This is a guide to the free US public registers that list every school district, college, local government, hospital and nonprofit.
We’ll go through one example: a company that sells a work-order app to the facilities teams of schools, colleges, local governments and hospitals in Ohio.
Every count in this guide is real. I counted them on the official files.
The federal school district file has 19,630 rows, and every one has a phone number. The college file names the person who runs 5,935 colleges and trade schools. In Ohio alone, five of these files give you 4,392 institutions, each with an official ID and an address.
This guide is for founders and list builders who sell to schools, colleges, local governments, hospitals or nonprofits. My TAM playbook uses a register as the last check that a list is complete. In these markets, the register is the list itself. Read step 1, then go to the registers for your market.
The running example: say you work at Gantrybook, a company with a work-order app for facilities teams. Gantrybook wants every institution in Ohio that runs its own buildings.
1. Check whether a register lists your whole market
Some buyers exist because a law creates them, like school districts, counties and fire districts. Others need a license or a tax status, like hospitals and nonprofits. For all of them, a government agency keeps the complete list.
Contact databases are built mostly from company websites and professional profiles. That suits companies. It suits a township with a part-time clerk or a volunteer fire district much less, because they have little of either.
Three questions tell you whether a register is worth using:
- Does it list every member of the market, or only the ones that file something?
- Which contact fields does it have: phone, website, a person’s name, or only a role?
- Which ID connects it to the next file: a tax ID, a website domain, or the register’s own ID?
2. Schools: start with the NCES district and school files
What’s in it. NCES is the federal statistics office for education. Its Common Core of Data has a directory file of every public school district and one of every public school. The 2024-25 district file has 19,630 rows, all with a phone number and 88% with a website. The school file has 102,178 rows, all with a phone number and 69% with a website. Neither names a person.
Where. On the NCES website, under Common Core of Data: pick the nonfiscal directory file for the year you want.
Traps:
- Not every row is a district you’d call a district. The file holds 13,536 regular districts, 4,393 independent charter districts and 691 service agencies, among other types. Filter on the district type.
- Closed and future rows are in the file. Filter on the status and drop closed, inactive and future rows.
- Clean the website field. Some entries are “N/A” or “none” written as a web address, and 45 districts share sites.google.com. Never match districts on that domain.
ID: every district has an NCES district ID, and every school row carries the ID of its district.
In Ohio: 1,073 district rows. 619 of them are regular districts, and 613 of those have a website. Plus 3,611 schools.
3. Colleges: use IPEDS for the person in charge
What’s in it. IPEDS is the federal database of colleges. Every institution in a federal student aid program has to report to it, so it covers colleges, universities and trade schools. The 2025 directory file has 5,985 institutions. 5,935 of them come with a website, a phone number, and the name and title of the chief administrator. “President” is the most common title (3,024 rows), followed by chancellor and CEO. The tax ID (EIN) is filled for 99%.
Where. The IPEDS data center on the NCES website, under “Directory information”.
Traps:
- The file starts with an invisible character. If you read the CSV as plain UTF-8, the name of the first column starts with that character and no longer equals “UNITID”. Every lookup on the ID then fails without an error. Read it with
encoding="utf-8-sig". - Inactive rows are in the file. Keep the rows marked as active this year (5,916).
- One tax ID, many campuses. 395 EINs are shared by 1,690 institutions, mostly campuses of one system. Use the EIN to connect files, but don’t remove duplicates by it.
- It’s wider than colleges. 2,203 institutions are for-profit, and 617 of those have beauty, cosmetology or barber in their name. Filter on the control and sector fields.
ID: the IPEDS institution ID, plus the EIN to connect it to other files.
In Ohio: 270 institutions (106 public, 76 private nonprofit, 88 for-profit), and 268 of them have a named chief administrator.
4. Local governments: use the Census list of government units
What’s in it. One Excel file with every county, city, township, special district and independent school district. The Census Bureau pulled it from its master list of governments on August 13, 2026. It has 38,693 general-purpose governments (3,031 counties, 19,489 municipalities, 16,173 townships), 40,511 special districts and 12,525 school districts.
Each row has a mailing address, a website for some, and the title of the contact role (mayor, clerk, superintendent), never a name. Special districts are listed with their function, and each function is a market of its own: 5,974 fire protection districts, 3,553 water supply districts, 1,825 sewerage districts and 1,673 library districts.
