Somebody is researching your company right now and they are not using Google. They typed your category into ChatGPT, or asked Perplexity who the credible providers are, and they are reading a paragraph of prose with five or six little citation links underneath it.

If your business is not among those links, you were not shortlisted. There is no second page to be on. The answer either mentions you or it does not.

This guide covers how to rank in ChatGPT, what actually determines whether an AI search engine cites you, what works on each of the major platforms, and how to measure it. Most of it is not what people expect, because the tactics that move AI citations have very little in common with the tactics that move blue links.

How To Rank In ChatGPT: The Short Answer

To rank in ChatGPT and get cited by AI search engines, you need to be mentioned, described specifically and consistently, on the third-party pages the models already retrieve for your category: roundups, comparison lists, directories, interviews and profiles on domains that rank. Your own website matters for branded questions but is rarely cited for “who is the best X” questions, so the work is placement and PR rather than on-page optimisation. Progress shows up first on Perplexity and Google AI Mode, and last on ChatGPT.

How AI Search Engines Choose What To Cite

Start with the mechanism, because almost every mistake in this area comes from misunderstanding it.

When someone asks ChatGPT or Perplexity a question about a category, the model runs a search, pulls a handful of pages, and writes an answer grounded in what those pages say. The citations underneath are the pages it used. That means two separate things have to happen for you to appear. A page that mentions you has to be retrieved, and the sentence that mentions you has to be useful enough to survive into the summary. The discipline of shaping that outcome has a name, and our complete guide to generative engine optimization covers the theory in more depth.

The critical detail is which pages get retrieved. For a question like “who are the best X companies”, the retrieved pages are almost never a vendor’s own website. They are third-party pages: roundups, comparisons, industry lists, directories, interviews and profiles. Models are built to avoid grounding a recommendation in the marketing copy of the company being recommended, for the obvious reason that everybody claims to be the best.

This is why so much AI visibility work fails. Companies rewrite their homepage, add FAQ schema, restructure their service pages, and see nothing change. They optimised the one source the model was never going to cite for that question.

Why Your Own Website Is Not Enough

Your site still matters for branded questions. If someone asks an assistant “what does Acme Corp do”, your own pages are the natural source and clear, well-structured content helps.

But branded questions are not where new customers come from. The valuable queries are unbranded, the ones where someone is choosing rather than checking, and those answers are assembled almost entirely from third-party pages. You cannot write your way into them from your own domain.

What you can do is get mentioned on the pages the models already trust. That is the whole game, and it is a public relations problem wearing an SEO costume.

How To Rank In ChatGPT: What Actually Works

Get listed on the roundups that already rank

Search your category the way a customer would, in plain language, and note which third-party pages come back. Those are the pages that get retrieved when an assistant answers the same question. Being present on them is the single highest-leverage action available, and it is a placement problem rather than a writing problem.

Prioritise by domain authority, because retrieval favours pages that already rank. A mention on a strong domain that holds page one is worth more than five mentions on pages nobody surfaces.

Be described in a sentence a model can lift

Models reproduce specific, self-contained claims. “Acme builds HIPAA-compliant patient intake software for mid-sized clinics” survives into an answer. “Acme is a leading innovator in the healthcare space” does not, because it says nothing that distinguishes you and nothing that could be checked.

When you have influence over how you are described on a third-party page, spend it on specificity: what you do, for whom, and one concrete qualifier. That single sentence does more work than a page of adjectives.

Publish interviews and executive profiles on domains that get cited

Interviews and profiles are unusually effective here for a structural reason. They are third-party pages, they are attributed to a named person, and they contain direct quotes, which is the format summarisers favour when they need something to attribute.

A profile of a founder on an authoritative domain gives a model something it can cite for both a person query and a company query. Two entry points from one page.

Get the same facts repeated across several sources

Consistency across independent pages raises confidence. If four separate domains describe you the same way, that description becomes the one the model reproduces. If every page describes you differently, none of them wins and the model reaches for a competitor with a clearer story.

Decide on your one-sentence description before you start placing anything, then keep it identical everywhere.

Answer the question directly, in your own content

For informational queries where your own site can be cited, structure matters. Lead with a direct answer in the first two sentences, then explain. Use question-shaped headings that match how people ask. Keep the answer near the heading rather than three paragraphs below it.

This is ordinary good writing, and it is also what makes a passage easy to extract.

Keep dates and numbers current

AI search engines lean heavily on freshness for anything comparative, and a page dated two years ago is far less likely to be retrieved for a “best of” question. Numbers with dates attached are strong extraction targets, so state them plainly and update them.

Do not bother trying to trick the model

Hidden text, prompt injection in page content and keyword stuffing aimed at retrievers do not work, are actively filtered, and put your domain at risk. There is no shortcut layer here. The systems reward being genuinely present in the places that discuss your category.

