Methodology

What this data is, how it is built, how accurate it is, and what it cannot tell you. We would rather lose your trust here than in a grant application.

Where the data comes from

Every US foundation and tax-exempt organization files an annual return with the IRS, and the IRS publishes them in bulk. Private foundations file Form 990-PF, which lists every grant paid during the year. Public charities — community foundations, United Ways, larger operating nonprofits — file Form 990, and report grants on Schedule I.

We process both, in full, for the four most recent submission years available: 1.9 million returns, 11.4 million individual grants, and $1.2 trillion in giving, covering tax years 2019 through 2024. Not a sample — every return the IRS published in those years.

How grants get matched to organizations

Form 990-PF identifies grant recipients by name and address only — no tax ID. Connecting those to real organizations is the hard part, and it is where most free grant data falls down.

Schedule I filers, unlike 990-PF filers, do report the recipient's tax ID. That gives us hundreds of thousands of verified name-to-ID pairs contributed by the filers themselves, which we use both to resolve recipients directly and to test our matching against known answers.

Measured against those known answers, with the answer key hidden from the matcher: 96–98% of matches are correct. We currently resolve about 66% of grant records to a specific organization, and 43% of the distinct organizations named. Those two numbers differ because matching succeeds most often on frequently-funded organizations, and we would rather say so than quote whichever is flattering. We would also rather leave a record unmatched than attach it to the wrong nonprofit.

Foundation pages show every recipient a foundation reported, including the ones we could not match to a tax ID. Those organizations are real; we simply could not confirm which registered nonprofit they are, so they appear exactly as the foundation wrote them.

The reporting lag — the most important caveat

This is historical data, not a feed of open opportunities. Foundation filings are published roughly a year after the tax year ends. Nothing here is a live application deadline.

It is still useful, because foundation giving is unusually sticky. Measuring across consecutive years, a typical foundation funds about two thirds of the same organizations again, and only about 3% turn over their entire grantee list. Smaller foundations are the most consistent of all. So past giving is a strong predictor of future giving — but it is a prediction, and we will not pretend otherwise.

What we mean by "accepts unsolicited requests"

Form 990-PF asks foundations to indicate whether they make grants only to preselected organizations. Of the 91,758 private foundations in our data that actually made a grant and answered the question, 78% said yes. When we mark a foundation as accepting requests, it means it did not check that box on its most recent return.

This is the foundation's own statement, not our judgement, and it can be out of date or filled in carelessly. Treat it as a strong signal for prioritising your time, not a guarantee.

Known limitations

• Grants to named individuals (scholarships, hardship grants) are excluded from organization matching — they are a different thing.
• Cause categories come from the recipient's IRS classification. An arts grant to a university may be classified as education.
• Organizations that lost their tax-exempt status disappear from the IRS master file, so some historical recipients cannot be matched at all.
• Foundations that give only through donor-advised funds may report far less detail than those that grant directly.

Corrections

If something here is wrong about your organization or your foundation, tell us and we will fix it: [email protected].