I built a market study last week for coin-operated laundries across Nevada and Utah. Then I priced what it would have cost to commission it, because that number turns out to be the least interesting thing about the exercise.
Here is what got built. Twenty-five scripts, about 3,900 lines. Two hundred and seventy-four ZIP-level trade areas and 247 places scored on demand. Seventeen upstream data sources, from three Census programs to OpenStreetMap to seven separate municipal water and sewer rate schedules. Three finished documents with a financial model at the end. Start to finish it ran from about three in the afternoon to seven in the evening.
What a human charges for that
I am not guessing at the hours. I am adding up the work that actually happened and pricing it at what people charge.
| Phase | Hours | Why it takes that long by hand |
|---|---|---|
| Methodology design | 8 to 16 | Choosing the demand denominator, deriving a benchmark, designing a score you can defend |
| Data acquisition | 20 to 40 | Seventeen sources, two Census programs that use different NAICS digit counts, twelve demographic tables across three states, and seven city utility rate sheets that are each formatted differently |
| Joining and analysis | 15 to 25 | 274 trade areas joined against three separate supply sources, distance rings, suppression handling |
| Financial model | 10 to 20 | A pro-forma driven off the actual local water and sewer rates |
| Producing the documents | 15 to 25 | Three deliverables, charts, formatting |
Call it 90 hours. At that size, here is the market:
| Who does it | Typical rate | What it costs you |
|---|---|---|
| Freelance analyst | $60 to $125 an hour | $5,400 to $11,250 |
| Boutique market research or site-selection firm | $150 to $250 an hour, usually quoted as a fixed fee | $15,000 to $35,000 |
| National real-estate advisory, team-staffed | $300 to $500 an hour | $50,000 to $75,000 |
The substitutes do not actually substitute. An industry report from one of the big research publishers runs about $1,000 to $1,500 and is national, so it will never tell you which ZIP code. A location-analytics subscription runs $10,000 to $35,000 a year and hands you demographics rather than an answer. A site-selection consultant charges roughly $2,500 to $7,500 per site, and this covered 274 trade areas. Equipment distributors will run you a free demographic study, but they are selling you machines, so read it knowing which way the incentive points.
The expensive version is usually wrong
This is the part worth your attention, and it is not a story about speed.
The whole study turns on one obscure fact. County Business Patterns, the federal file that nearly every market study is built on, only counts businesses that have employees on payroll. Coin-operated laundries are mostly unattended, so they have no payroll and the file cannot see them. Of the 20,293 laundromats in the United States, 9,403 do not appear in it at all.
If you do not know that, you sum ZIP Business Patterns across every ZIP in the country, get 5,234 stores, and build everything on top of a number that understates the market by 3.9 times. In one specific direction: it makes every trade area look emptier than it is. You would pay $15,000 for a study that tells you to open where there is already competition.
Two more of the same shape. Federal agencies suppress any figure that would identify an individual business, so a county with two laundromats reports zero. A suppressed zero and a genuine zero look identical in a spreadsheet, and one of them is a market and the other is not. And if you model income as "lower is better," which is the intuitive way, you rank the poorest ZIP in the state first. Demand for self-service laundry actually peaks in a band, roughly $30,000 to $65,000, because below it people cannot spend and above it they own machines.
None of those three are exotic. They are just things you either know or you do not, and the invoice looks the same either way.
So the honest comparison is not 90 hours against four. It is 90 hours for an answer that is probably wrong in a specific and expensive direction, against an afternoon for one where the benchmark was derived from the data rather than borrowed, and every zero was checked against an independent roster before being believed.
What actually changed
The data has been free and public this whole time. The Census API key takes a minute to request and costs nothing. OpenStreetMap is free. The bulk demographic files need no key at all. Nothing in this study sat behind a paywall.
What changed is that the cost of using it collapsed. The barrier was never access. It was the 90 hours, and the fact that those hours had to belong to someone who already knew which federal file lies to you and how.
That is the shift worth naming. Not that a machine wrote some code quickly, but that a category of analysis which was priced as a five-figure consulting engagement is now something a small business owner can run on a Tuesday. The same thing already happened to website audits, to contract review, to bookkeeping. Market research was simply further down the list.
What it still does not do
I would rather be straight about the ceiling than oversell this.
It cannot see quality. Four tired laundromats may leave a trade area more open than two good ones, and no federal file records which is which.
It lags. The demographic estimates are a rolling five-year average, the business counts run about two years behind, and a ZIP that added 2,000 apartments last year still looks like it did before they were built.
It cannot see leases. Plenty of genuine white space is white because no suitable building exists in it at a rent that works.
And it cannot tell you whether the gap is real or explained. A metro where most apartments already include in-unit laundry should carry fewer laundromats, and that is correct rather than an opportunity.
What comes out is a ranked shortlist of hypotheses. Somebody still has to drive those trade areas and look at the buildings. The work after the shortlist is still the actual work, and it is still yours.
The difference is where you start. You used to start by deciding whether the question was worth $15,000. Now you start with 274 trade areas already ranked, and you spend your money on the five that survived.
If you want the method rather than the price tag, the companion piece walks through the whole thing step by step, including the three traps above and the exact API calls, so you can point it at any business type in any part of the country.
Related reading
- How to Size Up Any Local Business Market With Public Data: the full method behind the study priced above, with the federal API calls and the traps that make the data lie to you.
- The Free US Small Business Help You Are Already Paying For: the public advisory network that will review an analysis like this with you at no cost.
- How a Small Business Runs AI Agents Without a $47,000 Surprise Bill: the other side of the cost story, and how the cheap version stops being cheap.
- Claude for Small Business Just Shipped: a plain-English walkthrough for a company with no technical staff.
- AI Employees for Small Business: 10 Real Stacks: documented cases of the same substitution happening in other functions.
Fact-check notes and sources
Consulting rates are market ranges rather than published prices and vary by region and scope; the federal figures below are exact and were pulled on 10 August 2026.
- Laundromat counts: 10,890 establishments with payroll from County Business Patterns 2023 (NAICS 812310), plus 9,403 without payroll from Nonemployer Statistics 2022 (NAICS 81231), totalling 20,293. Both via the Census Data API.
- The 3.9x undercount: summing the coin-operated laundry line across all ZIPs in ZIP Business Patterns 2023 returns 5,234 against the 20,293 above.
- Disclosure suppression: the Census Bureau suppresses cells that would disclose an individual establishment, per its statistical safeguards policy. Three counties in this run reported zero from both federal programs while public directories listed operating stores.
- Scope of the study described: 25 scripts totalling roughly 3,900 lines, 274 ZCTAs and 247 places scored, 17 distinct upstream hosts, three PDF deliverables. Measured directly from the project directory.
- Hours and rates: my own estimate of the labour, broken down by phase above, priced against commonly quoted US rates for freelance analysts, boutique research firms, and national advisory practices in 2026. Treat the dollar figures as ranges for comparison, not as quotes.
This post is informational, not investment or business advice. The cost ranges are estimates for comparison and are not quotes from any named firm. No research publisher, analytics vendor, consultancy, or distributor referenced generically here is affiliated with this site, and no criticism of any specific company is intended or implied.
