# Predictive Donor Analytics: How AI Finds Your Next Major Gift By EvolveFundraising editorial team Published: 2026-10-27 Updated: 2026-10-27 Category: AI and fundraising Canonical: https://getevolvefundraising.com/blog/predictive-donor-analytics ## Summary Predictive donor analytics uses the data you already hold (giving history, engagement and relationships) to estimate how likely each person is to give, give more or stop giving. The output is a ranked list: who to call this week, who is ready for a bigger ask and who is drifting away. It matters more than ever because fundraising growth now depends on a narrowing group of larger donors, and teams can't afford to spend time on the wrong people. ## Key takeaways - Start with reliable giving and engagement history. - Use scoring to prioritize human attention. - Evaluate predictions against actual results. Predictive donor analytics uses the data you already hold (giving history, engagement and relationships) to estimate how likely each person is to give, give more or stop giving. The output is a ranked list: who to call this week, who is ready for a bigger ask and who is drifting away. It matters more than ever because fundraising growth now depends on a narrowing group of larger donors, and teams can't afford to spend time on the wrong people. ## Why prediction matters now In 2025, dollars grew 5.0% while donor counts fell 3.6%, and the growth was driven almost entirely by major and supersize donors, according to the [Fundraising Effectiveness Project](https://afpglobal.org/news/fundraising-effectiveness-project-reports-strongest-revenue-growth-five-years-even-fewer). Bequests also jumped, per [Giving USA 2026](https://afpglobal.org/news/giving-usa-2026-what-fundraisers-need-know-year-end), which recommends leaning into planned giving. Translation: the next big gift is probably already in your database. Prediction helps you find it before it goes elsewhere. Few nonprofits use these tools yet. Nonprofit Tech for Good's 2025 survey found less than 1% of nonprofits use real-time fundraising intelligence ([Nonprofit Tech for Good](https://www.nptechforgood.com/101-best-practices/ai-marketing-fundraising-statistics-for-nonprofits/)). That's an opening for early adopters. ## What does predictive donor analytics predict? | Prediction | Question it answers | What you do with it | | --- | --- | --- | | Likelihood to give | Who in our database will give in the next 90 days? | Prioritize appeals and calls | | Upgrade potential | Who could give more than they do now? | Mid-level and major gift asks | | Lapse risk | Which current donors are drifting? | Personal outreach before they lapse | | Monthly conversion | Which one-time donors would give monthly? | Targeted monthly invitations | | Planned giving likelihood | Who might include us in their will? | Legacy conversations | ## Which signals matter most? **Giving behavior (the classic RFM model)** - **Recency:** how recently they last gave. - **Frequency:** how many gifts over time. FEP data consistently shows donors with more gifts retain far better. - **Monetary:** gift size and trend (rising, flat, falling). **Engagement signals** - Event attendance and volunteering. - Email opens and clicks. - Membership tenure and renewals (for associations, often the strongest signal). - Committee or board service, credentials, awards. **Relationship and capacity signals** - Relationships to existing major donors or board members. - Employer and matching gift eligibility. - External wealth screening, where you use it. The best models blend all three. Giving behavior tells you about the past; engagement tells you about the direction. ## Do you need a data scientist? No. You need three things: 1. **Clean data on one record.** Duplicates and split systems break any model. A donor who appears as three people gets three weak scores instead of one strong one. 2. **Enough history.** Two or more years of gift data is ideal. 3. **A system that scores and explains.** A score without a reason ("78 because: attended the conference three years running, gave twice last year, recently opened every email") won't be trusted or used. ## How to act on predictive scores Scores only matter if they change what staff do each week. 1. **Weekly top-25 list.** Each gift officer gets the 25 highest-potential people not contacted in 60 days. 2. **Lapse-risk alerts.** Donors whose risk jumps get a personal check-in, not another appeal. 3. **Segmented appeals.** High-likelihood, low-capacity donors get mail and email; high-capacity donors get a meeting. 4. **Measure.** Compare response rates and average gifts for top-scored vs. unscored groups each quarter. ## Common pitfalls - **Scoring dirty data.** Clean first ([post 17: Donor Data Cleanup Before Migration: Duplicates, Households and Soft Credits](https://getevolvefundraising.com/blog/donor-data-cleanup)). - **Ignoring non-giving engagement.** Especially costly for associations, where members' engagement predicts giving. - **Black-box scores.** Staff need the "why." - **Bias.** Check that the model isn't systematically down-ranking groups of donors because of how historical data was collected. ## How Evolve Fundraising approaches donor intelligence Evolve Fundraising scores relationships using the full picture on one record: gifts, pledges and recurring giving alongside membership tenure, event attendance, volunteering and email engagement. Scores come with plain-language explanations and suggested next actions, built on the same approach EvolveAMS uses to explain member renewal risk. Staff can also ask the AI assistant directly ("Which members who attended the last two conferences have never given?") and get a list in seconds. ## Frequently asked questions ### What is predictive donor analytics? The use of statistical models and AI on your donor data to predict who will give, upgrade, convert to monthly or lapse. ### What is RFM analysis in fundraising? Scoring donors by Recency, Frequency and Monetary value of their gifts. It's the foundation of most donor scoring. ### How much data do you need for predictive scoring? Ideally two or more years of clean gift history on deduplicated records, plus engagement data. ### Can small nonprofits use predictive analytics? Yes, if it's built into their fundraising software. Small teams benefit most because they have the least time to waste. ### Is wealth screening the same as predictive analytics? No. Wealth screening estimates capacity from external data; predictive analytics estimates likelihood from your own data. They work best together. **Find your next major gift.** See donor intelligence in Evolve Fundraising: [Request information from EvolveFundraising](https://getevolvefundraising.com/request-information). *Sources: AFP / FEP Q4 2025; Giving USA 2026 via AFP; Nonprofit Tech for Good, 2026 AI statistics.* ## Sources and further reading 1. [Fundraising Effectiveness Project](https://afpglobal.org/news/fundraising-effectiveness-project-reports-strongest-revenue-growth-five-years-even-fewer) — afpglobal.org 2. [Giving USA 2026](https://afpglobal.org/news/giving-usa-2026-what-fundraisers-need-know-year-end) — afpglobal.org 3. [Nonprofit Tech for Good](https://www.nptechforgood.com/101-best-practices/ai-marketing-fundraising-statistics-for-nonprofits/) — nptechforgood.com ## Related reading - [AI for Fundraising: A Practical Playbook for Nonprofits and Associations](https://getevolvefundraising.com/blog/ai-for-fundraising) - [Donor Retention Rate: 2026 Benchmarks, the Formula, and 9 Ways to Improve It](https://getevolvefundraising.com/blog/donor-retention-rate) - [Donor Data Cleanup Before Migration: Duplicates, Households and Soft Credits](https://getevolvefundraising.com/blog/donor-data-cleanup) - [Discuss your fundraising goals with the Evolve team](https://getevolvefundraising.com/request-information) ## Discovery [All articles](https://getevolvefundraising.com/blog) · [RSS feed](https://getevolvefundraising.com/rss.xml) · [Site guide](https://getevolvefundraising.com/llms.txt)