Methodology

The registers behind each list

Each list begins with a public source that defines which organizations belong in it. We keep the source's identifier on every row (for example the NPI, the SEC file number or the USDOT number) so you can check the row against the source yourself.

Each register has a free CSV of our counts by state, linked below, licensed under CC BY 4.0.

CMS NPPES

The National Plan and Provider Enumeration System (NPPES) is the federal register of healthcare providers, published by CMS as a monthly full file with weekly updates. It feeds the dentist, orthodontist, physician, physician specialty, pharmacy, physical therapist, optometrist and chiropractor lists. We use the file of 2026-09-14.

A register count is the number of distinct registered practice addresses (address line 1 plus five-digit ZIP, with suite numbers kept) of active NPIs that carry the list's taxonomy code. Individual and organization records both count, except for pharmacies, where only organizations count and the unit is a registered pharmacy address. The count includes stale, duplicate and billing addresses, so it is an upper bound on practices.

Each physician specialty list (dermatologists, cardiologists, oncologists, pediatricians, neurologists, urologists and radiologists) counts the registered practice addresses where at least one active NPI carries a code from that specialty's set in any of its 15 taxonomy slots. The sets come from the NUCC Health Care Provider Taxonomy code set and are listed on each specialty page. Radiation oncology counts with oncologists, not radiologists, and critical care is not counted as cardiology. Pediatric cardiology and pediatric hematology-oncology count in two lists each, so specialty counts can't be added together. The specialty is the code the provider chose, not a board certification.

NPPES also lists individual clinicians. We use them to count clinicians per address and to name a practice that has no organization record. Our lists will hold no column of clinicians' names and no home addresses.

Providers rarely update or deactivate an NPI when they retire or move. In the file of 2026-09-14, the median dental record was last changed 10.0 years earlier. So we will confirm each practice through its website or other current evidence before it becomes a row, and leave out rows we cannot confirm.

Data files: dentists, physicians, pharmacies, physical therapists, optometrists, chiropractors.

SEC Form ADV

The SEC publishes the Form ADV data of every investment adviser registered with it, and it feeds the RIA firm list. We use the file of 2026-10-02.

A register count is an SEC-registered adviser firm with status "Approved" or "120-Day Approval" and a main office in the 50 states or DC, counted once at its main office. Smaller advisers register with their state securities regulator instead of the SEC, so they are not in this file, and firms that withhold their main office address have no state and are left out. Form ADV is self-reported and updated at least once a year, so assets and employee counts can be up to a year old.

Our ranking of the largest RIA firms orders these firms by regulatory assets under management (Form ADV Item 5F(2)(c), self-reported, including non-discretionary assets).

Data file: RIA firms.

Private equity, hedge fund and venture capital advisers

The same SEC release, file of 2026-10-02, feeds the private equity, hedge fund and venture capital lists. It has two files: SEC-registered advisers and exempt reporting advisers. Exempt reporting advisers file parts of Form ADV without registering with the SEC, and many venture capital and smaller private fund advisers file this way. An adviser counts when it answers yes to that fund type in Form ADV Item 7.B and has a main office in the 50 states or DC. An adviser in both files is counted once, from its registered filing. Advisers that report several fund types appear in each of those lists, so the three counts can't be added together. Form ADV fund types follow the SEC's definitions, so a private equity count can include private credit, infrastructure and secondaries managers. Advisers whose main office the SEC withholds as a private residence have no state, so they are left out of the counts and the rankings.

Suspect filings. We remove two kinds of exempt reporting filing from the counts and the rankings. A filing is removed when its gross private-fund assets are an odd figure (not a whole number of millions) that three or more different advisers report exactly, or when its business name contains the word exchange, token, crypto or RWA. In the file of 2026-10-02 the first rule caught a cluster of new filings, most of them giving a main office in Colorado, that all reported one identical gross figure. The filter removed 172 filings from the private equity count, 15 from the hedge fund count and 11 from the venture capital count. It may also remove a few genuine small advisers, and it does not catch every doubtful filing.

Rankings. The private-fund rankings list "focused" advisers, whose funds of that type are at least half of the private funds they report. They are ordered by the gross assets of the private funds the adviser reports, not fund size or assets raised. Each row is one SEC filer, so a group that files under several advisers can appear more than once. We publish no totals of these assets, because feeder funds, funds of funds and sub-advised funds are reported more than once.

Data files: private equity firms, hedge funds, venture capital firms.

CMS Hospital General Information

CMS publishes Hospital General Information in its Provider Data Catalog (dataset xubh-q36u). CMS describes it as "a list of all hospitals that have been registered with Medicare". It feeds the hospital list. We use the file of 2026-07-22.

A register count is a hospital in the file with a state in the 50 states or DC, one row per CMS Certification Number (CCN). Hospital type, ownership and emergency services come from the file's own fields. Hospitals not enrolled in Medicare are missing, and one CCN can cover several campuses, so the count is of certified hospitals, not buildings. The file has no rehabilitation hospital type and few long-term hospitals, so it counts fewer hospitals than the American Hospital Association. The hospital list will hold hospital organizations and their published main and department inboxes, never named executives or staff.

Data file: hospitals.

FMCSA census and operating authority data

The Federal Motor Carrier Safety Administration (FMCSA) publishes a census of every entity with a USDOT number, and it feeds the trucking company list. We use the rows of 2026-10-03.

