People type "how to rank in ChatGPT" into search engines every day, and the honest answer starts with a correction: ChatGPT does not publish a list of ten blue links with a number one at the top. It writes an answer, and sometimes that answer cites or links to sources. The goal is not a rank. The goal is to be one of the sources the answer is built from, or the source it names.
By the iSuggest.ai Team ยท Updated for 2026
That is a different job from classic SEO, but it is not a mysterious one. Most of it comes down to making your pages easy to fetch, easy to understand, and easy to quote. This guide walks through eight steps in the order we would do them, with a note on what is within your control and what is not. If you want the background first, our complete guide to GEO explains the discipline this all sits inside.
First, understand what "ranking" means inside an AI answer
When ChatGPT answers a question, two broad things can happen. It can answer from what it learned during training, with no citation at all. Or it can search the web, read a handful of pages, and write an answer that references them. Only the second path produces a visible source, and it is the path you can influence in the short term, because it depends on pages that exist on the web right now.
That is why the useful question is not "where do I rank?" but "when ChatGPT goes looking for sources on my topic, will it find my page, understand it, and trust it enough to use it?" Everything below is a way of answering yes to each part of that sentence. We cover the mechanics of getting cited in more depth in how to get cited by ChatGPT, so treat this post as the step-by-step checklist version.
It is also worth being realistic about the size of the prize. Industry data shows that overall website traffic from AI search engines grew 16x between 2024 and 2026, and that ChatGPT's absolute referral traffic still grew 27% year over year, even as its share of the category slid because competitors grew faster (SEO Sherpa, PPC Land). The channel is real and growing, but it starts from a small base, so treat it as an addition to your search strategy rather than a replacement.
Step 1: Let the right crawlers in
Nothing else matters if the crawler cannot read your page. OpenAI documents separate user agents for different jobs, and the distinction matters: one is used to collect content that may be used for training, another is used for search features, and a third fetches pages when a user asks ChatGPT to open a link. Blocking the wrong one in robots.txt can quietly remove you from ChatGPT's search results while you believe you only opted out of training.
Open your robots.txt file and read it line by line. Look for blanket Disallow: / rules under User-agent: *, for security plugins that add AI blocks by default, and for rules a previous developer added years ago. Our post on robots.txt syntax mistakes that block every AI crawler at once lists the patterns that cause the most damage. If you deliberately block training crawlers, that is a legitimate choice, but make it on purpose and keep the search crawler allowed.
Also check that your server does not return errors to unfamiliar user agents. Some firewalls and bot-protection services challenge anything that is not a known browser. A page that returns a 403 to a crawler is, for the purposes of citation, a page that does not exist.
Step 2: Make sure your content is in the raw HTML
A browser runs JavaScript before it shows you a page. Many crawlers fetch the raw HTML and do not run your scripts the way your browser does. If your headline, your main copy, or your product details are inserted by JavaScript after load, the crawler may see an empty shell. This is one of the most common reasons a page looks perfect to a human and invisible to a machine, and we explain it in detail in what client-side rendering hides from AI crawlers.
The simplest test is to view the page source (not the inspected DOM) and search for a sentence from your main content. If it is not there, a crawler that does not execute scripts will not find it either. The fix is usually server-side rendering or pre-rendering for the pages that matter most, not a rewrite of your whole site.
Step 3: Put the answer first
Language models quote pages that answer a question directly. A page that spends four paragraphs warming up before it says anything useful gives a model very little to lift. Lead each important page with a clear, specific, self-contained answer in the first 100 words or so, then expand underneath.
A good test: could someone read only the opening paragraph and walk away with a correct, useful answer? If yes, you have given an AI system a passage it can quote without distortion. If no, rewrite the opening. We go deeper on this style in how to write content AI models want to quote, but the short version is: one claim per sentence, concrete nouns, and numbers framed with their source.
- Use headings that match real questions people ask, such as "How much does X cost?" rather than clever titles.
- Keep paragraphs short enough that each one carries a single idea.
- Define your terms the first time you use them, since a quoted passage loses its surrounding context.
Step 4: Make clear who you are with structured data
Models need to know what entity a page is about. Structured data in JSON-LD form is a machine-readable label that says "this is an organization, here is its name, here are its official profiles" or "this is an article, here is its author and date". It does not replace good writing, but it removes ambiguity, and ambiguity is what makes a model hesitate.
