
| Offices with 5 to 49 staff | CBP size bands | 260 | $2,808,000 |
| Within a 20-minute drive (SAM) | CBP by ZIP code | 143 | $1,544,400 |
| Open to a quote | Your phone survey, 9 of 30 | 43 | $464,400 |
| Won in year one (SOM) | Your close rate, checked against crew capacity | 14 | $81,900 billed in year one |
Look at what the table gives the reader. They can disagree with your close rate, or your survey, or your drive radius, and re-run the arithmetic with their own figure. That is exactly what you want an underwriter doing, because it means they are engaging with your numbers instead of discarding them.
The year-one win is about 10% of the offices in your SAM. That share is an output of the steps above, not an input. The difference is the whole point.
Selling to households instead
The same chain works for consumer businesses, using household data instead of business counts. A house cleaning service would start from the ACS 5-year estimates for the census tracts it can reach, filter to households above an income level using the household income table, and then apply its own research for how many already pay a cleaner or would start to. Use the 5-year figures for anything smaller than a large city: the 1-year estimates only cover areas of 65,000 people or more.
The house cleaning sample plan, set in Omaha, models $196,400 of revenue in year one. Divide a target like that by what one customer pays you in a year and you get the customer count your market section has to support. Work it in that direction too, as a check. If the count is larger than your SAM, the forecast is wrong, not the market.
Checking a forecast you already have
If your financial model already exists, run the example backwards. The janitorial services sample plan models $518,400 of year-one revenue for an office cleaner in Kansas City. The figures are fictional. At an illustrative $1,500 a month per contract, that is about 29 full-year contracts, or more if contracts start through the year. A reader of that plan would want the market section to show where those contracts come from: how many offices, of what size, inside what radius.
The dog grooming sample makes the same point at a smaller scale. It models $168,200 of year-one revenue for a Denver groomer. Divide that by a price per groom and by how often a regular customer comes back, and you have the number of dogs on the books that the market analysis needs to justify.
Five things that trip up a bottom-up count
- Establishments are not firms. CBP counts locations. A group of five dental practices is five establishments and one buyer. If one decision-maker controls several sites, count buyers, not sites.
- CBP leaves out the self-employed. It covers establishments with paid employees only. Sole traders appear in Nonemployer Statistics, which matters most when you count competitors.
- The data are old by design. The latest CBP year is 2023, released in mid-2025. Say which year you used, and do not present it as this year’s count.
- Small cells are protected. The Census Bureau applies disclosure protection to CBP so that no individual business can be identified, which makes the smallest areas the least precise. If a ZIP code shows a flag instead of a number, use the county figure and a stated share.
- The UK business data only see registered firms. The ONS release covers businesses registered for VAT or PAYE, so very small unregistered traders are missing from it.
Fill in your own count
Copy this into your market analysis and replace every bracket.
- Customer: [industry code, or household type and income band] in [area].
- Total count: [number] from [dataset, table, reference year]. Value per customer per year: [$]. TAM: [count × value].
- Size or profile filter: [rule], leaving [number]. Source: [size bands or income table].
- Reach filter: [radius or areas], leaving [number]. This is your SAM. Source: [ZIP codes, tracts or districts summed].
- Willing to buy or switch: [x of y asked], leaving [number]. Source: [your survey, pre-orders or letters of intent, in the appendix].
- Won in year one: [number], using [close rate], and within capacity of [number]. This is your SOM.
- Year-one revenue: [customers] × [price] × [average months or visits billed] = [$]. This must match the first line of your revenue forecast.
- Competitors serving the same customers: [names, prices, and the count from business data].
If line seven does not match your financial model, fix one of them before you write any prose around it. A market section and a forecast that disagree are the first thing a careful reader finds.
Your next step
Open data.census.gov, or your country’s equivalent from the table above, and pull the count for your industry code in your county or district. Write that single number into line two of the box, with its table name and year. The rest of the section builds from it.
Doing it yourself tends to stop working when the market is not a clean industry code: a product sold across several customer types, a national online business, or an investor plan that needs a credible TAM as well as a local SOM. Our business plan writing service builds the count and ties it to the financial model for you. If you would rather keep doing it yourself, the DIY toolkit links the market inputs to the revenue forecast so the two cannot drift apart.