HOW TO RANK
IN CHATGPT.
"Ranking" in ChatGPT is not the same game as ranking in Google. There are no ten blue links to climb. Either the model names you when a buyer asks what to choose, or it names someone else. This guide covers how ChatGPT actually builds an answer, and what moves whether you are in it.
How do I rank in ChatGPT?
To rank in ChatGPT you have to be the answer it can assemble, not the page it can crawl. ChatGPT builds responses from its training data plus live web results it browses, and it leans heavily on third-party sources like reviews, "best of" lists, and comparison pages. So get named in the sources it reads, make your own site machine-readable about who you are and what you do, and shape your pages to answer the exact questions buyers ask. Then measure by asking the models directly and watching which pages they cite.
How ChatGPT sources an answer
ChatGPT answers in one of two ways, and the difference decides how you get in.
By default it answers from training data: patterns learned from a huge slice of the web at a point in time. No live lookup. Here you win by being mentioned often enough, in enough trusted places, that the model has "learned" your brand as part of the category.
When the question needs fresh or specific information, it browses the live web and footnotes the pages it used. Browsing relies on a search index and OpenAI's own crawlers, so a page that is not indexed and reachable simply cannot be cited.
There is a useful distinction here. A mention is when the model names you from memory with no link. A citation is a clickable source it pulled in while browsing. You want both, and they are earned in different ways: mentions come from broad presence over time, citations come from a single page being the cleanest available answer in the moment.
Get into the third-party sources ChatGPT trusts
This is the part most people miss. The majority of the sources behind commercial AI answers are not the brand's own website. They are independent pages that discuss the category: reviews, lists, comparisons, and authoritative editorial. ChatGPT trusts a third party describing you far more than it trusts you describing yourself. So the highest-leverage work is often off your own site.
- + "Best of" and listicle pages. Recently updated "best X" lists are among the most cited page types in AI answers. Earn a place on the credible ones in your category, and pitch publishers who maintain them.
- + Reviews and ratings. Volume and recency of genuine reviews on the platforms that matter for your vertical feed the model's sense of who is real and who is liked. For hotels and ecommerce this is decisive.
- + Comparison content. Pages that put options side by side are gold for an answer engine, because they hand the model a ready-made structured comparison. Being present and well described in those comparisons matters.
- + Authoritative editorial and awards. A mention in a respected publication, or an award or accreditation, is a strong trust signal the model can lean on when deciding who to name.
Make your site easy for a machine to understand and quote
Off-site presence gets you considered. On-site clarity makes you quotable when the model browses. The aim is simple: be unmistakable about who you are, and write answers a model can lift cleanly.
If your pages are not indexed and your robots file blocks AI crawlers, you cannot be cited at all. This is the floor. Confirm AI user agents are allowed and your key pages are reachable.
Be consistent about your brand name, what you sell, and where you operate across your site and everywhere else. Mixed signals make the model unsure who you are, and unsure brands do not get named.
Structured data such as Organization, Product, and FAQ schema spells out facts in a format machines read without guessing. It will not invent authority, but it removes ambiguity from the facts you do have.
Use questions as headings and put a direct, self-contained answer in the first lines beneath each one. That is the passage a model can extract whole. The same structure that wins featured snippets now wins AI citations.
One format earns its own line: comparison pages on your own site. An honest page that lays out the real options for a buyer, including where you fit and where you do not, gives the model exactly the structured, decision-ready content it wants to cite. Keep these current. Freshness is a real signal; a page revisited recently tends to be favoured over one left untouched for years.
How to measure whether it is working
Keyword rank does not map to AI answers. The metric is whether you get named, and whether your pages get cited. You can check this yourself.
- + Ask the models the buyer's questions. Run the actual prompts a customer would type, such as "best X for Y," across ChatGPT, Perplexity, and Gemini. Note who gets named and in what order.
- + Read the citations. In browsing mode the answer links its sources. Those footnotes tell you exactly which pages taught the recommendation, so you know which lists, reviews, and comparisons to target next.
- + Track share of voice over time. Re-run the same set of prompts on a schedule and watch your mentions and citations move. That trend, not a single ranking, is the scoreboard.
GEO is not a replacement for SEO. It is the next layer.
Optimising for AI answers is called GEO, or Generative Engine Optimization, with AEO (Answer Engine Optimization) as the on-site half of it. It runs on the same fundamentals SEO always has: clear entities, sound structure, and real authority earned off-site. The difference is the goal. SEO earns the click on a results page. GEO earns the mention inside the answer, before any results page appears. You still want both, because the page that ranks in search is often the same page a model reaches for when it browses. The brands that get named tomorrow are the ones structuring for it while their competitors are still arguing about whether it matters.
Want this done on your brand instead of done yourself? That is the work behind my GEO and AEO services: getting you into the sources AI reads, then structuring your site to be the answer it cites.
Send me your URL. I'll show you where you're invisible.
A short video audit: the prompts your buyers use, what ChatGPT and Perplexity answer, and exactly where a competitor gets named instead of you. If it's useful, we talk.
Rather just talk it through?
15 minutes. No pitch. We find where AI is naming a competitor and what it takes to flip it.