Machine Learning in Fundraising

AI is eating the world… but nonprofits are still serving sandwiches. While startups sprint ahead with AI, most nonprofits are stuck debating if ChatGPT is “ethical.” AI is NOT optional. It’s the single biggest force multiplier in history. Yet, most nonprofits are: Drowning in admin work Burning out on low-impact tasks Struggling with donor engagement Meanwhile, AI-driven orgs are: Automating back-office work Personalizing donor outreach Running impact programs with 10X efficiency Let’s talk about what nobody tells nonprofits about AI (with real evidence). 1. AI can 10X donor engagement. Most nonprofits still send generic mass emails. AI changes that. Harvard research shows personalized donor messaging increases retention by 80%. How? AI tools like Rasa and Drift tailor responses in real time. ChatGPT-style assistants craft hyper-personalized donation asks. AI sentiment analysis ensures every message hits the right emotional tone. Nonprofits using AI in fundraising see a 44% increase in donor conversion. 2. AI slashes admin work (so teams can focus on impact). Nonprofits waste 40% of their time on admin. AI eliminates that. AI automation can: Process tax receipts Automate grant applications Manage volunteer scheduling Example? GiveDirectly uses AI to verify beneficiaries, cutting admin costs by 70%. 3. AI predicts & prevents crises. Most nonprofits react after disasters strike. AI-driven analytics change that. Example? Red Cross uses AI to predict hurricanes and deploy aid faster. AI processes satellite data, social media, and weather reports. Early warnings improve response times by 50%. More lives saved, less money wasted. 4. AI makes small teams operate like big ones. Think AI is only for giant NGOs? Think again. Mama Hope used AI chatbots to handle donor FAQs, freeing 30% of staff time. Charity: Water automates donor follow-ups to boost retention. Team Rubicon uses AI logistics to deploy volunteers faster than FEMA. AI levels the playing field. 5. AI doesn’t replace humans, it amplifies them. Biggest fear? “AI will take our jobs.” Reality? AI eliminates low-impact tasks so teams can focus on real mission work. AI writes reports—humans build relationships. AI analyzes data—humans make decisions. AI sends emails—humans inspire action. The question isn’t “Will AI replace us?” The question is “How fast will we fall behind if we ignore it?” Nonprofits that adopt AI now will dominate the next decade. The biggest threat to nonprofits isn’t funding, it’s irrelevance. Want to get started? Pick ONE thing to automate this month: AI-powered donor messaging? (Try ChatGPT or Jasper) AI-driven grant writing? (Check out Grantable) AI for impact measurement? (Look into DataRobot) The nonprofits that embrace AI will scale 10X. The ones that don’t? They’ll keep serving sandwiches. With purpose and impact, Mario

Most prospect research is stuck in 2010. Here are 5 techniques that actually work in today's environment: 1. Behavioral data mining > wealth screening: Instead of just looking at capacity markers, analyze digital engagement patterns. Donors who consistently open emails about specific programs are signaling interest. We've found these behavioral signals predict giving 3x better than wealth indicators alone. 2. Social listening with AI tools: Set up automated monitoring of prospects' social media for life events, interests, and values alignment. The tools have become sophisticated enough to flag genuine opportunities without being creepy. 3. Collaborative intelligence gathering: Create systems for program staff, volunteers and board members to log prospect interactions in real-time. The collective intelligence of your entire organization is more powerful than any research database. 4. Relationship mapping visualization: Use software to visually map connections between current donors and prospects. These relationship webs reveal non-obvious pathways to new prospects that traditional research misses. 5. Predictive modeling for mid-level donors: Apply machine learning to identify which donors under $1,000 have major gift potential. The algorithms now accurately predict upgrade potential 18-24 months before traditional qualification methods. The organizations seeing breakthrough results aren't just gathering more data - they're gathering different data and analyzing it more intelligently. Which of these techniques have you tried? Let me know which one you'll implement next.

