What a grantee file can actually tell us
A grant decision begins with a simple question: what do we know about this organization today? We followed that question through IRS status records, filings, restriction snapshots, and the newer sources in our dataset. The useful finding is how often a blank space calls for another check, rather than a verdict.
By the CharitySense Data team
Published Updated
01/Start with the decision
Start with the decision, then choose the evidence
We often talk about a nonprofit as if it has one diligence score. A funder actually faces several different questions. Is the recipient legally eligible for this particular grant? Can its finances support the proposed work? Who else is connected to it? What is happening now that an annual return cannot show? The answer to one does not settle the others.
The vehicle matters too. An individual donor can use our companion guide on vetting a nonprofit. A sponsoring charity controls grants from a donor-advised fund, while a private foundation can face expenditure-responsibility rules for grants to certain recipients. A family office follows the rules of the vehicle it uses. Corporate giving may bring its own policies and review process. We would write the vehicle and intended recipient in the file before applying a generic checklist.
| The question to resolve | |
|---|---|
| Donor-advised fund | Which organization does the sponsor legally control the grant to, and what recipient classification or extra conditions apply? |
| Private foundation | Is the recipient a qualifying public charity or exempt operating foundation? If not, does an exception, equivalency determination, or expenditure responsibility apply? |
| Family office | Which legal vehicle is making the grant, and who owns the required checks and records? |
| Corporate program | What do the program's eligibility policy, payment controls, and reputational review require for this recipient? |
The IRS describes the sponsor's legal control of DAF assets and the public-charity classifications on which grantors may rely, subject to limits. Its DAF guidance and reliance criteria are a better legal starting point than an all-purpose claim that every grant must go to a listed public charity.
02/first screen has three
The first screen has three different kinds of answer
We matched IRS-linked organization records to the status evidence available in our current snapshot. In 3,696,792 records, 1,389,088 had a Publication 78 listing. Another 542,923 carried a deductibility code from the IRS Business Master File without a Publication 78 listing in this compiled view. For 1,764,781, the status was unknown. That last number is the largest group. It needs verification; eligibility cannot be inferred from a missing status.
What the compiled record can establish
Nearly half need a fresh eligibility answer
Share of 3,696,792 IRS-linked organization records. Hover for the exact count; select a row for its interpretation.
Publication 78 itself is not a complete roster of eligible donees. The IRS search guidance notes that some churches, group-ruling subordinates, and governmental units may be eligible without appearing there. The same IRS tool exposes automatic revocations and later reinstatements. A name match alone cannot resolve those changes; use the EIN, current classification, and relevant determination information.
State registration has its own coverage boundary. Our current source tracks California; 3,305,615 IRS-linked records are marked "not applicable" because the organization is headquartered outside that tracked jurisdiction. That status says nothing about another state's requirements. For a grant where state registration matters, the relevant state registry belongs in the file.
Missing evidence is a reason to look again. It is not a negative finding about the grantee.
03/watchlist match
A watchlist match is a research task
The expanded dataset now has a separate restriction layer drawn from OFAC and SAM source snapshots. Our September 28 review of 3,696,792 IRS-linked records found 397 records marked for restriction review and five marked confirmed. Those labels describe matches in a dated compiled source. They do not, by themselves, establish that a proposed transaction is prohibited. A potential name match has to be resolved against the official entry, identity details, applicable program rules, and the current list. The OFAC match guidance makes the need for that human check explicit.
There is another useful distinction inside the data. Its separate list-screening factor can identify a listing on other watchlists, but a listing is not the same legal claim as an OFAC or SAM restriction. In the same snapshot, only three records were marked for review on that separate screening factor. Provider coverage was incomplete in records we inspected, so a clear label only means no match among lists actually checked. We would record the source snapshot and rerun the relevant official checks at the grant decision.
04/filing tells us about
The filing tells us about capacity, with a lag
A full Form 990 gives a reviewer reported revenue, expenses, assets, liabilities, compensation, governance answers, and program descriptions. A 990-EZ or 990-PF answers a different set of questions. A 990-N is a short electronic notice, so we cannot read a balance sheet from it. The current corpus contains 8,868,546 990-series filings, including 1,540,498 990-N notices. Those are filing counts across years, not counts of organizations without financial information today.
We would compare the latest available years, then ask whether the proposed grant fits the organization's reported scale. A thin reserve or operating loss can prompt a budget conversation. So can a sudden revenue change. None of those alone proves weak stewardship or poor program outcomes. The filing is historical, and recent tax years are still arriving. The overhead guide explains why one expense ratio is too narrow a test; the shape of nonprofit money shows how uneven the financial comparison group can be.
The public filing is still valuable for reconciliation. We can inspect whether amounts in schedules agree with the base return and whether a reported grant or related-party payment has a named recipient. A mismatch is a question for the filing and its notes, not a finding of wrongdoing. Sometimes it reflects an amended return or a different reporting boundary.
