A foundation can keep an open application portal for a decade and still
fund the same thirty organizations every year. That difference is visible
in its own tax filings, and almost nowhere else.
What this shows. For each foundation: how many of the
organizations it funded last year were ones it had not funded before,
how that compares to similar foundations, what it typically grants, and
how much confidence the underlying name-matching supports. It reports
what the filings say. It does not tell you where to apply — that
judgment depends on things no tax return contains.
Preview — a first cohort. These are real foundations and real
figures, computed from their own Form 990-PF filings for tax years 2020
through 2024. The cohort is a slice, not the sector: the largest filers
holding $10M to $500M in assets with at least four consecutive years on
file. If you spot something wrong in your own organization's row, that is
exactly what this stage is for — tell me and
I will show you the filings behind it.
Screen a set of funders
Filter the list, open any row for the evidence behind the figure, and
add rows to a screen you can export. Nothing is stored on a server and
there is no account.
Loading…
Your screen
Add foundations from the list above to build a comparable table you can export.
How the figure is computed
Every US private foundation files Form 990-PF, and Part XV of that
return itemises every grant it paid, with the recipient's name. Line
those lists up across consecutive filing years and you can see which
organizations were funded for the first time.
That does work the application checkbox cannot. Whether a foundation
says it accepts unsolicited proposals is self-reported and never
audited. The grant ledger is behaviour.
Where it is uncertain, and why that is shown
Form 990-PF records a recipient's name and address but no EIN, so
deciding whether this year's grantee is last year's grantee is a
name-matching problem, not a lookup. Match rate is published on every
row for that reason. Below 85% no comparison is shown at all.
One limitation worth stating plainly. Match rate does not detect
a renamed grantee. When a preparer changes and the same organization is
written differently, the record does not become ambiguous — it becomes
a second entity, which reads as a new grantee and pushes the figure up.
A high share of one-year-only recipients is flagged on the row where it
appears, but distinguishing genuine turnover from a renaming still
requires reading the filings.
What it does not tell you
Whether your work fits what the foundation funds — no tax return contains that.
Whether someone on your board already knows a trustee, which is how a great deal of foundation money actually moves.
What has changed since the last filing. These returns run twelve to twenty-four months behind.
Anything about foundations that state they do not accept unsolicited requests, beyond that they say so.
Status
Pipeline built to spec — clustering, drift guards, publication gate, 39 correctness checks passing
IRS extraction live — 594,053 Form 990-PF filings indexed across 2020–2024, covering 143,131 foundations
Now: widening the cohort beyond the first 118, and hand-checking a sample against the filings
Then: per-foundation pages, and an index of how often foundations fund anyone new
Then: opt-in board-history check, using only officer listings from public filings