Where. The Census Bureau’s Government Organization datasets page.
Traps:
- The websites are self-reported. The file’s own documentation says only limited quality control was done on them. They’re listed for 42% of general-purpose governments, 30% of special districts and 94% of school districts.
- Names are abbreviated, like “BLUFFTON EX VLG SCH DIST” for Bluffton Exempted Village Schools. That breaks name matching against NCES unless you write out the abbreviations.
- Some units are dormant. They have no financial activity but still count as governments. The active field marks them.
ID: the Census ID, plus the state, county and place codes.
In Ohio: 88 counties, 923 municipalities, 1,308 townships, 991 special districts and 665 school districts.
5. .gov domains: add the CISA list
What’s in it. CISA, the federal cybersecurity agency, runs the .gov domain registry and publishes the full list of registered .gov domains. Each row has the domain, the organization, its type, city and state. The list had 16,915 domains on the day I counted: 9,206 city, 2,715 county, 1,438 state and 1,111 special district domains, plus federal, tribal and election domains.
Where. CISA’s public dotgov-data repository on GitHub. The files are updated daily when anything changes.
Traps:
- One government, many domains. Police departments, election offices and county agencies register their own.
- A domain isn’t always a website. CISA says not every registered domain offers an online service.
- Most local governments don’t use .gov. Of the 16,202 general-purpose governments with a website in the Census file, only 1,653 list a .gov address.
- The security contact column is for reporting security problems. Keep it out of any outreach.
Why it’s worth adding anyway: matching exact name and state, I connected 4,598 municipalities in the Census file to a .gov domain. For 1,048 of them the Census file had no website at all. For 2,888 it listed a different address, like a .com, .org or .us domain.
In Ohio: 705 domains (446 city, 135 county, 37 special district).
6. Hospitals and care: use the CMS provider files
What’s in it. CMS, the federal agency that runs Medicare, publishes a file of every hospital registered with Medicare. It lists 5,419 hospitals, every one with a phone number, plus the hospital type, the ownership and whether it has an emergency department. There’s no website and no name. Types include 3,107 acute care hospitals, 1,382 critical access hospitals (small rural hospitals), 624 psychiatric and 94 children’s hospitals.
The same site has files built the same way for 14,690 nursing homes, 12,460 home health agencies and 7,490 dialysis facilities.
Where. The CMS Provider Data Catalog, “Hospital General Information”.
Traps:
- The download link carries a version number. Read the current link from the dataset’s description on the site instead of saving it in your script.
- Ownership decides who buys. “Government, hospital district or authority” and “Voluntary non-profit, private” are different sales processes.
- The facility ID is a CMS number. It connects CMS files and nothing else. Connect to other files by name, city and phone.
In Ohio: 193 hospitals (118 acute care, 33 critical access, 29 psychiatric, 8 children’s).
7. Nonprofits: use the IRS master file and Form 990
What’s in it. Two layers. The first is the IRS Exempt Organizations Business Master File, a list of every tax-exempt organization: 1,964,958 records, as CSV files by state or region. Each record has the EIN, name, address, an “in care of” name, an activity code (filled for 71%) and asset and revenue amounts. It has no phone number and no website.
The second layer is Form 990, the yearly tax return of larger nonprofits, as XML files. The 2025 index lists 748,906 returns, and 376,920 of them are full Form 990s. I read one batch of 9,264 full 990s. Every return had a phone number and a signing officer, 99% named their officers or directors, and 68% listed a website.
Where. The IRS website: “Exempt Organizations Business Master File Extract” for the first layer, “Form 990 series downloads” for the second.
Traps:
- Some batches use Deflate64 compression. The two largest I checked, 2025_TEOS_XML_05B and 2026_TEOS_XML_05A, use it, and Python’s built-in zip library can’t open them. 7-Zip can.
- The batches are large. The ones I checked ran from 104 MB to 521 MB. Filter the yearly index on your EINs first, then download only the batches you need.
- “In care of” can be a person, a firm or an office. Treat it as a hint, never as the decision maker.
ID: the EIN.
In Ohio: 76,947 tax-exempt organizations in the master file.
8. Local businesses: use Overture where no register exists
What’s in it. Overture Maps is a free map dataset from a group of companies including Meta and Microsoft. Its September 2026 release has 81 million places worldwide, 59 million of them from Meta. Fields include name, category, a confidence score, websites, phone numbers, addresses and whether a place is open or closed.