05/Relationships add context
Relationships add context that a ratio cannot
Our current filing analysis found 455,980 organizations with named counterparty records. They connect a filer to counterparties named in its most recent return, where the source supplies enough identity evidence for the link. Related organizations, grant recipients, and other reported money flows can then be reviewed with their filing citations. That matters when the proposed grantee operates through affiliates: the organization on the application may be only one member of the network. A blank graph is not evidence of independence, because the return may not name a counterparty or the identity may not resolve.
Available federal-award records add another lens for 26,902 EINs in this source snapshot. Award history can tell a reviewer which federal relationships are visible and which agency records deserve a closer read. Absence from those records is not proof of no government funding. These two layers are useful for choosing follow-up questions, while the actual grant file still needs the recipient's own explanations and current documents.
If you are tracing a specific grant path, our article on where the money goes is a more detailed companion. The account of Schedules B and R also explains why named relationships and anonymous contributors must not be conflated.
06/What a website adds
What a website adds to the file
We had captured public-website evidence for 33,506 organizations when we checked on September 28. That is a deliberately partial enrichment layer beside the larger filing corpus. A website may describe a current program, a monitoring practice, or an outcome claim that no annual return could contain. We would retain the capture date and source page, then compare that public claim with the organization's latest report and, for a consequential grant, ask for direct evidence of the result. A missing web record tells us nothing about whether the organization publishes useful information elsewhere.
The limit cuts both ways. A filing can establish what the organization reported for a tax year, but it does not measure whether a program worked. A website can tell us what an organization says it does now, but that claim may change or lack an independent evaluation. We use the two as different kinds of evidence, with dates attached to both.
07/What still belongs
What still belongs in the grant file
For a private foundation, expenditure responsibility can require a pre-grant inquiry, written terms, monitoring, and reporting when a grant goes to a recipient covered by the rule. A foreign grantee can also be assessed through an equivalency determination under the IRS conditions. The IRS expenditure-responsibility explanation and foreign-grant guidance set out the distinctions. The applicable route depends on the grant and recipient; neither a blog post nor a database label can choose it for a funder.
We would keep a dated record of the exact EIN and official status check, the latest relevant return, the questions raised by finances or relationships, any resolved list matches, and the grantee's answers. For a material grant, the budget, governance documents, audited financial statements where available, and evidence of program results sit beside the public data. The review is strongest when someone can tell which claim came from which source and when it was observed.
A searchable starting list helps compare mission, geography, financial scale, and available status evidence. We would still carry each finalist into a separate, documented review. A shortlist helps us find questions; it is not an approval queue.
Behind the Data
We reviewed the current CharitySense filing corpus on September 28, 2026. Its latest coverage summary, calculated September 26, includes 2,222,250 public profiles and 8,868,546 filings. Of those filings, 1,540,498 are short Form 990-N notices. For eligibility questions, we separately examined 3,696,792 organization records linked to IRS status evidence. The deductibility, restriction, screening, and state-registration figures each use those 3,696,792 records as their denominator. We also counted 455,980 organizations with named counterparties in their latest filings, 26,902 with federal-award records in the available source, and 33,506 with captured public-website evidence. These are different populations and none is a complete census of grant-eligible organizations. Hero counts in millions are rounded to two decimals; exact counts appear here. Source snapshots predate our analysis. Recent filings and external lists can change, and this article is not legal or tax advice.
Sources
- CharitySense analysis of IRS Form 990-series filings, IRS organization-status records, named filing relationships, available federal-award records, and captured public websites; reviewed September 28, 2026.
- IRS Tax Exempt Organization Search and IRS grantor reliance criteria, for current status, exceptions, and classification checks.
- IRS donor-advised fund guidance, expenditure responsibility, and foreign-grantee guidance, for the legal distinctions described above.
- OFAC charitable-sector guidance and potential-match guidance, for dated list checks and human resolution of possible matches.
- IRS Form 990-N guidance, for the limited information in an e-Postcard.
Ask the research team
What question about a potential grantee would you want us to investigate next? Tell us what evidence would make the answer useful to you.
08/Final Thoughts
Final Thoughts
The expanded data changes how we think about grantee diligence. A larger corpus helps us find a candidate and locate the evidence around it. The most informative result can be an unresolved status: an IRS classification to verify, a dated sanctions snapshot to refresh, or a promising program claim that needs a current report. We learn more when we treat those gaps as work to do than when we turn them into a score.
For the broader view of who appears in the corpus at all, see the filing blind spot. The organization-level review begins by asking what the proposed grant needs to establish, then collecting the freshest evidence capable of answering it.
About the author
The CharitySense Data team
Research team
We read public nonprofit filings and other sources to understand how the sector works. This blog shares what we find, along with the limits of the evidence. Each article explains how its figures were measured and cites the records or primary sources behind them.