How to Get Cited by ChatGPT: My 0/5 GEO Audit

I build an AI-visibility product. Then I ran my own two domains through five buyer-intent queries on Perplexity, the exact questions a customer would ask, and got cited zero times out of five. My competitors got cited in every single one.

That sting is the best lesson in generative engine optimization I can give you, because it shows the gap precisely. Here is what GEO is, why those competitors win the citation, and the one structural pattern all of them share.

The short answer (what GEO is)

Generative Engine Optimization (GEO) is optimizing your content to be cited as a source inside an AI’s synthesized answer, not to rank in a list of links. When someone asks ChatGPT or Perplexity a question, the model pulls a few sources into one answer. GEO is the work of being one of those few. It overlaps with SEO but it is not the same: a page can rank #1 on Google and still never appear in an AI answer.

Why your SEO rank does not measure this (the problem)

Search engines hand the user ten links and let them choose. Answer engines choose for the user and cite a handful of sources. Those are different jobs:

• SEO rewards backlinks, rank position, click-through.
• GEO rewards extractability: how cleanly a model can lift a direct, structured answer from your page and trust it.

If you only measure rank, you are blind to the entire AI answer layer, which is exactly where I was invisible. The uncomfortable part is that the two can diverge completely. A page can sit at the top of Google for a term and contribute nothing to the AI answer for that same term, because the model is not reading the SERP; it is assembling a response from sources it judges clean, direct, and trustworthy.

What my audit actually found (the data)

Five queries on Perplexity, logged in, 2026-06-24:

Query	Did it cite me?	Who got cited instead
best AI visibility audit tool	No	turboaudit.ai, usesift.net
how to get my website cited by ChatGPT	No	llmgeokit.com, cleversearch.ai
what is GEO and who offers it	No	wikipedia.org, searchengineland.com
AI search optimization services	No	webfx.com, searchbloom.com
how to check if AI recommends my brand	No	loamly.ai, metricusapp.com

Zero for five. And note: this ran on my own account with personalization on, biased toward me, and I still did not appear. That makes the zero a stronger signal, not a softer one. If the engine that knows my search history will not cite me, a stranger’s session certainly will not.

The pattern the winners share (the mechanism)

I pulled the live HTML of the top three competitors and the structure was nearly identical. Every one of them does four things:

1. The headline IS the question. Their H1 and H2 are the buyer’s exact query (“Is AI recommending your competitors instead of you?”, “Be the answer in AI search”). The page is written to be the answer, not to bury it.
2. A direct answer in the first paragraph. No throat-clearing. The liftable answer is up top, in plain sentences a model can quote without editing.
3. Q&A structured data (FAQPage + Question + Answer schema). All three wrap their content in machine-readable Q&A. This was the single most consistent signal across the set; it hands the model a pre-formatted answer with the question already attached.
4. They name the engines and own the vocabulary. “ChatGPT, Perplexity, Gemini, Claude” appear in the title, and “AI citation”, “GEO”, “AI visibility” are used as the page’s core terms, not as throwaway phrases.

That is the citation mechanism. It is not magic and it is not paid placement. It is structure, applied deliberately, on a page built for one question.

The two channels: your own pages vs. everyone else’s

There is a second half to the data that is easy to miss. Look again at who got cited: some are the competitors’ own domains (turboaudit.ai, cleversearch.ai), and some are third parties (wikipedia.org, searchengineland.com, directory and roundup sites). Those are two distinct channels, and they need two distinct plays.

• Owned-page citation is the channel you control directly. The four-part pattern above is the entire lever. This is where you start, because you can ship it today.
• Off-site citation is the slower channel: getting named inside “best AI visibility tools” roundups, comparison listicles, and reference pages that the engines already trust. You earn these with outreach, a clean entity presence (consistent Organization data, a Wikipedia or Wikidata footprint), and being genuinely list-worthy. It compounds, but it is not a same-day fix.

Most teams obsess over the second channel (PR, links) while ignoring the first, which is backwards. Fix your own pages first; they are the cheapest citation you will ever buy.

How to apply it to your own site

• Pick one buyer question. Make it your H1, verbatim.
• Put the answer in the first two or three sentences, in plain language a model can quote.
• Add FAQPage schema with the real questions your buyers ask, and answer each in two or three sentences.
• Name the engines and use the category terms consistently through the page.
• Give the page one job. A page that tries to answer five questions answers none of them cleanly enough to be cited.
• Then measure, do not guess whether it worked.

How to check if AI cites you (do this first)

Before optimizing anything, get a baseline: ask the four engines your category’s buyer questions and record the cited domains. You can do it by hand, query by query, or run it automatically. I built a tool that tests citation across ChatGPT, Perplexity, Gemini, and Claude with real API calls and shows you exactly who is cited instead of you: run your own free citation audit on Citability. (Yes, the same tool whose own site scored 0/5 above. That is the point. Now I can measure the fix.)

The baseline matters because GEO is iterative. You ship the structured page, you re-run the same queries a week later, and you watch whether your domain enters the cited set. Without the measurement loop you are decorating pages and hoping.

FAQ

Q: How do I get cited by ChatGPT? A: Publish a page whose headline is the buyer’s exact question, put a direct answer in the first paragraph, wrap it in FAQPage schema, and name the engines and category terms. Then measure citation across engines and iterate.

Q: What is generative engine optimization (GEO)? A: Optimizing content to be cited as a source inside an AI’s synthesized answer, rather than to rank in a list of links. It overlaps with SEO but targets extractability and trust, not rank position.

Q: Is GEO different from SEO? A: Yes. SEO optimizes for ranking among links; GEO optimizes for being one of the few sources an AI pulls into its answer. A page can rank well and never be cited.

Q: How do I know if it is working? A: Measure citation directly. Test your buyer queries across the AI engines and track whether your domain appears in the cited sources over time. Baseline it with a free audit.

Q: Does FAQ schema guarantee a citation? A: No. Schema makes your answer easy to extract and attribute, which is necessary but not sufficient. You still need a direct answer, topical relevance, and enough trust signals. Schema removes friction; it does not manufacture authority.

I build tools that measure AI visibility, and my own site failed the test I sell. That is not embarrassing; it is the reason the work matters, because you cannot fix what you refuse to measure. I am rebuilding chudi.dev’s pages to the pattern above, and I will publish the next audit, win or lose, so you can watch the method work in real time.