Trussell: When AI becomes part of the fundraising journey

Anti-poverty charity, Trussell recently received what it says was the UK's first donation made entirely through an AI agent.

Developed by GoCardless as part of the Financial Conduct Authority's AI live testing programme, the technology allowed someone to ask about Trussell's work, decide how much they wanted to give and set up a Direct Debit, all through a conversation with an AI assistant.

It's a pretty big development for charity fundraising. Someone can go from learning about a cause to donating without having to visit a website or fill out a separate form.

But we're more interested in what happens before the donation.

If people start asking AI where they should give their money, charities will have another thing to think about. What does AI actually know about their work, and  is it giving potential supporters the full picture?

It needs to know more than you do

Charities have spent years answering the questions people have before they donate: Where does my money actually go? What difference will £10 make? How do I know this organisation is effective? Why should I support this charity rather than another one?

A lot of charity communications is built around answering those questions with websites, impact reports, case studies, media stories and campaigns all helping someone build a picture of an organisation before deciding whether to support it.

If AI becomes part of that journey, the questions don't really change. But who is finding and sharing the answers does.

We've talked before about generative engine optimisation (GEO), which is about making sure organisations can be found and understood by AI tools. Until now, a lot of that conversation has focused on appearing in AI-generated answers. The Trussell example makes the stakes a little higher.

If someone asks AI to recommend a homelessness charity, knowing which organisations exist is only part of the job. Ideally, it should also be able to explain what makes their approaches different, what they've achieved and what a donation would help them do.

Charities need to consider whether the information they're putting out there gives AI enough to work with.

An impact report might explain all of this in detail, but will an AI assistant pick up the most important parts? Will it understand the difference between a charity providing immediate support and one campaigning to change the systems causing the problem?

There's a lot more to a charity than a short description of what it does, and that shouldn't get lost  because someone is asking AI instead of visiting its website.

Is AI just another channel?

Communications teams already have enough platforms to worry about.

There's the website, Google, Instagram, TikTok, subscriber newsletters, podcasts, media coverage and whatever else audiences happen to be engaging with.  AI assistants are now  joining that list.

It sounds exhausting, but that doesn't necessarily mean charities need to start creating a whole new set of content just for AI.

We already think about what people need from us across different channels. A journalist wants evidence, expertise and a story. Someone on Instagram needs a reason to stop scrolling. Someone arriving through Google probably has a question they want answered.

AI is another way people might find that information, but charities have much less control over what comes back.

You can spend months developing a campaign or carefully explaining your work on your website, only for an AI assistant to summarise your organisation in a couple of sentences. 

So charities need to think about whether the information they're putting out there gives AI enough to work with and is presented in a suitable format . Is their impact reporting easy to find and understand? Does it answer questions users might be searching for,  that AI will pick out and reshare? Are they getting their expertise and perspectives into the wider conversation, rather than relying solely on their own channels?

That makes PR, thought leadership, impact reporting and website content feel less like separate parts of the comms mix and more like a footprint for AI to work with.

Being recommended isn't the same as being represented

Imagine asking an AI assistant where to donate £20 a month to help tackle food poverty.

It gives you three charities to choose from, each with a short description of what they do. It's a good starting point if you're unsure which charities to support, but is it enough to make a decision on who gets your money?

One charity might provide emergency food parcels, while another works with communities to tackle the reasons people are struggling to afford food in the first place. Both are doing important work, but they're approaching the problem in different ways.

If AI reduces that information to a couple of sentences, will someone really understand the difference?

There's also the question of which charities get mentioned. Larger organisations tend to have more information available online, while smaller community charities might not have the same resources to publish research, secure media coverage or maintain a detailed website.

For charities, this is why GEO needs to be about more than getting your name into an AI-generated answer.

It's about making sure there's enough information available for AI to explain what your organisation does, how it approaches the problem and what someone's donation would support.

Trussell has shown how much easier the process of donating could become with AI. But making a donation easier shouldn't mean giving people less information about where their money is going.

Charities have spent years encouraging donors to look beyond the donation itself and understand the work they're supporting. It would therefore be a shame if making giving more convenient meant losing some of that understanding along the way.