Charity Leaders & AI: Where Do We Start? 🤖 I've spent the last few years helping charities embed digital (and increasingly AI) into their core mission. AI was today's topic on the Third Sector Lab x SCVO Digital Senior Leaders Programme with me, John Fitzgerald and Maddie Stark Here's the questions charity leaders need to ask plus a few practical ways to move the conversation from hype to strategy 👇 The Big Questions We Need to Ask❓ - Where is AI already affecting our mission—positively or negatively? - How empowered (or anxious) do our staff and volunteers feel about AI? - Which parts of our work could AI actually improve (reach, impact, efficiency)? - Do we understand the risks—data, ethics, trust? How will we keep our values central? - Who else in our network is experimenting with AI and what are they learning? Five Practical Steps for AI-Ready Leaders 5️⃣ AI Impact Mapping 🗺️ Bring your team together. Map every touchpoint where AI could play a role - from fundraising and supporter comms to governance and frontline service. Pinpoint where the real wins and risks are for your charity. Staff & Volunteer Pulse Check 🩺 Run a session where people role-play different AI scenarios. What opportunities and anxieties bubble up? (Be ready for honest feedback!) Use it as a way to shape your AI literacy and support plans. Debate Real-World AI Use Cases 👥 Share case studies: the good, the bad, and the complex. Chatbots for helplines? Automated grant app sorting? Data-driven supporter segmentation? Debate - don’t sell - the practicalities and ethical red lines. Risk & Governance Tabletop 🎲 Role play as trustees, comms, digital leads, service staff—respond to an data breach as a result of AI usage or staff concerns about AI bias in recruitment. Work out who needs to be in the room when things go wrong, and what new protocols may be needed. Quickfire AI Experiment 🧪 Have your team test a popular AI tool - draft a donor email, summarise a board paper, generate a campaign image. Use Co-Pilot, ChatGPT, Perplexity, Claude, Gemini or whatever tool is most relevant to your needs. Compare notes: What worked, what failed, where was human oversight crucial? Make Space for Messy Conversations 🪢 - Is AI use visible or happening “off the books?” - What would success - or failure - with AI look like for us next year? - How can we work across the sector for stronger, more ethical approaches? - What are the values we refuse to compromise on, no matter what shiny AI tool we see? Don’t Forget: Make It Actionable 💪 - Finish your next senior team meeting with a commitment - Run a staff survey on AI - Pilot a small AI project - Join or create a sector AI peer group If you’ve taken baby steps, had a tough internal debate, or even failed spectacularly, or you just want to share a handy resource - I want to hear about it in the comments 👇

AI is only as powerful as the problems it solves. For nonprofits, one of the most fundamental challenges is knowing how much to ask for, and when. Ask too high, and you risk discouraging a gift. Ask too low, and you leave potential impact on the table. That’s why we’ve taken Intelligent Ask Amounts to the next level for GoFundMe Pro partners. Grounded in deep user research and powered by GoFundMe’s AI models, this improved version gives nonprofits the ability to dynamically optimize campaigns for what matters most: one-time revenue, conversions, recurring gifts, or a balanced mix. The ask amounts adapt in real time to donor behavior and campaign goals—helping nonprofits drive more sustainable giving. The best part? These improvements are to a product that has already delivered results. For example: the National Civil Rights Museum used Intelligent Ask Amounts during key giving moments and saw a 62% increase in average gift size on December 31st year-over-year, along with other strong gains. (I’ll link the case study with more details in the comments!) What makes me proud isn’t just the AI, it’s the teamwork behind it. Three product pods, Applied Science, Research, CX, Legal, Marketing, Comms and more all came together to turn a complex fundraising challenge into a solution that’s both powerful and practical. Because at the end of the day, innovation is only meaningful when it helps nonprofits raise more with less friction—so they can focus on their mission. 👉 Learn more here: https://gfme.co/47CvtSc 