A register count is an active FMCSA-registered carrier with an MCS-150 update in the last 24 months: the USDOT number is active, the entity is registered as a carrier, its physical address is in the 50 states or DC, and it filed its biennial MCS-150 update within two years. The count includes passenger carriers, intrastate carriers and private fleets that haul their own goods. Fleet sizes are self-reported on the MCS-150 and can be up to two years old.

The hazmat carrier list takes the carriers in that count whose census record carries the hazardous-materials indicator. The flag doesn't show that a carrier hauls hazardous materials now, and it is a different register from PHMSA hazmat registration.

The freight broker list counts active property-broker operating authorities by docket (MC number). FMCSA moved registration to its Motus system, and its legacy authority file stopped updating on 2026-05-14. Where Motus holds a broker record for a docket, Motus decides its status. Otherwise the legacy status as of 2026-05-14 stands, so an authority revoked since then without a Motus record is still counted. An authority is a license to broker freight, not proof that a company is brokering loads now. Carrier and broker counts use different units and are never added together.

Some census records carry a contact email given to FMCSA on the MCS-150. Where we use one, the row will name the register and its file date as the source, not a web page.

Data file: trucking companies. It holds the carrier counts by state; the hazmat and freight broker counts by state are on their own pages.

Overture Maps places and BLS counts

Overture Maps places is an open map dataset that combines Meta, Microsoft and other sources. It feeds the church and veterinarian lists, from the release of 2026-09-23.

A church count is a church location: an open place whose own category is a Christian place of worship, or a generic place of worship whose name reads as a church, deduplicated on name and ZIP code. A veterinary count is a veterinary clinic location listed on public maps, chosen the same way from veterinary categories and names. Both are counts of map listings, not of congregations or practices. One congregation can have several listings, several can share a building, and a place with no listing is missing.

For veterinary clinics, the official figure is the Bureau of Labor Statistics (BLS) Quarterly Census of Employment and Wages (QCEW): 36,619 private establishments in NAICS 541940 Veterinary services in 2025 Q4. QCEW counts only establishments with employees on payroll, so the two counts differ. Our statistics pages also cite the BLS Occupational Employment and Wage Statistics (OEWS), which counts wage and salary jobs and leaves out the self-employed, so it measures people in jobs, not offices.

Church pages cite single figures from the 2020 U.S. Religion Census (Grammich et al., Association of Statisticians of American Religious Bodies, 2022), which counts congregations of all faiths. Those figures are copyright ASARB, and our CSV holds only our map counts.

Contains data from Overture Maps Foundation, licensed under CDLA Permissive 2.0. Some Overture places come from Foursquare under the Apache License 2.0. Our pages show only counts. A list file will either carry Foursquare's notice with those rows or leave them out.

Data files: churches, veterinarians.

Register counts versus list rows

A register count is the size of the source before our checks, so it is an upper bound on a list. Each list will hold fewer rows, because we will remove closed, duplicate and unconfirmed entries, and institutions that are not practices. The counts on our pages are rounded; the CSVs keep exact values. Compare two counts only where their units match: a registered practice address, an adviser firm, a carrier, a hospital and a map listing measure different things.

Websites

We match each organization to its website using the address, phone and name, and we keep only matches where these agree. An organization without a website can still be in a list, with its address and phone but without an email.

Published contact details

We read the organization's own public pages (home, contact, about and a small number of other internal pages) and record:

• emails that appear on the page, with the page URL and the date we read it
• phone numbers, which we normalize; where a list offers a line type, each number is labeled landline, mobile or VoIP

We respect no-sharing notices. If a site states that it does not give out, sell or share its email addresses, we collect no email from that site. This is the notice CAN-SPAM refers to in its address-harvesting rule.

We don't guess emails. We never build addresses from name patterns (such as firstname@domain) or test random combinations. If a business has not published an email, its row has none.

Email verification

We check each email with the receiving mail server without sending a message. Each email gets one status:

The catch-all guide explains the second and third statuses and how to send to them safely.

Re-checks. We update and verify your list when you order, so you receive a fresh list. The last-verified date on each row shows the date of the latest check of that email.

At order time

We update and verify your list when you order, so you receive a fresh list. We update the records from the latest public register files, verify the emails in your selection, apply the suppression list (removal and opt-out requests) and check the file against the data dictionary before we send it. See how it works.

Email coverage

Many businesses publish no email address, so a list holds fewer emails than rows. In our test of 500 dental offices in California, Ohio and New Hampshire (29 September 2026), 21% had an email published on their own website and 11% had one our verifier marked valid. Rates differ by list and by state, and the README in every file gives that file's own counts.

What the data cannot tell you

• Ownership. NPPES names an authorized official for each organization, which does not prove who owns the practice. Group affiliation, where a list has it, is a separate, labeled field built from public evidence. It can be incomplete.
• Individual roles. A role address such as office@ reaches the office, not necessarily a particular person.
• Changes after the last check. An office can close or change its email the day after we verify it. The last-verified date tells you how recent our information is.
• Everything outside the register. An organization that is not in the source register (for example a practice that never registered in NPPES, a state-registered adviser or a hospital not enrolled in Medicare) can't appear in the list.

Removals

Anyone can ask us to remove a business or a person. Once processed, the request applies to every future file. See data removal.