How to Use AI to Predict Which Major Donors Will Reduce Giving

Capacity represents the donor's financial wherewithal to make gifts at various levels. AI systems analyze multiple wealth indicators to assess both current capacity and changes over time that might signal shifts in giving ability.

Key capacity indicators AI models analyze:

• Real estate holdings: Individuals with at least $2 million invested in real estate are 17 times more likely to donate than average, making property ownership a powerful capacity signal.
• Stock ownership and SEC filings: Public filings reveal stock trading activity and ownership stakes, indicating both wealth and liquidity for potential gifts.
• Political contributions: Donors giving over $2,500 to political campaigns are 14 times more likely to make charitable contributions, signaling both capacity and engagement.
• Corporate board service: Board positions indicate leadership capacity, professional networks, and typically significant wealth.
• Business ownership: Equity stakes in private companies, particularly in successful ventures, represent substantial but often illiquid wealth.

Critically, AI models don't just assess static capacity—they monitor changes over time. A major donor who sells a business, experiences a stock market downturn affecting their portfolio, or enters retirement may face reduced capacity even if their commitment remains strong. Predictive models that incorporate real-time wealth screening data can flag these capacity shifts, allowing fundraisers to adjust asks accordingly rather than inadvertently requesting gifts the donor can no longer comfortably make.

Propensity measures how likely someone is to give based on their behavior, engagement patterns, and philanthropic history. Unlike capacity, which looks at financial ability, propensity focuses on demonstrated giving behavior and engagement signals that indicate ongoing commitment—or warning signs of decline.

Behavioral signals AI models track for propensity assessment:

• Giving recency, frequency, and monetary value (RFM): Changes in these foundational metrics often precede larger shifts. A donor who moves from quarterly to annual giving, or who gives late in response to appeals when they previously gave proactively, may be signaling declining engagement.
• Event participation: Attending galas, site visits, or donor appreciation events indicates active engagement. When a previously regular attendee begins declining invitations, it's an early warning sign.
• Communication engagement: Email open rates, response times to fundraiser outreach, and participation in surveys all signal interest level. Declining engagement across these channels often precedes giving reductions.
• Volunteer activity: Major donors who volunteer, serve on committees, or take on advocacy roles demonstrate deeper commitment. Withdrawal from these activities can indicate wavering support.
• Gift designation changes: Shifts from unrestricted to restricted giving, or from general operations to specific programs, may indicate declining trust in organizational leadership or desire for more control over gift use.

Machine learning excels at identifying subtle patterns in these behaviors that human fundraisers might miss. For example, a donor who continues giving at the same level but whose engagement across multiple other dimensions has declined by 20-30% over six months represents a high-risk retention case—even though their giving hasn't changed yet. Predictive models can surface these multi-factor patterns and score donors on propensity to continue, reduce, or cease giving.

Affinity measures how well your mission aligns with a donor's values, interests, and philanthropic priorities. Even donors with high capacity and demonstrated propensity may reduce giving if their priorities shift or if they perceive your organization is moving away from the issues they care about most.

Affinity indicators AI can analyze:

• Giving to peer organizations: Tracking public gifts to similar nonprofits reveals shifting interests. A children's hospital donor who begins making major gifts to environmental causes may be signaling evolving priorities.
• Board service at other nonprofits: Taking leadership roles elsewhere often indicates where donor passion truly lies. This doesn't always mean reduced giving to your organization, but it's a factor to monitor.
• Program-specific engagement: Donors who attend events or make gifts tied to specific programs reveal their affinity focuses. If your organization shifts priorities away from these areas, you risk losing their support.
• Personal connections: Relationships with specific staff members, board members, or beneficiaries often drive major donor commitment. When these individuals leave or relationships weaken, affinity can decline rapidly.
• Life changes: Major life events—children graduating from your school, personal health issues shifting priorities, relocation, or family changes—can all affect mission affinity.

AI tools can monitor public philanthropic activity, track engagement with specific programs through your CRM data, and even analyze communication sentiment to gauge affinity levels. When combined with capacity and propensity data, affinity scores help create a complete picture of donor retention risk. A donor with high capacity and propensity but declining affinity represents a different intervention opportunity than one whose capacity has changed but whose mission commitment remains strong.

Before investing in any AI tool, evaluate your donor data quality. Predictive AI will only be as good as the data it's built on. Organizations with poor data hygiene waste money on sophisticated models that produce unreliable results.

Data quality checklist:

• Complete giving history: At minimum, you need 2-3 years of donation data including amounts, dates, designations, and appeals responded to.
• Engagement tracking: Event attendance, volunteer activities, email interactions, and meeting notes should be systematically recorded in your CRM.
• Demographic information: Location, employment, board service, and other affinity indicators help models assess retention risk.
• Relationship data: Assignment to specific development officers, relationship strength ratings, and stewardship activities all contribute to predictive accuracy.

If your data is incomplete or inconsistent, dedicate 3-6 months to data cleanup before pursuing predictive models. Organizations using even basic predictive analytics improved donor retention by 12% on average—but only when working with clean, well-structured data.

The most common implementation failure isn't technical—it's human. Organizations invest in predictive analytics but don't train development staff to interpret scores and act on insights. The result: powerful predictions sit unused while donors quietly disengage.

Essential training components:

• How to read retention-risk scores: Understand what a 0.7 risk score means versus 0.9, and what thresholds trigger intervention.
• Which factors drive each score: Look beyond the number to understand why a donor is flagged (capacity change? Engagement drop? Affinity shift?).
• Appropriate intervention strategies: Different risk factors require different responses. Capacity issues need one approach; affinity concerns need another.
• How to integrate predictions into portfolio management: Build regular review of high-risk donors into weekly routines, not just annual planning.
• When to trust AI versus personal judgment: Models are powerful but not infallible. Fundraisers need frameworks for when to follow predictions and when to rely on relationship knowledge.

Schedule quarterly refresher training and create simple job aids that help development officers quickly translate retention-risk scores into action plans. The organizations seeing 12%+ retention improvements from predictive analytics are those that make AI insights a core part of daily fundraising practice, not an occasional report to review.

Knowing a donor is at risk is only half the battle. Your team needs clear, proven intervention strategies to deploy when the AI flags high-risk donors. These playbooks should match intervention tactics to specific risk factors.

Example intervention playbook structures:

Document these playbooks clearly so any development team member can execute them consistently. Track which interventions work best for different risk profiles, and refine your approaches over time based on results.