What AI search means for charity websites and what to do about it
AI search is answering the questions that used to bring supporters to charity websites. What's changing, and the practical steps worth taking now.
SparkToro is a marketing-analytics firm that studies how people use search. Its research now puts the share of Google searches that end without a single click at roughly 68%, which is a sharp rise on previous years.
The answer appears on the results page, but the user never visits a website. Add AI Overviews, Perplexity, and ChatGPT's browse mode, and the trend sharpens. People get answers without going anywhere.
Publishers often call this "Google Zero", a world where Google sends them little to no traffic.
For charities, this is a slow erosion of the moments that turn a curious person into a donor.
In this post, I will look at what it means for organisations that depend on their websites for supporters and fundraising income, and what we can do about it now.
The problem with informational traffic
Charity content used to follow a reliable formula. You'd write about the issues, they'd rank on Google, and some of the people who find you go on to follow your charity or even donate.
AI search breaks that because now, if someone types "how do I help a friend with depression?" they might get a synthesised answer pulled from several sources. Your guide might have actually fed that answer, but your website never entered the picture.
You provided the knowledge, but were cut out of the conversation.
The bigger and more generic the search query, the worse this problem gets. Searches like "charities that help refugees" or "how does homelessness happen?" are now regularly answered by AI without a click required. These were once reliable top-of-funnel entry points for many organisations. For a long time, charities could reliably trust that optimising for these "long tail phrases" would yield a fruitful source of traffic.
Smaller charities feel it most acutely. A household name like the British Red Cross will get searched by name regardless. Brand recognition acts as an inoculation against AI searches. But a smaller organisation whose web traffic depends on ranking for cause-related terms has far less protection.
If an AI summary answers the question, there's nothing left to click.
The donation journey is getting interrupted
Beyond informational traffic, there's a more direct concern: AI can start to influence which charities people actually give to.
When someone searches "best charity for children in poverty" or "where should I donate after the floods?", they're in an active giving mindset. Historically, charities that ranked well for those terms had a real fundraising advantage. Now AI tools are generating curated shortlists in response to those exact queries and the criteria for inclusion aren't fully transparent.
What we can see is that AI systems seem to favour organisations with a strong digital footprint. That means lots of credible coverage, well-structured websites, and clear mission statements. It also means a presence in trusted sources like the Charity Commission, Wikipedia, and the main news outlets. An organisation that does extraordinary work but has a thin online presence is at a disadvantage, but that has nothing to do with its impact.
There's a misinformation risk too. AI tools make errors. They confuse similar-sounding charities. They repeat out-of-date figures. They describe a charity's work from old press releases rather than what it does now. If your web presence is sparse or inconsistent, you have little ability to correct what AI says about you, and potential donors may never know the information was wrong.
What actually helps
The encouraging thing is that what AI search rewards is largely the same thing that makes a charity website useful to the people it's for. A few areas are worth prioritising.
Original content that can only come from you
AI systems are trained on what already exists on the internet. They can synthesise, summarise, and rephrase but they can't generate content that doesn't exist yet (or, if they can, they can't do it accurately).
This could be your case studies, your data, your field reports from frontline workers, your survey of the people you support. All of that is something AI can't invent.
AI search leans more and more towards first-hand content. Google's own quality guidelines make this explicit: they want content that shows real experience with a topic, not just knowledge of it.
Original numbers appear to get cited more than almost anything else in AI answers. If an AI quotes your figure and attributes it to you, that's both a credibility signal and a reason for curious readers to find your site.
A good example is the Out Of The Shadows Index. We built the site with Together for Girls (a global partnership to end violence against children) and Economist Enterprise, who led the research. Rather than publishing a static report, we made it a deep, interactive data tool. It gives journalists, researchers, and campaigners something to dig into. A resource like that gets cited and linked to, from academic papers, news coverage, and policy documents, the places AI systems treat as trusted. The Economist Enterprise branding doesn't hurt them, either. If you can get credible, well-known partners to put their name alongside yours on something solid, that association does real work in how AI systems judge your standing.
Depth over volume
A single page titled "mental health in young people" that touches on twelve issues superficially is unlikely to be cited by AI for anything. But a well-evidenced piece on one issue eg: the link between exam pressure and self-harm in 16–18 year olds stands a far better chance to, especially if it cites research, it includes real insight from your clinical team and it's kept up to date. AI systems love these reference points.
It means going properly deep on the topics where you really have authority, rather than spreading yourself thin across everything near your cause.
Content signed off by a named expert, like your clinical director, your lead researcher, your policy head, with their credentials listed carries weight that anonymous content doesn't and this authority matters in both traditional SEO and AI optimisation. Put real names on important content. Include short bios (check the bottom of this very post!). Make it clear who is speaking and why they're qualified.
Being a source, not just a website
AI systems judge credibility partly by how many reputable sources reference you. For example, being quoted in The Guardian, having your research cited by a university, appearing in a House of Commons briefing paper.
This puts PR back on the agenda in a serious way. It means making your spokespeople available to journalists, publishing findings worth reporting, and building relationships with the sector publications and policy bodies that AIs see as trustworthy sources.
