Building a Funder Prospect List That Fits the Mission, Not a Directory Dump

A new grant-writing engagement almost always starts the same way: a client hands you a mission statement, a program budget, and a request to "find some foundations that might fund this." It's tempting to open an AI assistant and type something like "give me a list of foundations that fund youth literacy programs in Ohio" and treat whatever comes back as a starting point you can refine. Don't. Unless your assistant has live web browsing turned on for that specific query — and you verify the results anyway — a general-purpose AI model isn't looking anything up. It's generating a plausible-sounding list based on patterns in its training data, and a plausible-sounding foundation name, contact, or award range is not the same thing as a real one.

This isn't a hypothetical risk. It's the exact limitation our own funder-research prompts are built around and disclose up front: "Do not ask a general-purpose AI assistant to 'find' or 'list' real foundations from memory alone — without live web access, it can generate plausible-sounding funder names, contact details, or figures that are wrong." That's not a footnote we're hiding from you to sell a toolkit — it's the starting assumption behind how the whole funder-research process has to work, whether you use AI at all. (For a broader look at why AI drafts need fact-checking in the first place, see our article on what the hallucination research actually says.)

So what does work? A funder prospect list that actually holds up is built from primary sources first, and AI — if you use it — comes in afterward, to help you process and compare what you've already verified. Here's the order that actually protects you and your client.

Start with what your client already has

Before searching anywhere else, ask the client for their existing donor and funder history: past grants received, board members' professional and philanthropic connections, and any funders who've expressed interest informally. This step gets skipped constantly because it isn't as exciting as "researching new funders," but a funder who has already given to this organization, or whose program officer already knows a board member, is a warmer and more verifiable lead than anything a directory or search engine will surface. It also costs nothing and takes an afternoon.

Search a real directory or the open web — but for leads, not answers

Once you've exhausted existing relationships, search subscription directories your client or agency may already have access to (Candid's Foundation Directory Online and tools like Instrumentl are the ones most commonly used in the sector) or run structured searches on the open web. Either way, the output of this step is a list of candidate names to verify, not a finished prospect list. This is also the one place an AI assistant can help without the "finding funders from memory" problem: instead of asking it to name foundations, ask it to help you build better search-query strings and category terms for the databases and search engines you're already using — varying by population served, geography, and funder type. That's a genuinely different (and safe) use of AI than asking it to name funders directly, because you're the one running the search and seeing the real results.

Verify every candidate against primary sources before it goes on the list

This is the step that actually separates a usable prospect list from a directory dump. For every candidate name, before it earns a spot on the tracker:

• Read the funder's own current website or guidelines. Priorities, exclusions, and deadlines change year to year — a summary from a directory, a blog post, or an AI assistant's memory can be a cycle or two out of date. The funder's own current page is the only source that reflects this year's actual rules.
• Confirm the organization is real and currently tax-exempt. The IRS's own Tax Exempt Organization Search lets you check an organization's exemption status and eligibility to receive tax-deductible contributions (Pub. 78 data) directly from the source, free.
• Pull the funder's actual tax filings. ProPublica's Nonprofit Explorer is a free, searchable database of Form 990 filings by name or EIN — it shows revenue, expenses, executive compensation, and (for private foundations filing Form 990-PF) the actual list of grants the foundation paid out that year. This is possible because exempt organizations and the IRS are both required to make these returns available for public inspection — a Form 990 is not a leak or a workaround, it's a public record by design.
• Read the actual grants list, not just the summary financials. A private foundation's Form 990-PF includes a supplementary-information section listing grants paid or approved during the year, and many foundations describe their own application process — where to send inquiries, what format, any submission deadlines — right there in the return, which matters most for the smaller family foundations that don't maintain a full public website.

Treat anything you can't confirm through one of these primary sources as unverified, and either keep digging or drop the candidate — don't carry an unconfirmed name forward into a client-facing list.

Use the funder's real past-grantee list to check fit and realism

Once you have a funder's actual grants list — from their Form 990-PF, an annual report, or a "past grantees" page — it's worth a closer look before committing hours to a full application: what's the rough range of award sizes, are there recurring themes in program type or geography, and does this funder mostly fund new organizations or renew existing relationships? Reading that pattern by hand works fine; if the list is long, an AI assistant can also help spot the pattern faster, as long as you're pasting in the funder's actual published list rather than asking it to guess. Either way, treat the pattern as a planning input, not a guarantee — award ranges shift year to year, and the funder's current guidelines are still the final word on what they'll actually consider.

Log what you verified, and when

A prospect list is only as useful as its documentation. For every confirmed prospect, record the funder's name and official website, a program officer or contact if you have one, focus area and any geographic restriction, the typical award range you actually found (with a note on where it came from), whether a matching contribution or Letter of Inquiry is required, the application deadline, and your fit assessment. The point of writing down the source alongside each field — "$5,000–$15,000, per the funder's 2026 guidelines page" rather than just "$5,000–$15,000" — is that six months from now, you or the client can tell at a glance what's a confirmed fact and what needs a re-check before the next application cycle.

If you'd rather not build that spreadsheet from scratch, the Funder Prospect Tracker in the toolkit described below is built with exactly those columns, so all that's left to do is fill it in with prospects you've already verified — it doesn't do the research for you, and it isn't meant to.