Børneriget Fonden exists to fund Mary Elizabeth's Hospital, a specialist hospital for critically ill children, young people and expecting families. The foundation's fundraising team is small, and every hour spent on manual donor research is an hour not spent building the relationships that fund the hospital. The challenge was a lead generation problem: a universe of 390,000+ potential donors, very little public signal on the most relevant ones, and a team with finite time to work through them. A key group in that landscape, Denmark's family foundations, is also the hardest to assess as very little data is typically publically available. However, a lack of public data does not mean a lack of relevance.

A lead-generation engine built around two questions.

1. Who to reach out to: a ranked shortlist of the ~300 Danish companies and foundations most aligned with Børneriget's cause — financially capable, purpose-matched, and confirmed reachable.
2. How to reach out: a network map that traces the shortest path from Børneriget's existing contacts to every high-priority prospect, written in plain language so the team knows exactly who can open the door.

A composite scoring model that pulls financial filings for capacity, then uses an LLM to assess each company's stated purpose, industry classification and published language for mission alignment. Both signals feed into a single relevance score. A network-graph model runs in parallel, tracing Børneriget's contacts against every candidate's board, ownership and founding structure up to four degrees of separation, and returns the shortest introduction path per prospect.

~300 high-priority prospects identified from 390,000+ active Danish companies.

Network context for every lead containing the shortest route from a known contact to the company, with up to four degrees of separation.

A tool Børneriget owns and can tune: scoring weights, financial thresholds and purpose keywords are all editable with no developers needed.

The scoring weights, keyword groups and financial thresholds reflect how Børneriget thinks about relevance i.e. which signals matter, and how much. Turning that expertise into a model requires close collaboration between domain experts and data experts.

Philanthropic giving runs on relationships. Knowing which companies are worth approaching is only half the problem; knowing who in your own network can open the door is what makes the list actionable. Solving both together changes how a fundraising team operates.

When thousands of candidates are scored across financial strength, purpose alignment and network proximity, a number alone is hard to act on. By combining parametric scoring with LLM-generated explanations for each recommendation, the output becomes a white box: the team can see what drove a candidate's ranking and walk into each conversation already knowing why this prospect is worth the call.