Guide

Why your site is not showing up in ChatGPT and AI answers

Your pages are indexed in Google, yet ChatGPT and other AI answers never mention your site. Here is why that gap exists and what to check.

By Sunny Patel · Updated 2026-07-15

You searched a question your business answers, typed it into ChatGPT, and your site was not there. You checked Google Search Console and the page is indexed. That is the confusing part: Google clearly found it, so why did the AI answer skip it entirely.

The short answer is that indexed and cited are different states, checked by different systems, and passing one does not carry over to the other. A page can be perfectly crawlable, fully indexed by Google, ranking on page one, and still never get named or linked in an AI generated answer. That gap is common, it is measurable in parts, and most of it comes down to a short list of concrete things worth checking on your own site before assuming the content itself is the problem.

Being indexed by Google is not the same test AI systems run

Each AI answer engine runs its own crawler under its own rules, separate from Google's index, and in some cases fetches a page live rather than relying on a stored index entry at all. A page Google indexed years ago can still be invisible to every one of these systems if the relevant crawler was blocked, never found the page, or fetched a version of it that did not read the way the page looks in a normal browser.

Google's own AI features still require indexing, and still do not guarantee it

Google's documentation on AI features in Search states that a page has to be indexed and eligible to be shown in Google Search before it can ever appear inside an AI Overview, and adds that meeting every requirement does not guarantee a page gets crawled, indexed, or served. That is Google describing its own AI feature, built on top of its own index. ChatGPT, Perplexity, and Copilot answer to none of it.

OpenAI runs two different crawlers for two different jobs

OpenAI documents two separate crawlers. GPTBot crawls content that may be used in training OpenAI's generative AI foundation models, while a second crawler, OAI-SearchBot, is the one that surfaces websites in ChatGPT's search results and answers. Blocking GPTBot opts a site out of training data. Blocking OAI-SearchBot instead means the site will not be shown in ChatGPT search answers at all. A site can allow one and block the other, and each choice changes a different outcome.

Perplexity draws the same line

Perplexity's documentation states that PerplexityBot is designed to surface and link websites in search results on Perplexity and is not the crawler Perplexity uses to train its models. The pattern across all three providers is the same: training crawlers and answer crawlers are separate systems with separate rules, and Google's index sits outside both.

Crawlable, indexed, cited: three separate states

It helps to name the three states explicitly, because most of the confusion comes from treating them as one continuous thing when they are really three different checkpoints, each run by a different system.

Crawlable is the simplest condition: the page returns a normal response, robots.txt does not block the crawler in question, and a working path leads in from the rest of the site's internal link graph. Orphan pages and shallow-versus-deep click paths both describe ways this first condition fails quietly. Indexed goes a step further: a search engine or AI system has actually processed the page and added it to whatever internal store it uses to answer future queries. Google Search Console can tell you whether Google has indexed a page. Most AI systems offer no public equivalent, so there is often no direct way to confirm a page has been indexed anywhere beyond Google's own index.

Two guides on this site cover the mechanics of the crawlable stage in more depth: orphan pages and click depth.

Cited is the state that actually matters commercially. A real person asked a real question. The AI system's answer named or linked your page as a source. A page can be crawlable and indexed and still never get cited, because the system judged another page a better match for the specific way people were asking, or because the page's content did not answer the question as directly as the retrieval step rewarded. Retrieval systems behind AI answers typically favour content that answers a specific question clearly, with unambiguous claims, over content that is technically crawlable but vague, thin, or buried behind unnecessary preamble. A page can pass every crawlability and structure check and still lose out to a competitor's page that simply answers the underlying question more directly.

What to check on your own site

None of the checks below can tell you whether an AI has actually cited a page. What they can tell you is whether the structural and access barriers that commonly cause the gap are present on your site right now, which is the part you can fix directly.

AI-bot access

Check robots.txt for a rule that blocks GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, or Google-Extended. A blanket Disallow: / under any of those user agents removes that system from the running before content quality ever comes into it. This is the single fastest thing to rule out, because it is a yes or no answer rather than a judgement call.

Site structure: orphans and click depth

A page with zero inbound internal links, or one sitting many hops deep from the homepage, is harder for a link-following crawler to reach at all. This applies to AI crawlers with the same or greater force than it applies to Google, because most AI-only crawlers lack Google's decades of accumulated URL history to fall back on when internal linking falls short. See internal links for AI crawlers for the full mechanics.

What the raw HTML actually contains

A page that looks complete in a browser can still serve a near-empty shell to a crawler that reads raw HTML and does not execute JavaScript. If the words a person sees only appear after client-side rendering, a crawler that skips script execution reads a page with little or nothing on it, regardless of how much content exists once the page finishes loading.

AI readiness score

A composite score across content, access, structure, metadata, and schema gives a single number to track over time and a breakdown of which part is weakest on a given page. A low score flags a real structural weakness worth fixing. A high score does not mean a page will be cited, only that the common structural barriers are not the thing holding it back.

If the structure is clean and the site is still not cited

Once robots.txt allows the relevant bots, the page is not orphaned or buried deep in the link graph, and the raw HTML contains real content, the remaining gap usually sits in how directly the page answers the specific question people are asking. That is a content and framing problem rather than a technical one, and it is a different diagnosis from anything a structural scan can run. Fixing the structural side first is still worth doing, because a page that fails an access or crawl check never gets the chance to be judged on its content at all.

Frequently asked questions