Welcome to the Future of Fundraising. The best question you can ask yourself when evaluating AI is “Will this directly drive revenue or will this create efficiencies?” If your answer is revenue, you're probably looking at Autonomous AI. If your answer is efficiency, you’re looking at AI Enablement. Developing a clear grasp of Autonomous AI versus AI Enablement is a skill all fundraising leaders need to develop now, because today’s choices will drive tomorrow’s growth. When I co-founded Gravyty almost a decade ago, I was a frontline fundraiser who needed to operate more efficiently to reach more of the donors in my portfolio. What we created was the first AI Enablement tool for fundraisers that could self-write emails for me to edit and send to keep me on top of outreach. This is a great example of AI Enablement, tools that draft emails, summarize insights, predict giving potential, analyze CRM data, or prioritize donor outreach lists. Those key words–draft, summarize, predict, analyze, prioritize–are often akin to AI Enablement. AI Enablement tools are measured in the efficiencies that they produce, essentially helping employees do their current job well. Autonomous AI is an entirely different category. Unlike AI-enabled tools, Autonomous AI is responsible for an entire job from start to finish, independent of its human colleagues, as a standalone solution. In fundraising, this critical difference means that it is accountable for the same outcomes as a staff member. Unlike AI Enablement, in our industry, Autonomous AI can be measured on direct revenue generation and pipeline growth. Autonomous Fundraising, and the work of the Virtual Engagement Officer, exemplifies this difference. Bucknell University’s VEO, Lauren, manages a 1,000 donor portfolio and has raised $450,000 while outperforming a control group on every metric: dollars raised, renewals, participation, and gift increases. The VEO operates just as a traditional gift officer would, using cultivation activities that lead donors to give. For this reason, we can measure the VEO by the same revenue-generating standards as every other fundraiser on the team. Rather than focusing on doing the current scope of work well, Autonomous AI has the unique ability to be applied to scale areas of growth that were previously thought impossible. As we evaluate AI and bring it into our organizations to improve fundraising, the donor experience, and ultimately our missions, asking critical questions about outcomes will become increasingly important.

I recently analyzed 10,000+ donor records using AI. The results were shocking. The traditional wealth screening had missed: * 12 highly engaged major gift prospects hiding in the under-$100 donor pool * 8 donors who had capacity to give 10x their current level and had been giving signals for years * 1 past board member who'd introduced us to several in his network followed by... crickets The difference? AI doesn't just look at wealth indicators. It analyzes behavioral patterns, engagement history, and external factors that traditional methods miss. The future of donor research isn't about having more data. It's about having smarter insights. What's your biggest question about implementing AI in your fundraising process? PS: Salesforce had something similar happen. Marc Benioff said they identified over 100,000 people nobody had followed up with!

#WednesdayWisdom Most nonprofits won't get to choose whether they deal with AI. The only real choice is whether you do it strategically or reactively. 𝗥𝗲𝗮𝗰𝘁𝗶𝘃𝗲 𝗔𝗜 looks like this: A peer organization announces an "AI-powered donor platform." Your board asks, "Why aren't we doing this?" You rush to buy a tool so you can say you're "doing AI." The result? Tools that don't fit your workflows. Staff who were never brought along. Budget tied up in tech that quietly creates more work instead of freeing capacity. I've watched this pattern repeat across nonprofits and foundations including some of the most sophisticated organizations on paper. 𝗦𝘁𝗿𝗮𝘁𝗲𝗴𝗶𝗰 𝗔𝗜 adoption feels very different: ✔️ You map where staff time actually goes and name your mission bottlenecks. ✔️ You choose one workflow where automation would create the highest mission ROI. ✔️ You bring staff in early as co-designers, not passive recipients. ✔️ You pilot small, measure honestly, and expand based on real results. ✔️ You build board literacy around capacity, risk, and governance — not just technology trends. Take the American Cancer Society's approach: in a 2022 fundraising campaign, they used machine learning to optimize ad strategy—driving donation revenue 𝟭𝟭𝟳% above benchmark and engagement rates of nearly 𝟳𝟬% on their rich media units. Their Director of Media Strategy put it plainly: "Every bit of our campaign spend needs to be optimized for the best possible performance." That's strategic AI starting with a clear mission outcome and building the technology around it. Before you sign a contract or add another tool to your stack, ask your leadership team: → Where is manual work currently limiting our mission delivery? → What would become possible if we reclaimed 10 hours per week of staff capacity? → Which staff, board, and partners need to be in this conversation from day one? → What does success look like in mission terms, not technology terms? This is the foundation I build with executive teams when we design AI roadmaps together, so AI becomes a 𝗰𝗮𝗽𝗮𝗰𝗶𝘁𝘆 𝗴𝗶𝗳𝘁, not a 𝘄𝗼𝗿𝗸𝗹𝗼𝗮𝗱 𝘁𝗵𝗿𝗲𝗮𝘁. If you're feeling that "do something with AI" pressure but don't have a clear first step, DM me with "ROADMAP" (or drop it in the comments). I'll share the framework I walk EDs, C-suite leaders, and boards through in our AI readiness sessions. Your mission is too important to automate on autopilot.

