What is LLM SEO?
LLM SEO is the work of making a website easy for large language models to find, understand and cite when they answer questions. A large language model (LLM) is the kind of AI that ChatGPT, Claude and Gemini are built on: it writes answers in plain language, and when it searches the web first, it links to the pages it used. For example, ask Perplexity "how long does a dishwasher last" and it answers in a paragraph with three or four sources; LLM SEO is about being one of those sources. The same work is also called generative engine optimisation (GEO) and answer engine optimisation (AEO).
How do LLMs decide which websites to cite?
LLMs learn about websites in two ways, and only one of them is quick to change:
- Training data. Before release, a model is trained on a large copy of the web, collected by crawlers such as GPTBot, ClaudeBot and CCBot. What it learns is months old by the time you use it, and it does not come with links.
- Live retrieval. When a question needs current or specific information, the assistant searches the web, fetches a handful of pages with crawlers such as OAI-SearchBot, Claude-SearchBot and PerplexityBot, and quotes the passages that answer best. This is where cited links come from.
Most of LLM SEO works on the second route, because you can affect it within weeks: let the search crawlers in, make sure they can read your pages without running JavaScript, and put a clear, specific answer where they look first. Our AI crawlers list shows which crawler does which job.
What does the LLM SEO checker look at?
The checker reads your robots.txt, your sitemap and up to 20 public pages, tests 17 AI crawler names, and runs the 40 checks of our AI visibility audit on what it finds. These are the ones that matter most to LLMs in particular:
| Check | Why it matters to LLMs |
|---|---|
| AI crawlers allowed in robots.txt | We test each AI crawler name. A search crawler that is told to stay out cannot fetch the page, so it cannot quote it. |
| No firewall block on AI user agents | Bot protection often turns AI crawlers away even when robots.txt allows them; we compare what an AI crawler and a browser receive. |
| Content is in the raw HTML | Most AI crawlers read the HTML as it arrives and do not run JavaScript, so text added by scripts is invisible to them. |
| Answer-first intro | LLMs quote passages, not pages. The two sentences under a headline should answer it without the rest of the page. |
| Clean, chunkable HTML | Pages are split into chunks before an LLM reads them; real paragraphs, lists and tables split cleanly, a wall of nested boxes does not. |
| Specific numbers and definitions | A passage with a figure or a definition is more useful to an answer than one that only describes. |
| Outbound citations | Claims linked to their sources are easier for an assistant to trust and repeat. |
| Organization data and a consistent brand name | Help an LLM connect your pages to one named organisation, and credit it correctly. |
| llms.txt published | A short markdown guide to your site for LLMs. Worth 1 point of 77; see below. |
Access is where most sites fall down. When we tested the checker on 32 well-known websites in October 2026, 6 turned our crawler away at the firewall and 2 delivered pages that stay empty until JavaScript runs: a quarter of them would be hard for an LLM to quote however good their writing.
How is LLM SEO different from ordinary SEO?
LLM SEO builds on ordinary SEO rather than replacing it, because AI assistants search an index before they answer. Four things change:
- The unit is the passage, not the page. A page can rank with a slow introduction; an LLM will quote whichever paragraph answers the question, wherever it sits.
- More crawlers. As well as Googlebot and Bingbot, each AI company has its own, and robots.txt and firewalls treat them separately.
- JavaScript matters again. Google renders scripts before indexing; most AI crawlers do not.
- Success is a citation, not a click. Many people read the answer and never visit, so being named and linked is the result to measure.
The audit page compares the two side by side.
Does llms.txt help with LLM SEO?
llms.txt might help a little, and it is cheap to add. It is a proposed file, placed at /llms.txt, that lists a site's most useful pages in markdown for LLMs to read.
It is not an agreed standard, and there is little public evidence so far that AI assistants use it when choosing what to cite, so the checker gives it 1 point
and puts far more weight on crawler access and the content of your pages. Our own is at citablereport.com/llms.txt as an example.
Questions and answers
Is the LLM SEO checker free?
Yes. Your score and six of the 40 checks are free, with no signup. You can check three sites a day. The full report, with every result and the fix for each problem, costs US$9 once.
Is LLM SEO the same as GEO or AEO?
Yes, for practical purposes. LLM SEO, generative engine optimisation (GEO) and answer engine optimisation (AEO) are different names for the same work: making a site easy for AI assistants to find, understand and cite.
Does LLM SEO replace ordinary SEO?
No. AI assistants usually search an index before answering: ChatGPT search uses Bing among other sources, and Google's AI Overviews use Google's own. A page that is not indexed is rarely cited, so LLM SEO adds to good SEO rather than replacing it.
Should I block GPTBot and other training crawlers?
That is your choice. Blocking training crawlers such as GPTBot, ClaudeBot and CCBot does not stop the search crawlers (OAI-SearchBot, Claude-SearchBot, PerplexityBot) from fetching and citing your pages, as long as robots.txt allows them by name. The checker marks a site that blocks only training crawlers as a partial pass on that check, and one that blocks search crawlers as a fail.
Does the checker tell me whether ChatGPT mentions my site?
No. It checks your website, not what assistants say today. To see that, ask ChatGPT, Claude, Perplexity and Gemini the questions your customers ask, once a month, and note which sites they cite.
For Google in particular, see our AI Overview checker; for an overview of every assistant, our AI visibility checker.