Start small. Add Organization markup with a complete sameAs list on your homepage, Article markup with author and dates on posts, and FAQPage markup where a page genuinely contains questions and answers. Our guides on structured data for AI search engines and FAQ schema show exactly what to add. Whatever you add must match what is visible on the page. Markup that describes content that is not there is worse than no markup.
Step 5: Build real mentions of your brand elsewhere
AI systems do not rely only on your own site. Industry analysis reports that brand mentions correlate with AI search visibility roughly three times more strongly than backlinks do (a correlation of 0.664 versus 0.218, per Media Officers). Correlation is not causation, and you should read that number as a direction rather than a promise, but it points the same way as common sense: a brand discussed in many credible places is easier for a model to recognize and trust.
That means earning coverage the honest way: useful contributions to industry publications, accurate listings in the directories people actually consult, real customer reviews, and clear, consistent descriptions of what you do. A word of caution on press releases. A Meltwater study of more than 8 million citations across eight major language models found that earned and news media made up 37.6% of citations, while press releases made up only 0.2% (Meltwater). Press releases can support a broader presence, but do not expect them to work as a citation shortcut. We unpack that further in earned media vs press releases.
Step 6: Keep pages fresh and honest about dates
When a model searches the web for an answer, stale pages are risky to cite. Show a clear published date and a clear updated date, keep them accurate, and update the content when facts change. Do not bump a date without making a real change; it is easy to spot and it erodes trust. For pages that cover fast-moving topics, schedule a review rather than waiting until traffic drops. Our guide to how often to re-audit your website suggests a practical rhythm.
Step 7: Check what a machine actually sees
You cannot improve what you have not looked at. A browser extension that shows green ticks tells you about the page as rendered in your own browser, which is not the same thing as the page a crawler fetches. The point of a GEO audit is to check closer to what a crawler sees: whether your page identity is clear, whether headings are structured sensibly, whether links and images are described, whether performance is acceptable, and whether structured data is present and valid.
iSuggest.ai runs that kind of audit on any URL for free, and audits are not metered. Paste a page into the homepage audit form, read the report, and fix the highest-impact items first. Our post on what an iSuggest.ai audit measures explains each signal so the report reads as a to-do list, not a mystery.
Step 8: Publish where AI systems look
Beyond your own site, you can publish a clean, factual snapshot of your page to the iSuggest.ai AI directory, which has pages for Gemini, ChatGPT and Claude (see the ChatGPT directory). A snapshot is a stable, structured, crawlable record of what your page says at a point in time. Publishing a snapshot spends credits, while auditing does not; the logic is explained in our credit system post. To be clear about expectations: publishing makes your content easier to find and read. It does not guarantee a citation, and no honest vendor can promise one.
What you cannot control, and what to do about it
Some parts of this are outside your hands. Model behavior changes between versions, and the exact way sources are selected is not published. Two identical sites can be cited at very different rates because of factors you cannot see. The sensible response is to focus on the controllable inputs above, measure them over months rather than days, and avoid anyone who sells guaranteed placement.
It also helps to keep ChatGPT in proportion. Google still sends the majority of web traffic, and industry data shows zero-click searches reached about 68% of US Google queries in early 2026 (Digital Applied). Doing the work that makes you citable in ChatGPT tends to help you in Google's AI Overviews and in other assistants as well, because the underlying requirements, readable, well-structured, trustworthy pages, are the same. That overlap is why we argue in GEO vs SEO: ranking is not enough that you should do both rather than choose.
A simple way to start this week
If eight steps feels like a lot, compress it into one afternoon. Check robots.txt for accidental blocks. View source on your top page to confirm the content is in the HTML. Rewrite the first paragraph so it answers the page's main question. Add Organization and FAQPage markup where it is honest to do so. Then run an audit and note your starting score. In a month, run it again and compare. The point is a repeatable loop: measure, fix the biggest gap, publish, measure again.
If you want to see where your most important page stands right now, see how an iSuggest.ai audit works and run it on your own URL. It takes about as long as reading this paragraph, and it will tell you which of the eight steps above deserves your attention first. Then create a free account at your account page when you are ready to publish a snapshot.