What Your Senior Fundraisers Know That Your CRM Doesn’t Yet
Every fundraising organization has at least one person who seems to know instinctively who to call. They look at a list of names and immediately identify the three worth pursuing this week. They remember that a particular donor went quiet after a leadership change and is probably ready to re-engage. They know that a parent who attended two campus events in the last year is worth a personal call even though their giving history is modest.
That judgment is not magic. It is pattern recognition built from years of working donor relationships, and it is genuinely valuable. It is also, in most organizations, almost entirely undocumented.
When that person leaves, retires, or moves to a different role, the judgment goes with them. The CRM has the data. It does not have the interpretation.
Why Expert Knowledge Does Not Transfer Automatically
CRMs are designed to store facts. A contact gave on this date, at this amount, and attended this event. Those facts are recorded accurately. What is not recorded is what those facts mean in combination — how a development officer weighs recent engagement against historical giving, or how they think about a lapsed donor differently from a non-donor with similar demographic characteristics.
That interpretive layer is where expert fundraising judgment lives. And because it was never designed to be captured, it rarely is. New staff members learn it over years of shadowing senior colleagues and making their own mistakes. It gets transmitted informally, inconsistently, and incompletely.
The result is that fundraising organizations are more dependent on individual expertise than they realize. The system works as long as the right people are in the right roles. When that changes, performance drops — not because the data got worse, but because the interpretation got lost.
This is one of the core problems that AI-powered donor scoring is well-positioned to solve, but only if the extraction happens intentionally. The data agent does not know how your senior development officers think about donor potential. You have to tell it. How to build a donor scoring system inside HubSpot covers how that translation from human judgment to scoring prompt works in practice.
What Expert Fundraising Judgment Actually Contains
Before you can capture institutional knowledge in a scoring system, it helps to be specific about what that knowledge actually consists of.
Most experienced development officers are applying a mental model that combines several distinct types of assessment, often simultaneously and without being fully conscious of it.
Profile assessment is the evaluation of whether a contact fits the characteristics of a likely donor: their relationship to the institution, their career stage, their demonstrated capacity, their connection to the mission. This is the equivalent of fit scoring in a sales context — it answers the question of whether this person is the right kind of prospect, before any outreach has happened.
Signal weighting is the judgment about which behavioral signals matter most and how to interpret them in combination. An alumna who has never given but has attended four events in three years is being weighted differently from one who gave once five years ago and has not engaged since. An experienced fundraiser does this weighting automatically. A scoring system needs it specified explicitly.
Timing judgment is the assessment of when a contact is likely to be in a receptive window. Career milestones, life events, reunion years, and giving campaign cycles all affect timing. Senior staff carry this awareness almost as second nature. It rarely gets written down anywhere.
Relationship history is the accumulated context from prior interactions — conversations, asks, responses, concerns raised. Some of this lives in CRM notes if staff have been diligent about logging it. Much of it does not. Your CRM has the data — so why can’t your team act on it? covers the gap between what is in the system and what experienced staff actually use to make decisions.
How to Extract the Knowledge
Getting institutional knowledge out of people’s heads and into a format a scoring system can use is a process, not a meeting. It requires asking the right questions and being willing to push past the first answers, which are usually too general to be useful.
Start with the output, not the input. Ask your senior development officers to describe the last five contacts they prioritized for outreach and explain why each one made the list. Then ask about the last five they decided to deprioritize and why. The contrast between those two sets of decisions usually reveals the criteria faster than asking directly about criteria.
From there, push for specificity. When someone says they look for donors who are engaged, ask what engaged means. How many events? Over what time period? Does type of event matter? When someone says they weight giving history heavily, ask how they think about lapsed donors versus consistent small donors versus one-time major donors. Each of those distinctions is a scoring rule that can be written into a prompt.
Pay particular attention to the edge cases. The contacts where the answer is not obvious — where a strong signal in one area is offset by a weak signal in another — reveal the most about how the scoring logic should handle complexity. Those are also the cases where a well-written prompt produces the most value, because they are exactly the situations where manual review is most time-consuming. Why your fundraising team is spending more time on admin than donor relationships describes what that time cost looks like in practice.
Translating Knowledge Into a Scoring Prompt
Once you have a clear picture of the criteria your senior staff use, the translation into a scoring prompt requires making three things explicit: what signals the system should look at, how to weight or combine those signals, and what output to produce for each possible combination.
The signals are the data points you identified in the extraction process — giving history, event engagement, relationship type, capacity indicators, suppression flags. Each one needs to map to a specific field or associated record in HubSpot that the data agent can access.
The weighting is where the nuance lives. A prompt that says score contacts based on their giving history and event engagement is not specific enough to produce consistent output. A prompt that says contacts with three or more gifts in the last five years and attendance at two or more events in the last two years qualify as high priority, unless a suppression flag is present, in which case they are low regardless of other signals, is specific enough to apply consistently across thousands of records.
The output should be as simple as the use case allows. For most donor scoring applications, high, medium, and low is sufficient. Adding more tiers creates complexity without adding proportional value, and it makes the output harder for fundraisers to act on quickly. What is HubSpot’s data agent and what can it actually do? explains how the prompt connects to the data agent’s output mechanism.
What Happens When the Knowledge Is Captured
When the scoring logic reflects the actual judgment of your best development officers, a few things change.
The first is consistency. Every contact in the database gets evaluated against the same criteria, applied the same way, regardless of who is on the team that week. A new staff member working from a scored list is making outreach decisions grounded in the expertise of the organization’s senior staff, not starting from scratch.
The second is scale. The judgment that one senior development officer can apply to a few hundred contacts manually gets applied automatically to tens of thousands. The contacts that would have been missed — because there was not enough time to get to them, or because they did not appear in the most recently reviewed list — are now in the output.
The third is durability. When institutional knowledge is captured in a scoring prompt, it does not leave when a person leaves. It can be updated as the organization learns more, but it does not disappear. That is a meaningful change for organizations that have experienced turnover in senior development roles and felt the impact on fundraising performance. Your donor database is full of answers — here is how to find them covers how all of these pieces fit together into a complete donor prioritization system.
When FSM worked with a university fundraising team to build their scoring system, the prompt development process started exactly here — with conversations with senior staff about how they actually thought about donor potential. The output of those conversations became the scoring criteria. The scoring criteria became the prompt. And the prompt, running against 100,000-plus records, produced results that matched what the senior staff would have produced manually, at a scale they never could have achieved on their own. A recent episode of AI in Action covers that process in detail.
Listen to the full podcast episode here.