To get cited by AI assistants like ChatGPT, Claude, Gemini, and Perplexity, give them content they can quote: direct answers near the top of each page, FAQ sections, schema markup, and consistent facts about your nonprofit everywhere they appear online. AI tools cite sources that are clear, specific, and easy to verify, and most of the work involved is plain writing and housekeeping rather than anything technical.
If someone asks ChatGPT which organizations are doing good work in your cause area, does your nonprofit come up? More donors, volunteers, and grant makers are starting their research inside AI assistants instead of a search bar. Those tools answer with a handful of names and links, and if yours isn’t one of them, the conversation moves on without you. The encouraging part is that the organizations getting cited are not necessarily the largest ones; more often they are the ones whose websites explain most clearly who they are and what they do.
Why AI assistants are a new front door
For years, being found meant ranking on Google. That still matters (our nonprofit SEO guide covers those fundamentals), but a second front door has opened. ChatGPT, Claude, Gemini, Perplexity, and Google’s own AI Overviews now compress the whole research process into a single conversational answer. Someone asks about food insecurity programs in their county and gets a short, confident summary that names two or three organizations.
A conversational answer works differently from a results page, because there is nothing beyond the answer itself to scroll to. If the assistant doesn’t know your organization, or doesn’t trust what it knows, you are simply left out. The work of earning a spot in those answers is called generative engine optimization, or GEO. Despite the name, most of it is ordinary writing and housekeeping, and almost everything below is something your team can start this quarter.
How do AI assistants pick which sources to cite?
AI assistants build answers in two ways. Some responses draw on what the model learned during training: a snapshot of the public web, including whatever it absorbed about your organization. Others are backed by live retrieval: the assistant runs a search behind the scenes, reads the top results, and cites the pages it used. Both paths reward the same qualities.
- Direct answers. Pages that resolve a question in the first few sentences are easier to quote than pages that wind up to the point.
- Clean structure. Clear headings, FAQ sections, lists, and structured data help machines parse who you are and what you do.
- Corroboration. When your founding year, mission, and location match across your site, Candid, Charity Navigator, and press coverage, confidence goes up.
- A recognizable entity. Assistants cite organizations they can identify without ambiguity: a distinct name, a clear mission, a complete About page.
Here’s how those signals translate into work you can actually put on a calendar:
| What assistants reward | What to do about it |
|---|---|
| Direct answers | Open every key page with a two- to three-sentence answer to the question the page exists for |
| Clean structure | Add question-style headings, FAQ sections, and schema markup |
| Corroborated facts | Audit your name, mission, EIN, and location everywhere they appear online |
| Entity clarity | Build out your About page and organization schema so machines know exactly who you are |
| Crawler access | Check that robots.txt isn’t blocking AI crawlers, and consider adding llms.txt |
Write content that answers the question first
Look at your program pages, your donation page, your most-read blog posts. Do they answer the visitor’s question in the first paragraph, or do they build up to it? AI assistants pull quotable passages, and a buried answer is much harder to lift than a direct one.
- Start each page with a short, direct answer to the question the page exists to address, then go deep.
- Use question-style headings. Assistants match them against the questions real people actually ask.
- Keep one idea per section, so a passage still makes sense when quoted out of context.
- Add an FAQ to program and service pages covering the questions your staff answer by email every week.
This also happens to be good writing for humans, which is the pattern with most of GEO: the same clarity that helps a machine helps a hurried donor. If producing this kind of content at scale is a bottleneck for your team, our AI Visibility & GEO services handle that production work.
Add schema markup so machines don’t have to guess
Schema markup is a layer of code that labels your content in a vocabulary machines share: this is an organization, this is its mission, these are FAQ questions and their answers. It removes guesswork. For nonprofits, the highest-value types are organization schema for who you are and FAQPage schema for the questions you answer.
We won’t repeat the implementation details here, because we’ve already written a full walkthrough: our guide to schema markup for nonprofits covers which types are worth adding and how to test them. For this article, the point is simpler: structured data is one of the few direct channels you have for describing your organization to AI systems in their own language.
