Tony Wang7 min read866 Businesses Share One Wyoming Address. The Denser Ones Are All Hospitals.
We found the densest points in US business data at 3.7cm precision. Three unrelated things produce the same signature, and only one is what you'd guess.
What is actually at the densest points
We aggregated every US listing carrying coordinates into geohash cells at precision 12 — about 3.7cm on a side, so a "cluster" means businesses sharing a point smaller than a dinner plate.
| # | Listings | What is there | Dominant category |
|---|---|---|---|
| 1 | 791 | Mayo Clinic, Rochester MN | occupational_therapist |
| 2 | 448 | 30 N Gould St, Sheridan WY | software_company |
| 3 | 435 | UTMB, Galveston TX | nurse_practitioner |
| 4 | 413 | Mayo Clinic (second building) | cardiologist |
| 5 | 371 | Bassett Medical Center, NY | doctor |
| 6 | 364 | Penn State Health Hershey | nurse_practitioner |
| 7 | 362 | MetroHealth, Cleveland | pediatrician |
| 15 | 257 | 1910 Pacific Ave, Dallas TX | software_company |
| 18 | 247 | 7901 4th St N, St. Petersburg FL | business_management_consultant |
| 19 | 245 | Mobile home park, Pennellville NY | (none) |
| 25 | 232 | 5900 Balcones Dr, Austin TX | software_company |
The ranking is dominated by one mechanism. Mayo Clinic, UTMB, Bassett, Penn State Hershey, MetroHealth, Geisinger, Cleveland Clinic, Johns Hopkins, UPMC, Mass General, Dartmouth-Hitchcock — large hospitals list every physician, nurse practitioner and therapist as a separate business, all at the building's coordinate. That single practice produces most of the extreme density in American business data.
The four that aren't hospitals
Sorted by address rather than by exact point, the non-hospital clusters are larger than the coordinate view suggests.
| Address | Listings at this address | Top categories |
|---|---|---|
| 30 N Gould St, Sheridan WY | 866 | software_company 97 · mgmt consultant 40 · clothing_store 30 |
| 7901 4th St N, St. Petersburg FL | 733 | mgmt consultant 31 · consultant 29 · software_company 24 |
| 5900 Balcones Dr, Austin TX | 481 | software_company 35 · non-profit 13 · recruiter 13 |
| 1910 Pacific Ave, Dallas TX | 326 | software_company 13 · marketing agency 10 · mgmt consultant 9 |
The category mix is the tell, and it is the opposite of a hospital's. These are software companies, management consultants, marketing agencies, recruiters and non-profits — business types that need no premises at all. A hospital's density comes from many people at one real place; this density comes from many companies at one mailbox.
What 30 N Gould Street is, precisely
This is a well-documented address, and it is worth being exact about what our number does and does not show.
30 N Gould St hosts 26 commercial registered agents between them representing roughly 300,000 of Wyoming's 830,000 LLCs. Registered Agents Inc, one tenant, accounts for more than 40% of all new Wyoming incorporations between 2019 and 2024. The address has been the subject of ICIJ's Pandora Papers reporting, which traced pandemic relief funds to companies registered there, and of sustained local coverage in Sheridan.
What we add is narrower and, we think, more interesting. Those hundreds of thousands of LLCs exist almost entirely on paper — a filing and a mailbox. Our measurement is that 866 of them went further and built a consumer-facing Google Business listing: a name, a category, in many cases a website and a photo. A storefront for a mailbox.
The one that is neither
Rank 19 is 245 listings at a single point in Pennellville, New York, and none of them are businesses in any ordinary sense. Their names are "744 County Rte 10 Lot A038", "Lot 7", "Lot 23" — a mobile home park where every individual lot became its own Google Business listing. All carry exactly one review and no category at all.
It is a useful reminder that "many listings at one point" is a signature, not a diagnosis. Three completely unrelated things produce it: a hospital listing its staff, a registered agent listing its clients, and a landlord listing its lots.
How we did this, and the caveats
| Decision | What we did |
|---|---|
| Source | Crawlora's Google Business index, 20.0M US listings, queried directly rather than through paginated search |
| Clustering | Geohash aggregation at precision 12 (~3.7cm per cell) over every US listing carrying coordinates |
| Sizing | Coordinate clusters to find groups; street-address matching to size them |
| Identification | Building occupants identified from the listings' own addresses; the Wyoming address corroborated against external reporting |
Coordinate clustering undercounts a co-located group. Listings at one street address do not all carry an identical coordinate: 30 N Gould St spans at least five distinct exact points, the largest holding 436 of its 866 listings. So the 3.7cm view is the right tool for finding dense addresses and the wrong one for measuring them. Both numbers appear above, labelled.
This is not a geocoding artifact. The obvious failure mode would be a geocoder falling back to a ZIP or city centroid and piling unrelated businesses onto one point. These coordinates are building-precise, to 7–15 decimal places, and the listings at each carry consistent street addresses with suite numbers — not the signature of a centroid fallback.
The top 30 is not a census of the phenomenon. There are certainly other registered-agent addresses and other hospitals below the cutoff. We report what the 30 densest points are, not how many such addresses exist nationally.
Occupant identification is inference from the listing data. We read the addresses and business names in each cluster; we did not independently verify every building. The Wyoming address is the exception, and it is corroborated by outside reporting rather than by us.
Find the clusters in your own market
The Google Business dataset behind this analysis carries coordinates, category, website and phone on 132M+ listings — enough to run this clustering on any city or industry you care about. 2,000 free credits a month, no card.
Related reading
Frequently asked questions
What is the densest single point in US business data?
The Mayo Clinic in Rochester, Minnesota, with 791 listings sharing one 3.7cm geohash cell. Large hospitals list every physician, nurse practitioner and therapist as a separate business at the building's coordinate, and that single practice accounts for 26 of the 30 densest points in the country.
Why do 866 businesses share one address in Sheridan, Wyoming?
30 N Gould St hosts 26 commercial registered agents, which between them represent roughly 300,000 of Wyoming's 830,000 LLCs. Registered Agents Inc alone accounts for more than 40% of new Wyoming incorporations between 2019 and 2024. Most of those companies exist only as a filing and a mailbox; our finding is that 866 of them also built a consumer-facing Google Business listing.
Are the businesses at that Wyoming address fraudulent?
Nothing here shows that, and we do not claim it. Wyoming's low fees, simple reporting and privacy protections are legal and widely used by legitimate startups, holding companies and small online businesses, and reporting on the address consistently makes that point. Listing data cannot separate a remote-first software company that used a registered agent from a shell company. The finding is about visibility, not legitimacy.
How do you tell a hospital cluster from a virtual-office cluster?
By category, and cleanly. Hospital clusters resolve to doctor, nurse_practitioner, pediatrician and cardiologist. The registered-agent addresses resolve to software_company, management consultant, marketing agency and recruiter — business types that need no premises. A hospital's density is many people at one real place; the other is many companies at one mailbox.
Could these clusters just be a geocoding error?
That is the obvious failure mode — a geocoder falling back to a ZIP or city centroid would pile unrelated businesses onto one point. It is not what is happening. The coordinates are building-precise to 7 to 15 decimal places, and the listings at each carry consistent street addresses with suite numbers, which a centroid fallback would not produce.