
Hub
AI search optimization
How to optimize for AI search products in general, from crawler access to quotable answers.
AI search
ChatGPT SEO is the work of making your site easy for ChatGPT to find, read and cite when it searches the web for an answer. Nobody can promise you a place in ChatGPT's answers. You can control whether its crawler is allowed in, whether your pages say what they mean in plain HTML, and whether your answers are clear enough to quote.
ChatGPT SEO is the work of making a website a likely source when ChatGPT searches the web to answer a question. It covers letting OpenAI's search crawler in, making sure the page content is readable without JavaScript, writing clear answers that can be quoted, adding structured data, and staying well indexed in ordinary search engines.
People also call it SEO for ChatGPT, generative engine optimization (GEO) or LLM optimization. The names differ, but the job is the same: make it easy for an AI system to find your page, understand it and trust it enough to cite it. This guide covers ChatGPT specifically. For the wider picture across Google AI Overviews, Perplexity, Claude and Gemini, start with the guide to AI search optimization.
One thing to settle first: "ranking in ChatGPT" is not like ranking in Google. There is no list of positions that stays put. ChatGPT writes each answer for one conversation, and the sources it uses can change with the wording of the question, the user's location and what was said earlier. The realistic goal is to be a strong candidate every time it looks.
ChatGPT answers from two places: what the model learned in training, and what it finds by searching the web for the current conversation. Only the search part produces links and citations. When ChatGPT searches, it rewrites the question into queries, pulls results, reads the most relevant pages and quotes or summarizes passages from them.
OpenAI does not publish a ranking formula for this, so be wary of anyone who claims to know one. What is documented is the plumbing: which crawlers OpenAI runs and what each one is for. Those crawlers decide whether your page can be read at all, which is the part of ChatGPT SEO you fully control.
| User agent | What it is for | If you block it |
|---|---|---|
| OAI-SearchBot | Finding and indexing pages so they can be shown and linked in ChatGPT search | Your pages are unlikely to appear as cited sources in ChatGPT search answers |
| ChatGPT-User | Fetching a page when a user's request needs it, for example when someone pastes your URL or asks about your page | ChatGPT may be unable to open your page when a user asks it to, though OpenAI says robots.txt rules may not apply to these user-initiated visits |
| GPTBot | Collecting content that may be used to train OpenAI's models | Your content is excluded from future training; OpenAI says this is separate from search |
Read the current list on OpenAI's own page, Overview of OpenAI crawlers, before you change anything: names and rules can change. Google's crawler is a separate matter. For Google AI Overviews the crawler is Googlebot, and the Google-Extended token covers training and grounding for Google's Gemini models, not inclusion in Google Search.
ChatGPT SEO is mostly the same work as Google SEO, with a few different details. Both need crawlable pages, accurate content and a site other sources trust. The differences are the crawler names, answers that quote passages instead of listing links, and visits that happen live when a user asks about your page.
To rank in ChatGPT, or more accurately to become a source it is likely to cite, allow OpenAI's search crawler, make sure search engines index your pages, put your content in the HTML, answer real questions in quotable passages, add structured data, keep facts consistent across the web, and measure the visits that arrive.
Open yoursite.com/robots.txt and check for rules that block OAI-SearchBot, or block every crawler with User-agent: * and Disallow: /. Many sites block AI crawlers by accident through a security plugin or a copied template. Decide on GPTBot separately: blocking training does not have to block search. The free robots.txt generator writes rules for each AI crawler and shows what each one is allowed to read.
A robots.txt that says yes does not help if a firewall or bot protection service answers OpenAI's crawlers with a challenge page. Check your CDN or security settings for rules that block AI crawlers, and look in your server logs for requests from OAI-SearchBot and ChatGPT-User that received an error.
Submit your sitemap in Google Search Console and Bing Webmaster Tools, fix pages marked as excluded, and make sure your important pages are not set to noindex. Pages that search engines cannot find are hard for any AI search product to find.
Crawlers may read the HTML your server sends and nothing more. If your prices, product details, FAQs or main text only appear after a script runs, a crawler may see an empty page. Test what your page says before JavaScript with the crawlability test.
Phrase headings the way people ask, such as "How much does X cost?", and follow each one with a short answer that names the subject and stands on its own. Use real lists for steps and real tables for comparisons. Put the answer first, then the detail. Do not pad: a clear page is easier to quote than a long one.
Schema markup states facts in a form machines can read without guessing: your organization, products, prices, articles and FAQs. Check yours with the schema markup validator. An llms.txt file adds a plain summary of your site for AI tools. OpenAI has not said ChatGPT uses llms.txt, so treat it as a cheap extra, not a lever.
Make sure your name, prices, policies and descriptions match on your site, your profiles and the directories that list you. Then watch your analytics for referrals from chatgpt.com and links tagged utm_source=chatgpt.com. That tells you whether people actually arrive, which matters more than whether one prompt shows your name.
For ChatGPT search, robots.txt should allow OAI-SearchBot and ChatGPT-User to read the pages you want cited, and block only private areas such as the cart or account pages. Whether to allow GPTBot is a separate decision about model training. A robots.txt that blocks everything by default blocks ChatGPT search too.