Keep your facts consistent across the web
Assistants cross-check. If your website says you were founded in one year, your Candid profile says another, and an old press release says a third, the model’s confidence in all three drops, and it becomes more likely to cite an organization whose facts it can verify. Entity clarity means every description of your organization tells the same story. Audit these in particular:
- Your About page, the anchor document for who you are (here’s what a strong About page includes)
- Your profiles on Candid, Charity Navigator, and other nonprofit directories
- Your Google Business Profile and social media bios
- Wikipedia, if your organization has a page
- The boilerplate at the bottom of your press releases and annual reports
Pick one canonical description (name, mission, founding year, location, EIN) and make everything else match it. A clear, memorable mission statement does double duty here; if yours has drifted, our roundup of effective nonprofit mission statements is a useful calibration exercise.
Should your nonprofit add an llms.txt file?
llms.txt is an emerging convention: a plain-text file at the root of your site (yoursite.org/llms.txt) that gives AI systems a curated, readable map of your most important pages: what you do, where your programs live, how to donate. The idea is to offer a straightforward orientation to your site so that crawlers don’t have to work it out from your navigation.
One caveat: the standard is young and support is uneven, so not every AI platform reads the file yet. But it takes about an hour to create, carries no downside, and makes your most important content easy to find. While you’re at it, check your robots.txt file. Some organizations blocked AI crawlers wholesale during earlier debates about scraping and are now invisible to the very tools their donors use. Blocking AI crawlers is a legitimate position, but it should be a deliberate decision rather than a setting left over from that period.
How to test whether AI assistants cite you
Testing this doesn’t require any special tools. Open ChatGPT, Claude, Gemini, and Perplexity, and run the questions your supporters would plausibly ask.
- “What does [your organization] do?”
- “Is [your organization] a legitimate charity?”
- “What are the best nonprofits working on [your cause] in [your city]?”
- “How can I volunteer with organizations focused on [your cause] near me?”
Score what comes back. Is the description accurate? Which peer organizations appear when you don’t? Which sources does the assistant cite — your site, a third-party profile, an old news article? Repeat the exercise monthly, because answers shift as models update and your fixes get recrawled. Pay particular attention to the wrong answers, since each error points at a specific page to fix or profile to update.
For the underlying mechanics of how assistants read your site, see AI discoverability for nonprofit websites.
Getting cited by AI: FAQs
What is generative engine optimization (GEO)?
GEO is the practice of making your website easy for AI assistants to understand, trust, and cite. It builds on traditional SEO but focuses on direct answers, structured data, and consistent facts rather than rankings alone.
Does SEO still matter if people are using AI assistants?
Yes. Most AI assistants pull from search indexes when they answer, so a site that ranks well is more likely to be retrieved and cited. GEO works with SEO, not instead of it.
How long does it take to get cited by AI assistants?
There’s no fixed timeline. Assistants that search the live web can pick up improvements once your pages are recrawled, while changes to a model’s trained knowledge take longer. Test monthly and expect gradual movement.
What is an llms.txt file?
It’s a plain-text file at the root of your site that gives AI systems a short, curated guide to your most important pages. Support varies by platform, but it’s quick to add and carries little risk.
How do I find out what ChatGPT says about my nonprofit?
The most direct way is to ask. Open ChatGPT, Claude, Gemini, and Perplexity and ask what your organization does, whether it’s trustworthy, and who does similar work in your area. Anything that comes back wrong or missing gives you a list of pages and profiles to update.
AI search is moving quickly, and most nonprofits haven’t started this work yet, which means the organizations that begin now are the ones assistants will learn to cite. If you’d like a partner for any of it, our SEO and GEO services cover everything in this article, from schema markup to entity cleanup. Elevation has worked exclusively with nonprofits since 2007, and we’re glad to take a look at where you stand. Tell us about your project, and we’ll go from there.
