An AI visibility audit checks whether AI assistants (ChatGPT, Claude, Perplexity, and Google’s AI Overviews) mention, accurately describe, and cite your nonprofit when people ask the questions your supporters ask. It takes about an hour, costs nothing to run yourself, and tells you exactly where you stand in the fastest-growing discovery channel.
More supporters now start with an AI assistant instead of a search box: “what are the best food-security nonprofits near me?”, “is this organization legitimate?”, “how do I donate?” If the answer doesn’t include you, or describes you wrong, you never learn about the person who asked. Here’s how to find out, step by step.
Step 1: list the questions your supporters actually ask
Don’t audit with your own vocabulary — audit with theirs. Write down 10–15 real questions across four intents:
| Intent | Example question | What you’re testing |
|---|---|---|
| Discovery | “Best [your cause] nonprofits in [your city]” | Do you appear at all? |
| Brand | “What does [your organization] do?” | Is the description accurate and current? |
| Action | “How do I donate to / volunteer with [your organization]?” | Does it route people correctly? |
| Trust | “Is [your organization] legitimate? Who runs it?” | Do trust signals surface? |
Step 2: ask the major AI assistants
Run every question through ChatGPT, Claude, Perplexity, and a Google search that triggers an AI Overview. Use fresh sessions so your history doesn’t color the answers, and save each response: you’ll want the before-and-after when you fix things.
A worked example
Say you run a food bank serving three counties. You ask each assistant: “What are the best food-security nonprofits near Riverside?” Perplexity names two national organizations and one local competitor, with citations — you’re absent. ChatGPT describes your organization when asked directly, but calls you a “soup kitchen” and names a program you retired two years ago. Google’s AI Overview pulls a sentence from a third-party directory with your old address.
That single session tells you three different things: a discovery gap (you’re not in the consideration set), an accuracy problem (stale facts in the training data and on your own site), and a source problem (a directory you don’t control is outranking your own pages as the machine’s source of truth). Each has a different fix, which is exactly why you audit before you optimize.
What a good answer looks like
You’re not aiming for flattery; you’re aiming for an answer a stranger could act on. A strong AI response about your organization typically has four properties:
- You’re named in discovery answers for your cause and service area: in the set, not necessarily first.
- The description is current: right mission, right programs, right geography, right leadership.
- Your website is the cited source, not a directory, not a news story from 2019.
- The action path is correct: donate, volunteer, and contact answers route to the right pages.
Screenshot the best answer you find and treat it as the benchmark. The audit’s job is to make every other assistant’s answer look like that one.
Step 3: score what you find
- Mentioned? Are you in the answer at all? And for discovery questions, are peers named ahead of you?
- Accurate? Right mission, programs, service area, and leadership? Or stale facts from an old site?
- Cited? Does the assistant link or attribute you as a source, or describe you second-hand?
- Actionable? When someone asks how to give or volunteer, does the answer route them to the right place?
Score each question simply: present/absent, accurate/wrong, cited/uncited. The pattern across 15 questions is your visibility profile.
Keep score like you mean it
A spreadsheet turns a one-off experiment into a program. One row per question per assistant, re-run quarterly:
| Question | Assistant | Mentioned? | Accurate? | Cited? | Notes |
|---|---|---|---|---|---|
| Best [cause] nonprofits near [city] | Perplexity | No | N/A | N/A | Two national orgs + one local peer shown |
| What does [org] do? | ChatGPT | Yes | Partly | No | Names a retired program |
| How do I donate to [org]? | AI Overview | Yes | Yes | Directory | Cites third-party page, not ours |
The quarter-over-quarter movement is the point: after you publish answer-first pages and schema, you should see “mentioned” flip on discovery questions first, then citations shift from third-party sources to your own site.
Why nonprofits fail the audit
Four causes explain most gaps. Your pages bury answers instead of leading with them. Your basic facts (founding year, service area, program names) differ between your website, your profiles, and third-party directories, so machines can’t tell which version to trust. Your site has no schema markup, leaving AI to guess at structure. And you simply haven’t covered the questions people ask. A page that answers “what does your organization do?” plainly is astonishingly rare.
How to fix it: the GEO playbook
The fix is Generative Engine Optimization (GEO, and how it relates to SEO and AEO), and it’s concrete: answer-first pages, FAQ sections with matching schema, consistent facts everywhere you appear, and clear entity signals about who you are. Our complete guide to getting your nonprofit cited by ChatGPT walks through each fix; the broader search fundamentals live in our nonprofit SEO guide.
What you can control (and what you can’t)
Be clear-eyed about the mechanics. Perplexity and Google’s AI Overviews retrieve live web results, so your fixes can show up there in weeks. ChatGPT and Claude blend training data with live browsing, so stale descriptions can linger even after your site is perfect, which is why consistency across every place your organization appears matters as much as your own pages. And every assistant varies its answers run to run; that’s normal. Audit for patterns across questions, not any single response.
You also can’t force a citation — no one can, and vendors who promise guaranteed AI placement are selling something they don’t control. What you control is being the clearest, most consistent, best-structured source about your own work. That’s the whole strategy, and it’s enough.
Who should own this
This is a communications-team habit, not an engineering project. The audit itself takes about an hour a quarter and needs no tools beyond the assistants and a spreadsheet. The fixes split naturally: content owners handle answer-first rewrites and FAQs; whoever manages the website adds schema markup; leadership signs off on the canonical fact sheet: one agreed version of your founding year, service area, program names, and impact language that everything else quotes.
Make it a habit, not a one-off
AI systems refresh their sources continuously, so visibility drifts the same way rankings do. Re-run the audit quarterly, after every major site change, and any time your programs or leadership change. Track the same questions each time so movement is measurable.
Once you’ve scored your visibility, the fixes live in our complete nonprofit GEO guide.
AI visibility audit FAQs
What is an AI visibility audit?
A structured check of how AI assistants (ChatGPT, Claude, Perplexity, and Google’s AI Overviews) describe and recommend your nonprofit when people ask the questions your supporters actually ask. You score whether you’re mentioned, described accurately, and cited, then fix the gaps.
Which AI tools should we test?
Start with ChatGPT, Claude, Perplexity, and Google’s AI Overviews. They pull from different sources and cite differently, so an organization can be visible in one and invisible in another.
Why doesn’t ChatGPT mention my nonprofit?
Usually because your content isn’t structured for machines to read and cite: no clear answer-first pages, inconsistent facts across the web, missing schema markup, or thin coverage of the questions people ask. Occasionally it’s simply that stronger, better-structured sources outrank you.
How do we improve our AI visibility?
The discipline is called GEO (Generative Engine Optimization): answer-first content, FAQ sections, schema markup, consistent facts everywhere your organization appears, and clear entity signals. Our guide to getting cited by AI walks through the full playbook.
How often should we re-run the audit?
Quarterly is a sensible rhythm: AI systems update their sources continuously, and your own content changes too. Re-run it after any major website change or rebrand.
Want the audit done for you and the gaps fixed? That’s exactly what our AI Visibility & GEO service does: we run the audit, structure your content so assistants can read and cite it, and keep your facts consistent everywhere. Tell us about your organization.
