Grant & Foundation Prospect Research

Score fit only after you have the foundation's current packet. Retrieve guidelines, program pages, recent grants lists, and the latest Form 990 when it exists. Write from those documents. A CRM note, a hallway briefing, or last year's calendar is not a source.

Name every source. Capture the document title or URL, the date you retrieved it, and the section you used. If the guidelines PDF and the website disagree, cite both and flag the conflict. The brief does not pick a winner. Grants staff resolve conflicts when they decide whether to pursue.

Form 990s describe history, not a promise. A 990 can show that a foundation has given in your sector, at a certain size, or to peer campuses. It cannot show that they will fund you, that the program still exists, or that this year's dollars are open. Treating a 990 as a commitment is a failure mode: filings lag the current cycle, restricted gifts look like open invitations, and a past grantee list is not an eligibility rule.

Advancement and finance systems (Salesforce, Ellucian, Workday, Anthology, and similar platforms) may already hold a foundation record, a prior ask, or a contact. Use them to see what your institution has already requested and received. Do not use them as a stand-in for the foundation's own materials. If the CRM lists a deadline the current guidelines do not, treat the CRM as stale until a person confirms it.

The brief has three working parts: program fit, funding cycle, and constraints. Each claim needs a cite to a document in the packet. If you cannot point to a sentence, the claim does not belong in the brief.

Fit is a comparison, not a compliment. State the foundation's published priority in its own language, then state the campus program in operational terms, then name the overlap and the gaps. "They fund education" is not fit. "Guidelines section 2 names early-career teacher residencies in high-need districts; the college of education runs a yearlong residency placed in three partner districts" is fit, with a cite.

Cycle is calendar language only when the packet contains calendar language. Open windows, LOI dates, invited-proposal dates, rolling review, and "no unsolicited proposals" all belong here when they appear in guidelines or on the program page. If the packet is silent, the cycle field stays empty. Inventing a deadline so the brief looks complete is a failure mode. A made-up date will migrate into a workplan, a dean's calendar, and a rushed narrative that misses the real window.

Constraints are the rules that kill or reshape an ask: geography, eligible applicant types, capital versus program, matching requirements, fiscal-sponsor rules, and "one proposal per institution." Cite each rule. Do not paraphrase away a restriction to make the opportunity look cleaner.

Illustrative pass: a college of education is considering a regional family foundation. The agent retrieves the current guidelines PDF, the grants-made page, and the latest 990. Fit is supportable because the guidelines name early-career teacher residencies and the college operates one; the brief cites that section. Cycle stays blank because neither the PDF nor the website lists an LOI or full-proposal date. Constraints include in-state eligibility and a no-capital rule, both cited. The 990 shows prior gifts to two peer residencies; those go in a history note labeled as history, not as a commitment. A reviewer should reject any sentence that inserts "LOI due March 1" when that date does not appear in the packet.

Quality here means a brief you can audit. Every filled field traces to a retrieved source. Empty stays empty when the source is missing, stale, or contradictory without a resolution.

Do not backfill from adjacent documents. A 990 year-end does not imply an application deadline. A "last funded in spring" note in the CRM does not imply this spring. A press release about a new initiative does not replace guidelines that have not been updated. If the only materials you have are a homepage and an old 990, say so and leave fit details, cycle, and several constraints blank rather than inferring them.

Stale packets are a separate problem from missing packets. Guidelines marked "updated 2019" with no later PDF should be flagged as potentially superseded. The brief can quote them, but it should not present them as current policy. Grants staff decide whether to call the program officer before any scoring is treated as actionable.

The output is a structured prospect brief per opportunity, not a ranked slate and not a go/no-go decision. Ranking belongs later, after humans have checked cites, conflicts, and campus capacity. The machine's job is to keep the evidence visible.

A person in grants or foundation relations reads the brief before anyone updates the opportunity record or starts a narrative. Review is not a courtesy pass. It is the step that catches invented dates, 990s treated as pledges, and fit language that quotes the campus strategy deck instead of the foundation.

Check three things in order. First, do the cites open the right documents and the right passages. Second, did empty fields stay empty, or did someone complete them from memory. Third, does the brief still read like research, or has it drifted into a draft proposal.

Do not submit from the brief. The brief is an internal research artifact. It is not an LOI, a full proposal, a budget, or a letter of inquiry. Submitting from the brief is a failure mode: unreviewed claims reach the funder, the wrong campus unit is listed as applicant, and a constraint you skipped becomes a decline reason you cannot walk back. Grants staff own the submission, including who signs, which program is named, and whether this cycle is even open.

After review, a human may write a short opportunity note into Salesforce, Ellucian, Workday, Anthology, or whichever system your shop uses for cultivation. That note should preserve cites and blanks. Flattening the brief into "strong fit, due soon" undoes the work.

Foundation prospecting sits beside, not instead of, individual giving research. A foundation brief answers whether a published program matches a campus project. It does not answer whether a trustee will make a personal gift, how a campaign should segment households, or whether an alum is a planned-giving candidate.

Use major gift prospect identification when the question is a person or family with capacity and a relationship, including foundation board members acting as individuals. Keep those files separate from the institutional foundation record so a personal ask does not collide with a pending proposal.

Use campaign segmentation optimization when you are grouping audiences for a campaign, not when you are scoring a single RFP. A foundation that has never accepted unsolicited proposals should not land in an email segment because a model likes the industry code.

Household models such as an alumni giving propensity model and a planned giving propensity model estimate likelihood for people. They are weak evidence for a foundation's published priorities. Do not let a high alumni score on a foundation staffer's record rewrite the guidelines.

Keep the artifacts distinct: one cited brief per foundation opportunity, reviewed by grants staff, with blanks where the packet is silent. That is the quality bar. Everything else (cultivation sequence, proposal, and submit/no-submit) stays a human decision.