What a finding means
We keep three things apart. A statistical finding is a fact: a county has fewer of one kind of business than counties like it. A market hypothesis is a reason it might matter: there may be unmet demand. A recommendation needs more evidence than the Census can give. The map stops at the first two. It tells you where to look. It does not predict that a business will succeed.
Sources and dates
- Businesses with employees: Census County Business Patterns 2023 (counts as of March 2023).
- One-person businesses: Census Nonemployer Statistics 2023, for the types where the Census publishes them.
- Population and income: American Community Survey 2019 to 2023, 5-year estimates.
- Population change: Census Population Estimates, Vintage 2024 (July 2020 to July 2024).
- Towns: Census Vintage 2024 place estimates.
- Startup costs and survival: the Groundwork Startup Industry Database.
Data read October 25, 2026. Counts are history, not today. A shop that opened or closed since early 2023 is not in them.
The Peer-County Method
For each county we find the 60 U.S. counties nearest it in population and median household income, with the two weighted equally in logarithms. We scale their counts to the size of the county and show the typical count and the range the middle half of them fall in. A type is marked fewer when the county sits below that range by at least 2 businesses and at least a quarter of the typical count. It is marked more under the same test above the range.
Small counts and the “not rated” label
The Census does not publish a business type in a county when it has fewer than 3 establishments. We never turn that into a zero. We show a range. A type whose count is a hidden range is never marked fewer. It is marked not rated, because our own checks found listings above the Census range for owner-run shops. Florists, gyms and bookstores are mostly owner-run and the Census counts only those with employees, so these three are never rated fewer, even when the count is exact. Of 9,380 county-and-type pairs, 456 are rated fewer, 1,332 are rated more and 267 are not rated.
What is counted
The Census sorts each business into one industry code. A gym, a fitness studio and a hospital wellness center can land in different codes, and a store inside a bigger store counts under the bigger store. Where the Census publishes one-person businesses we add them. Where it does not, we use businesses with employees only, for the county and its peers alike. We removed bakeries, grocery stores and pharmacies because the codes did not match what people see on the street.
Parishes
Louisiana has parishes, not counties. The Census counts them as county equivalents, and so do we.
What can mislead
- People shop across county lines. A county can have fewer shops because customers drive next door, or more because it serves its neighbors.
- Small counties swing. One shop is a large share of a county with 5,000 people.
- A gap shows where to look. It does not show that customers are waiting.
Audit log
Every check, what it found and what we changed. Listing checks use Google Maps and directories, which have their own errors, so we treat a difference as a reason to look, not as proof.
| Checked | How | Result | Action |
|---|---|---|---|
| Full recompute, Arkansas and Texas | Every county count (9,380 county-and-type pairs), every peer list (469 counties) and every fewer/more rating was rebuilt by separate code from the raw Census files and compared with what the pages say. | No differences. | None needed. |
| Jefferson County, AR: gyms | Census hides a count under 3 and showed a range of 0 to 2. Directories and Google Maps showed 5 gym listings plus 2 hospital wellness centers. | Listings found more than the Census range. | A hidden count now shows as a range, never zero. The type is left not rated or in line, never fewer. |
| Jefferson County, AR: coffee shops, pet grooming | Census counts compared with listings in Pine Bluff and White Hall. | Matches. | Marked Checked locally on the county page. |
| Garland County, AR: bakeries | Census range of 0 to 4 against about 9 bakery listings in Hot Springs. | Listings found more than the Census range. | Bakeries removed. Grocery stores and pharmacies removed for the same classification problem. |
| Lonoke County, AR: hardware stores | Census range of 0 to 2 for hardware stores against 5 to 8 listings in Cabot, Lonoke and Beebe. | Listings found more than the Census range. | Widened to hardware and building supply stores, which the Census counts as 7 for Lonoke. |
| Crawford County, AR: veterinarians; Hot Spring County, AR: hardware | Census counts of 4 and 4 against about 5 veterinary practices and 1 to 2 hardware listings (plus farm supply). | Consistent. The Census count was equal to or above the listings. | None needed. |
| Maverick, Ector and Wood Counties, TX: dentists, furniture | Census counts of 6, 7 and 7 against about 5, 5 and 6 listings. | Consistent. | None needed. |
| Liberty and Van Zandt Counties, TX: gyms | Census range of 0 to 2 in both. Listings showed about 2 in Liberty and about 6 in Van Zandt. | Van Zandt listings exceed the Census range. Many gyms are owner-run studios with no employees. | A hidden count is never rated fewer. It is shown as a range and marked not rated. By-type pages rank full counts only. |
| St. Tammany Parish, LA: florists; Acadia Parish, LA: hotels; Jefferson Davis Parish, LA: dentists | Census counts of 4, 5 and 5 against 10 or more florist listings, about 5 hotel listings and about 5 dental practices. | Florists: listings far above the Census count. Hotels and dentists: consistent. | Florists and gyms are never rated fewer, because the Census counts only the ones with employees. They have no by-type page. |
| Nueces County, TX: bookstores; Calcasieu Parish, LA: pet stores; Lafourche Parish, LA: sporting goods | Census counts of 4, 3 and 3 against about 7 bookstore listings (many independent), 4 pet store listings and 1 to 2 sporting goods listings. | Bookstores: listings above the Census count. Pet stores and sporting goods: consistent. | Bookstores join florists and gyms as never rated fewer. Pet stores and sporting goods stay rated. |
| Logan, Delaware and Creek Counties, OK: hardware, dentists, veterinarians | Census counts of 4, 5 and 5 against about 3 hardware and lumber listings, about 6 dental practices and about 5 veterinary practices. | Consistent. | None needed. |
| Rating rule | Fewer or more also needs a gap of at least a quarter of the typical count, and a hidden count is never rated fewer. | Arkansas fewer ratings fell from 108 to 46 after the new rules. | Rules applied to every state and published in the method. |
Audits run October 25, 2026. Listing checks cover a small share of all ratings. Pages with a listing check say Checked locally. Everything else rests on Census counts alone, and the full recompute confirms the pages match those counts.
Keeping it current
We refresh the counts when the Census publishes a new County Business Patterns year, and we run the full recompute before a new state goes up. Pages show the date the data was read.
Report a problem
Every county page has a correction form. We check the report against the source, fix the page and date the change in this log. Our corrections policy applies.