AI is quietly rewriting what it means to be good at fundraising. Most people in the sector are still using AI the same way. Write a prompt, get an output, tweak it, move on. Useful. But you're still touching every step yourself. That's not where this is going. The shift happening right now is toward agentic AI. The difference is simpler than it sounds. Today: "Draft a follow-up email for this donor." Agentic AI: "Follow up with our lapsed donors from last year." Same goal. Completely different relationship with the work. Instead of handing you a draft, it researches the donors, personalises messages based on their giving history, and schedules the sends. You set the goal. It figures out the steps. For fundraising teams, that changes two things worth paying attention to. The first is where your time goes. The job shifts from doing each task inside a workflow to designing the workflow itself, and making the judgment calls that AI can't. The second is what limits your capacity. A small team can do the work of a much larger one. The constraint stops being "do we have enough people" and starts being "have we designed this well enough." The fundraisers who thrive in this environment will be the ones who are best at designing excellent processes once, and letting AI run them at a scale no team could match manually. That's a different skill set than most fundraising careers have trained for. Worth starting to build it now.

Within two years, your donors will expect 𝗲𝘃𝗲𝗿𝘆 𝗹𝗲𝘁𝘁𝗲𝗿 𝘆𝗼𝘂 𝘀𝗲𝗻𝗱 𝘁𝗼 𝗳𝗲𝗲𝗹 𝗹𝗶𝗸𝗲 𝗶𝘁 𝘄𝗮𝘀 𝘄𝗿𝗶𝘁𝘁𝗲𝗻 𝗷𝘂𝘀𝘁 𝗳𝗼𝗿 𝘁𝗵𝗲𝗺. Not "Dear Susan" instead of "Dear Friend." That's not personalization. 𝗧𝗵𝗮𝘁'𝘀 𝗮 𝗺𝗮𝗶𝗹 𝗺𝗲𝗿𝗴𝗲 𝗳𝗿𝗼𝗺 𝟭𝟵𝟵𝟴. I'm talking about a fundraising letter that references the specific program Susan cares about, acknowledges how long she's been giving, mentions the outcome her last gift helped create, and makes an ask that reflects her actual capacity and history. 𝗘𝘃𝗲𝗿𝘆 𝘀𝗲𝗻𝘁𝗲𝗻𝗰𝗲 𝘁𝗮𝗶𝗹𝗼𝗿𝗲𝗱 𝘁𝗼 𝗵𝗲𝗿. Now multiply that across your entire donor file. This used to be impossible at scale. You could personalize letters for your top 50 major donors if your development director had the time (they didn't). Everyone else got the segmented version. Maybe three or four letter variants if you were sophisticated. 𝗧𝗵𝗮𝘁 𝘄𝗮𝗹𝗹 𝗶𝘀 𝗴𝗼𝗻𝗲. AI tools are already capable of generating truly personalized donor communications, not templates with fields swapped in, but letters where every paragraph reflects what you know about that specific donor. Their giving history. Their event attendance. Their stated interests. The same is true for emails, texts, and stewardship touches. And here's what most nonprofit leaders aren't thinking about yet: 𝘁𝗵𝗶𝘀 𝘄𝗶𝗹𝗹 𝘀𝗼𝗼𝗻 𝗳𝗲𝗲𝗹 𝗻𝗼𝗿𝗺𝗮𝗹. The first time a donor receives an annual report that reads like it was written for them personally, highlighting the programs they funded and the outcomes they care about, it will feel surprising. Maybe even a little unsettling. But that reaction won't last. Within a few years, donors will expect it. The generic version will feel lazy by comparison. The nonprofits that start building toward this now have a real window of advantage. Not because the technology is hard to access, but because 𝘁𝗵𝗲 𝗿𝗲𝗮𝗹 𝗯𝗼𝘁𝘁𝗹𝗲𝗻𝗲𝗰𝗸 𝗶𝘀 𝘆𝗼𝘂𝗿 𝗱𝗮𝘁𝗮. Personalization at scale requires clean, rich donor records. Giving history, interests, communication preferences, engagement data. If that information lives in your database, you can start producing truly personalized letters and emails today. If it doesn't, that's the project to prioritize this year. Not because AI told you to. Because your donors are about to start expecting it. Where is your organization on this spectrum? Still sending segmented letters, or already experimenting with true personalization?