The A.I. That’s Already Raising Money

Keep up with everything happening in fundraising by signing up for our Philanthropy Today newsletter and joining our LinkedIn group for fundraisers.

Ever since the emergence of ChatGPT and other more powerful artificial intelligence tools, some professionals have worried that A.I. will take their jobs. A new tech start-up, Givzey, will put that concern to the test in fundraising. The company has signed up 13 organizations that will spend 18 months using an autonomous fundraiser — an A.I. tool that will interact with donors in hopes of raising money for the institutions.

Adam Martel, CEO of Givzey, says his goal is for the A.I. to raise money from donors and that he doesn’t believe it is taking jobs away from real-life fundraisers.

“We have a staffing crisis in fundraising,” Martel says. “Anything we can do to help organizations get to more donors, in addition to helping them hire as many fundraisers as possible, is a good thing. This isn’t going to replace fundraisers because there’s such a lack of fundraising talent out there.”

A.I. and Fundraising

1. Technology

   Fundraisers Reach for A.I. to Boost Productivity

2. Opinion

   ChatGPT Poses Ethical Quandaries for Nonprofits. Here’s How to Confront Them.

3. Advice

   How Some Fundraisers Are Using ChatGPT Technology to Do Their Jobs Better

Martel says the goal of the autonomous fundraiser is to reach donors who don’t have a gift officer assigned to them — and that most organizations wouldn’t hire a gift officer to serve — but who could benefit from more personalized engagement.

“Eighty percent of the giving pyramid isn’t touched,” Martel says. “They’re not talked to, regardless of what level they’re giving at, because only 10 to 20 percent of the giving pyramid is managed by major-gift officers or fundraisers. They are wonderful, but they can’t scale.”

Martel views the autonomous fundraiser as a way to bring that personal touch to a bigger chunk of donors. While his positivity about A.I. in fundraising abounds, others in the field are approaching the idea of using A.I. to talk to donors with caution.

“I’m not anti-autonomous fundraisers,” says Nathan Chappell, co-author of The Generosity Crisis and an executive at the data firm DonorSearch. “But I am skeptical and concerned about this idea. Just because the technology allows us to do this, it doesn’t mean we should.”

How Does the A.I. Fundraiser Work?

Givzey’s autonomous fundraiser is a computer-generated avatar that looks like a real person and speaks to donors in plain language about the institution it represents. Current fundraiser avatars can be male or female, white or Black, and Martel hopes to expand the diversity of the avatars. Donors have to opt in to communicating with the autonomous fundraiser, and they are clearly informed that it’s not a real person.

Givzey is testing the autonomous fundraisers through research and development agreements at 13 institutions, mostly universities, along with a hospital and a nonprofit research lab. The organizations are paying to use the product. Martel declined to disclose the exact price, but did say that it was more than the $15,000 to $20,000 that Givzey hopes to charge if the company can get to scale.

The autonomous fundraiser, ideally, will communicate with donors about upcoming events, projects, and institutional offerings. It’s “a liaison between the donor and the organization,” according to Martel.

The A.I. fundraiser cannot currently have real-time conversations with donors. Martel says he hopes to have a beta version of an autonomous fundraiser that can carry on real-time conversations by October, with a completed version by the end of the year. Right now, its communication is asynchronous: The autonomous fundraiser sends messages via text or email, which are, at times, vetted by actual fundraisers.

For example, if organizations connect the A.I. tool to their donor-relationship management system, the autonomous fundraiser can give supporters information about previous gifts they’ve made to the university.

Another potential interaction: When an alum adds the “open to work” badge on their LinkedIn profile, the autonomous fundraiser can text or email to ask if the alum would like to connect with someone at the university’s career center who can help with their job search. Martel says that in early tests, “they’ve been taking us up on the offer, and there’s an automatic introduction between the donor and the head of career services.”

Autonomous Fundraiser in Action

The College of Charleston did a small round of tests with the autonomous fundraiser over the summer and — like all the organizations testing the A.I. — will offer it to 1,000 donors this year. In early tests, some donors have chosen not to work with the new technology. Fundraisers at the college want to use the tool to cultivate donors they are trying to encourage to give again or give more consistently. That includes people who have given between $25 and $50 and lapsed donors who have not given in the past year. The college will also be including some nondonors in that pool.

Data helps identify factors that drive donor retention, says Dan Frezza, chief advancement officer of institutional advancement at the college. “One of which is giving level: A $100 donor is infinitely more likely to renew than a $25 donor,” he adds.

Frezza wants to see if autonomous fundraisers can strengthen ties to donors who typically have been difficult to retain, including small-dollar donors and people whose donations came as part of an event ticket price. Holding on to donors, Frezza says, is much less costly than attaining new donors.

“For me, it’s: How does this compare to what a normal world would have otherwise looked like,” he says. “We know what our costs have been historically through the major channels of mail. How does this new channel compare? That’s what I’m looking for.”