An example for a store that wants to appear in ChatGPT search but opt out of training:
User-agent: OAI-SearchBot
Disallow: /cart/
Disallow: /account/
User-agent: ChatGPT-User
Disallow: /cart/
Disallow: /account/
User-agent: GPTBot
Disallow: /
Robots.txt groups are matched by user agent, and a crawler follows the group that names it most specifically, so a crawler named in its own group ignores the User-agent: * rules. Copy the private paths into each group. Robots.txt is a request that well-behaved crawlers follow, not a lock: keep anything truly private behind a login.
Content that ChatGPT can quote answers one question per section, puts the answer in the first sentence or two, names the subject instead of saying "it", and backs claims with facts a reader can check. It reads well to a person too. Writing for AI answers is mostly writing clearly, with structure that machines can follow.
| Hard to quote | Why | Easier to quote |
|---|---|---|
| "As mentioned above, it depends." | Points elsewhere and gives no answer | Name the subject and state the answer, then the conditions |
| A long story before the answer | The answer is buried deep in the section | Answer in the first 40 to 60 words, then tell the story |
| Steps written as one paragraph | Order and boundaries are unclear | A numbered list with one action per step |
| A comparison inside an image | Crawlers cannot read text in pictures reliably | An HTML table with labeled columns |
| "Industry-leading results" | A claim with nothing to check | A specific, sourced fact, or nothing |
Two cautions. First, do not invent statistics to look authoritative: an AI answer that repeats a made-up number with your name on it does you no favors. Second, keep a named author on guides and keep dates honest. Both help readers judge a page, whatever a model does with them.
The most common ChatGPT SEO mistakes are blocking OpenAI's search crawler by accident, hiding content behind JavaScript, treating one prompt test as a ranking, buying services that promise citations, and publishing thin AI-written pages in the hope that volume will get noticed.
Audit and rebuilds
The free CRO audit checks your page for people and for AI agents, with eight checks scored out of 100 and every finding, with how to fix each one. One of the checks covers AI visibility and agent readiness: it reads your robots.txt rules for AI crawlers, your llms.txt, your structured data and what your HTML says without JavaScript. It does not ask ChatGPT anything and does not track citations. The AI visibility checker runs the same audit with that check first. If you want the fixes built for you, the one page rebuild has Convertica's team build every fix the audit found on that page, and the full site rebuild covers your whole site.
In this series

Hub
How to optimize for AI search products in general, from crawler access to quotable answers.
Free tool
Check AI crawler rules, llms.txt, structured data and your HTML without JavaScript.
Free tool
Write robots.txt rules for OAI-SearchBot, GPTBot and other AI crawlers.
ChatGPT SEO means making your site easy for ChatGPT to find, read and cite when it searches the web. You cannot buy or force a citation. You can allow OpenAI's search crawler, keep your answers in plain HTML, write clear passages that answer real questions, and stay well indexed in ordinary search engines, which ChatGPT search also draws on.
ChatGPT SEO is the practice of making a website easy for ChatGPT to find, read and cite when it searches the web to answer a question. It covers crawler access for OpenAI's search crawler, content that is readable without JavaScript, clear answers to real questions, structured data and a good standing in ordinary search engines.
There is no fixed ranking in ChatGPT, because each answer is generated for one conversation. To be a candidate source, allow OAI-SearchBot in robots.txt, make sure your pages are indexed by search engines, put clear answers in the page HTML, and keep facts such as prices and policies consistent. None of this guarantees a citation.
OpenAI's crawler documentation treats them as separate choices. GPTBot collects content that may be used to train models. OAI-SearchBot is the crawler for showing sites in ChatGPT search. A site can block GPTBot and still allow OAI-SearchBot, so check both rules in your robots.txt.
Mostly not. The foundations are the same: crawlable pages, clear content, accurate facts and a site other sources trust. The differences are the crawler names, the fact that answers quote passages rather than list ten links, and that ChatGPT may fetch your page live when a user asks about it.
There is no public statement from OpenAI that ChatGPT reads llms.txt when choosing sources. It is a cheap, low-risk file that gives AI tools a plain summary of your site, so it is worth adding, but treat it as a helpful extra rather than a ranking factor.
Look in your analytics for referrals from chatgpt.com, and for utm_source=chatgpt.com on landing page URLs, which ChatGPT adds to many links it shows. Some visits arrive with no referrer at all, so treat the number you see as a floor.
Some tools send sample prompts to ChatGPT and record which sites appear. That gives a rough picture, but answers change with wording, location, memory and time, so a single result is not a stable position. Use those tools for trends, not as a ranking.
SEO for ChatGPT goes by several names: generative engine optimization (GEO), LLM optimization (LLMO), answer engine optimization (AEO) or AI search optimization. The names differ, but the job is the same: make it easy for an AI system to find a page, understand it and trust it enough to cite it.
No. AI answers that cite websites still need pages that can be crawled, indexed and understood, which is ordinary SEO work. What changes is the goal: being a quoted source inside an answer as well as a ranked link. Treat ChatGPT SEO as an extension of SEO, not a replacement for it.
Ask Convertica
Convertica
Hi, I'm Convertica's AI assistant. Ask me anything about the free CRO audit, the one page and full site rebuilds, or any of the free tools.