I pulled Ohio: 585,782 places with an Ohio address. 80% have a website, 88% have a phone number, and 16,225 are marked permanently closed.
Where. Free files on Amazon S3 and Azure, which you query by map area. DuckDB, a free database tool, reads them directly. No account needed.
Traps:
- The old category column is gone since the September 2026 release. Use the new basic category and taxonomy fields.
- Overture’s own docs warn about duplicates, a high junk rate and missing fields. Filter on the confidence score.
- Each release stays online for 60 days at most. Save the release you build from, or you can’t repeat the build.
- Categories are broad. Overture tags 2,155 Ohio places as hospitals, while CMS lists 193 hospitals in Ohio. Let the register decide which institutions exist, and use Overture to fill in phone numbers and websites.
ID: Overture’s place ID, which is meant to stay the same across releases.
9. Join the registers in a fixed order
Joining means connecting two files through a value both have, like a tax ID. Every register has its own ID. Keep all of them on every row, then connect the files in this order:
- EIN, where both files have one. 1,715 of the 1,785 private nonprofit colleges in IPEDS with a valid EIN are in the IRS master file. That gives you their 990s with the officers’ names.
- Website domain, after you remove the “http” part, “www.” and everything after the first slash. Never match on a domain that several rows share.
- Cleaned name plus city and state. Write out abbreviations, remove words like “school district” and “city of”, and give ties to a person to decide.
- Phone number, for Overture and CMS rows.
Paste this into Claude Code to build the joined file:
Build tools/registers.py for Ohio. Download into raw/ the current
NCES district directory file, the IPEDS directory file, the Census
government units file, the CISA .gov list and the CMS hospital file.
Find the current links on each agency's site. Read the IPEDS file
with encoding utf-8-sig.
Keep only Ohio rows, and only open or active rows.
Keep every register's own ID in its own column.
Join in this order: EIN, then website domain (lowercase, without
http, www. and everything after the first slash, never a domain that
several rows share), then cleaned name plus city and state.
Write institutions.csv: one row per institution, all IDs, the source
of every field and the match method. Print rows in and out per
register, and list every name match that fits more than one row.
Name alone is never enough. I matched Ohio’s 619 regular school districts against the Census file. The website domain connected 300 of them. Adding cleaned name plus city brought it to 559. The name by itself fit more than one district for 67 of them, because without words like “local” or “city”, different districts share the same name.
Give every row a link someone can open: the district’s page in the NCES district search, the college’s page on College Navigator, or the nonprofit’s page on ProPublica’s Nonprofit Explorer.
10. Pick the register for your market
| Market | Start with | Add | What you get |
|---|---|---|---|
| Public schools (K-12) | NCES districts and schools | Census school districts, CISA | Phone, website |
| Higher education | IPEDS | IRS master file and 990 via EIN | Chief administrator, officers, phone, website |
| Cities, counties, townships | Census general purpose | CISA .gov, Overture | Role title, current domain, phone |
| Utilities, fire, library, park districts | Census special districts | CISA, Overture | Role title, some websites |
| Hospitals and care facilities | CMS provider files | IRS for nonprofits, Overture | Phone, ownership, website |
| Nonprofits and foundations | IRS master file | Form 990 XML | Officers, phone, website |
| Local businesses in any category | Overture | Census business counts as a check | Phone, website |
For Gantrybook in Ohio, that’s 619 regular school districts, 270 colleges, 2,319 counties, cities and townships, 991 special districts and 193 hospitals. Together that’s 4,392 institutions before any filter, all from free files, each with a register ID.
Say Gantrybook keeps districts, colleges, cities, counties and water or sewer districts. The next step is finding the facilities director at each one. No register lists that person, so this is where your usual contact tools come in.
Pull every person with a facilities title at each institution, like director of facilities, facilities manager or director of operations, not one person per institution.
Before you buy emails, remove your current customers, open deals and everyone who asked not to be contacted. Keep only the emails an email verifier marks valid, and drop every “unknown”. Catch-all addresses, on mail servers that accept any address, go out only after a tool like LeadMagic or BounceBan confirms them.
Where to start today
You can get your market from a register today in three steps:
- Find your market in the table and download its register.
- Filter it to your states and to open rows.
- Join the other registers in the fixed order from step 9. That’s your list of every organization in your market.