How To Appear In Google AI Overviews

AI Overviews behave differently from ChatGPT because they are grounded in Google’s existing index. In practice this means conventional ranking still matters. Pages cited in an Overview are usually already ranking in the top ten for a related query, so classic SEO is the entry ticket rather than a separate discipline.

Two things help beyond ranking. Concise, direct answers placed high on the page get pulled more often than the same information buried mid-article. And breadth of coverage matters, because Overviews frequently cite several different domains that each answer part of the question rather than one page that answers all of it.

How To Get Cited By Perplexity

Perplexity is the most transparent of the major engines about its sources, which makes it the easiest to learn from. Ask it a question in your category and read the citation list. That list is a to-do list.

It tends to favour pages that read as comparative or evaluative, which is why roundups and versus-style pages appear so often, and it cites more sources per answer than most competitors. That makes it the most achievable early win. If you are going to test whether placement work moves AI citations, test on Perplexity first, because it will show movement soonest.

How To Measure Whether Any Of This Is Working

Most companies doing AI visibility work cannot tell you whether it worked, which is the main reason the discipline still gets treated as speculative. It is measurable, and there is now a whole category of AI visibility tracking tools built to do it.

Track two things, separately, because they answer different questions. Citations count how often your pages are used as a source in AI answers. Brand mentions count how often you are named in an answer without necessarily being linked. A brand can be widely mentioned but rarely cited, or the reverse, and the fix differs in each case.

For a concrete example of what these figures look like, here is our own citation profile at the time of writing (September 2026):

Platform	Citations of reverbico.com
Google AI Mode	121
Perplexity	79
Google AI Overviews	35
Gemini	31
Copilot	16
ChatGPT	10
All platforms	378, across 108 distinct pages

Two things in that table are worth noting for your own planning. Google AI Mode produces the most citations by a wide margin, and it is the one most companies are not tracking. And ChatGPT, the platform everyone asks about first, is the hardest to earn citations on and the slowest to move. Judge early progress on Perplexity and AI Mode rather than on ChatGPT, or you will conclude nothing is working while it is.

A Realistic Order Of Operations

Fix your one-sentence description first, because everything downstream repeats it and inconsistency is expensive to unwind later.

Then audit where you are mentioned today. Search your category as a customer would, list the third-party pages that come back, and mark which ones include you. The gaps are your target list.

Then get placed on the strongest of those gaps, prioritising domain authority over volume. A handful of placements on pages that already rank will move your citation count more than dozens on pages that do not.

Then measure, monthly, tracking citations and brand mentions as separate numbers. Expect Perplexity to move first, AI Mode second, and ChatGPT last.

Frequently Asked Questions

How do you get cited by ChatGPT specifically?

ChatGPT cites pages it retrieves through web search when it decides a question needs current information, and for category questions those pages are overwhelmingly third-party lists, comparisons and profiles. Getting cited by ChatGPT means being present, and specifically described, on the pages that already rank for your category query. It is the slowest of the major engines to reflect new placements, so give it months, not weeks.

Is ChatGPT SEO different from normal SEO?

Partly. Ranking in Google still helps, because AI search engines retrieve from pages that already rank, and Google AI Overviews are drawn almost entirely from the top ten. But ChatGPT SEO adds a layer traditional SEO ignores: your presence on other people’s pages. On-page work moves your own rankings; placements and consistent descriptions move your AI citations.

How long does it take to rank in ChatGPT search?

Perplexity typically reflects a new placement within weeks, Google AI Mode and AI Overviews within one to two months once the page is indexed and ranking, and ChatGPT last, often a quarter or more. Track all of them monthly and read the trend across platforms rather than judging on ChatGPT alone.

Does schema markup help you appear in AI overviews?

Only marginally. Schema helps Google understand a page, but pages appear in AI Overviews because they rank and answer the question directly near the top. Adding FAQ schema to a page that does not rank changes nothing. Ranking, a direct answer placed high on the page, and freshness do the work.

Can you pay to rank in ChatGPT?

Not directly. There is no ad product that places you in a ChatGPT answer. What you can pay for is presence on the third-party pages the models cite, through sponsored listings, interviews and profiles on authoritative domains, which is the same mechanism that earns editorial mentions. Whether paid or earned, the page has to rank and the description has to be specific for it to count.

Conclusion

Ranking in ChatGPT is not really an optimisation problem. It is a presence problem. The models answer category questions by reading what other people have published about your category, so the work is getting yourself credibly into those pages, described in a sentence worth quoting.

That is slower than a technical fix and considerably more durable, because a mention on a page that ranks keeps paying out every time somebody asks the question.

If you want help with the placement side of it, our Leader Spotlight interviews and Visionary Profiles exist precisely to put named people and companies onto pages that AI search engines already cite. And if your problem is the opposite one, an answer that mentions you but says the wrong thing, our list of ChatGPT reputation management companies covers the firms that work on that.