It's also worth making sure the basics are solid.
- Your Charity Commission listing should be current and complete.
- Your Wikipedia entry should be accurate and well-sourced.
- Your Google Business Profile should be claimed and updated.
These feel minor but they're often where AI systems look first for basic organisational facts. At Mutual we often talk about how one of the agendas for these companies is getting good data quickly and cheaply, it seems intuitive to us that they'd much prefer to source it from places like those sources than hunt around the entire web for nuggets, if they can help it.
Schema markup and structured data
Schema markup is a piece of code you add to your website that tells search engines and AI systems what things mean, not just what they say.
- An Organisation schema tells Google your registered charity number, your official name, your logo, and your social profiles.
- A DonateAction schema signals that you accept donations and links directly to your donation page.
- An FAQPage schema means your Q&A content can be extracted and displayed directly in AI-generated responses.
These are pretty easy to put into your site and can improve an engine's understanding of your organisation and content with minimal effort.
The FAQ question is worth treating seriously
Conversational AI search is built around questions. When someone asks Perplexity "how does [your cause] affect people in the UK?", the system is looking for content that directly and clearly answers that phrasing. Some charities have well-structured FAQ sections that cover the real questions their audiences ask, answered plainly and with some depth. They tend to do better in AI responses than charities that bury the same information in long pages of prose.
This isn't about stuffing questions in artificially. It's about working out what people actually want to know and answering it in a format that's easy to extract.
It's worth thinking about your own site search too, not just how external AI handles you. Working with the creative agency Bond & Coyne, we rebuilt the British Skin Foundation's site with an AI-powered search engine. It uses semantic search, the kind that understands natural language rather than matching keywords literally. Someone searching for "my skin gets really dry and itchy in winter" finds relevant content, not a blank results page.
More and more visitors arrive with conversational queries, whether they're typing into Google or into your own search bar. Sites that can't handle that will start to feel broken - traditional key word search is probably on its last legs.
Don't rely on search to do all the work
The harder shift is strategic. Accept that search traffic will become less reliable. The organisations best placed to cope are the ones with direct relationships with their supporters. In the publishing world, Conde Nast forecast as if Google traffic was effectively zero.
But an email list of engaged donors and volunteers is insulated from algorithm shifts. Someone who signed up to hear from you keeps hearing from you, regardless of what Google shows in its AI Overviews. Building that list through strong content, sign-up incentives, and a real reason to stay subscribed is the most durable investment a charity's digital team can make.
We've noticed that even when traffic is down, the traffic that does reach your site these days tends to be more engaged and more connected. This means they're far warmer to convert to newsletter subscribers.
The same applies to brand awareness more broadly because when people already know your name and search for it directly, AI search barely affects that journey. Branded searches are far more stable than generic topic searches right now. Anything that raises general awareness of your organisation (offline campaigns, partnerships, earned media, a presence in the community) feeds direct and branded traffic. And that traffic isn't as vulnerable to AI interception.
What to watch
If you want to track whether this is affecting you, the numbers to watch aren't just total sessions. You can monitor your click-through rate from search results, a falling CTR alongside stable rankings is a sign that AI Overviews are catching traffic before the click. Check now and then what AI tools actually say when asked about your cause or your organisation. And watch your direct and email traffic as a share of the total. If those are growing relative to organic, you're moving in the right direction.
SEO platforms like Semrush and Ahrefs now have specific reporting for AI Overview appearances.
AI search favours organisations that are credible, useful, and clearly who they say they are. That's not a new standard, it's just being applied with more reach and speed than before.
Charities that have always prioritised substance over optimisation are better placed than they might think. The work is making sure that substance is visible: attributed, structured, evidenced, and backed by a presence beyond your own website that tells the same story.
The charities that will feel this most painfully are those that built their digital presence around volume and keyword targeting without much underneath. For them, the fix is the slower work of becoming a real authority on the issues you care about most, and making sure the systems that now decide what people see can read you clearly.
Here to help
If you're looking for help to optimise your charity website and position yourself for Google-zero, feel free to each out to me emma.browne@mutual.agency, and we can have a chat about how we might be able to help.
Sources
- SparkToro / Datos — In 2026, less than one-third of Google searches still send a click (current ~68% zero-click figure)
- SparkToro / Datos — 2024 Zero-Click Search Study (earlier ~60% figure, for reference)
- Google — Creating helpful, reliable, people-first content (E-E-A-T guidance)
- Google — Adding Experience to E-A-T
- Out of the Shadows Index
- British Skin Foundation — site search
Emma is Operations Director at Mutual, leading client services and overseeing the ongoing management and optimisation of client websites. With over 20 years of agency experience and CIM marketing qualifications, she brings both strategic grounding and practical depth to digital performance.
Her optimisation work spans SEO and generative engine optimisation (GEO) — helping clients maintain visibility not just in traditional search, but in AI-generated answers and overviews. She works closely with clients to translate business goals into measurable improvements across search performance, content, and user experience.