This fall, Frezza plans to use the autonomous fundraiser in much the same way a development department might use a series of emails to contact donors. “It’s a yearlong relationship,” Frezza says. “It would say, ‘I am Dan, your autonomous fundraiser. I’d like to invite you to homecoming.’ Then maybe the week of homecoming, ‘You’re registered. Here is your schedule for homecoming this week.’ So all of those pieces, to me, is what differentiates it from just an email on a channel.”

Frezza says the goal would be to communicate with donors in different ways — both fundraising appeals and other types of information — and to build donors’ relationship with the college.

William & Mary also joined the autonomous fundraiser research project to connect with donors who don’t get enough personalized outreach.

“We’ve got hundreds of thousands of prospects, and it’s nearly impossible for all of our gift officers to be able to reach those individuals,” said Meghan Palombo, associate vice president for annual giving and philanthropic engagement at the university. “For us, this is a way for us to think about how might we deploy another resource to reach them.”

A test this summer with just 45 donors showed about a 9 percent response rate: People either engaged with the autonomous fundraiser or made a gift. “A 9 percent opt-in rate or engagement rate is not terrible when you look in the annual-giving space,” Palombo says. “But 45 donors is a really small subset, so we are looking to expand the sample size.”

Because the autonomous fundraiser is new, how it will work in practice isn’t clear. This summer, Givzey tested the autonomous fundraiser with 119 donors from William & Mary and the College of Charleston. Martel says that interaction led to 13 gifts totaling $3,400 — and that he hopes the larger 18-month trial will generate similar returns.

The institutions trying it this fall are already thinking of additional ways to use the tool. The College of Charleston wants to reimagine phone-athons, which were a popular fundraising tactic in the past because they used low-wage college students to call donors and ask them to give. But times have changed. With caller ID and more people refusing to pick up or being rude if they do, the amount of money phone-athons raise has decreased. Frezza wants to try a new twist on the model.

Givzey

“I can use our autonomous fundraiser to set up appointments with students,” he says. “Instead of having 70 students dialing all day, hoping to get a fraction of people to respond, I can now hire 10 students to fill five hours a day where our autonomous fundraiser can begin a relationship and actually schedule times with that student caller.”

Unanswered Questions

The institutions participating in Givzey’s research project are excited, but tech experts have concerns. Chief among them: whether autonomous fundraisers would take jobs from humans and if they can be trained to raise money ethically.

“I, generally, am wary of any A.I. solution that doesn’t involve humans,” says Afua Bruce, a technology consultant and author of The Tech That Comes Next. She worries these tools can’t “fail safely” and could harm people.

DonorSearch’s Chappell and other technology observers echo those concerns, arguing that the nonprofit world needs to set guidelines for the responsible use of A.I. in fundraising. For example, A.I.’s ability to store information and define best practices without regard to ethics is a big concern.

“A bot that’s read every book on human psychology, every book on consumer behavior, and actually is far better at convincing a human to do something than a human is, wouldn’t know the difference between asking someone for money and manipulating someone to give,” Chappell says. If A.I. involved in raising money is misused, he says, “that’s where we could diminish trust in our sector exponentially.”

Givzey’s autonomous fundraiser isn’t yet skilled enough to have real-time conversations and, as noted, is currently sending only asynchronous texts and emails. Martel says what is being sent to users is also being reviewed by fundraisers at the participating organizations to avoid problems.

Programs like ChatGPT are known for hallucinations — instances where they provide inaccurate and sometimes offensive information to users. Martel says the autonomous fundraiser’s most common early issues involve messaging with the wrong names.

“So with a husband and wife, it’ll address a letter to one, not both,” he says. “It doesn’t take into account the preference that that couple or that household has told us.”

Martel and the institutions participating in the project are still grappling with best practices for autonomous fundraiser contact reports. Right now, all interactions with donors and their questions are recorded. Givzey isn’t sure if that will continue as the technology improves and donors are able to interact with the autonomous fundraiser in real time.

Human gift officers know that while every interaction with a donor must be captured, some details shouldn’t be, says Frezza at the College of Charleston.

“In my career, there have been things I have alluded to but not put in — very personal things you don’t want to capture,” he says. “That’s where my unknown is right now. I think we have to find a way to safeguard certain things that don’t go into the contact record. It could be either things that are personal and sensitive we don’t want added. It also could be things that are salacious that somebody was just putting out there.”

Martel says the 18-month R&D contracts will be a time to help answer questions like this and others the company hasn’t even considered. The contract interval also allows time to see if other concerns emerge, and help work out kinks, so that autonomous fundraisers could be a valuable tool to any organization raising money.

“I believe that autonomous fundraising will change the world,” Martel says. “If we can get this right, there will be more food for families, there will be more shelter for the underserved and underprivileged. There will be more outcomes because more donors will be engaged, support the causes. And that’s ultimately my vision here — to drive the cost down so far that everybody, every organization, can benefit from autonomous fundraising.”