SEO for LLMs
SEO for LLMs: how to get cited by AI assistants
Large language models answer from two places: what they learned in training and what they fetch live. You influence the second far more than the first. This guide covers the practical parts: access, wording, and proof.

350+businesses served
8countries
7+years
In short
SEO for LLMs means making your pages reachable by AI crawlers, easy to quote, and credible. Allow the bots you want in robots.txt, serve readable text without relying on scripts, put clear answers in self-contained passages, state facts with sources, and earn mentions on trusted sites. Nobody can promise a citation, so measure by testing prompts.
Rather have it done for you? See our AI search optimization service, from $1,500.
On this page 6 sections
How models find your content
There are three routes. Training: the model learned from a snapshot of the web, and you cannot change that quickly. Live retrieval: assistants such as search-connected chatbots fetch pages when a question needs fresh facts, usually through a search index plus their own crawler. User-triggered fetches: a person pastes a link, or asks the assistant to open a page.
Most of what you can do concerns the last two. Pages that are indexed, crawlable and easy to extract answers from have a better chance of being retrieved. Pages that are blocked, slow, thin or hidden behind scripts usually do not.
The crawlers to know
Names and behavior change, so confirm each one in the vendor's own documentation before editing rules. These are the common ones at the time of writing.
| User agent | Run by | Typical purpose | Blocking it affects |
|---|---|---|---|
| GPTBot | OpenAI | Collects content that may be used to train models | Training use, not live search results |
| OAI-SearchBot | OpenAI | Finds pages for search features in ChatGPT | Appearing in ChatGPT search |
| ChatGPT-User | OpenAI | Fetches a page when a user asks | Pages opened on request |
| ClaudeBot | Anthropic | Collects content that may be used for training | Training use |
| PerplexityBot | Perplexity | Indexes pages for its answer engine | Appearing in Perplexity results |
| Google-Extended | A control token for use of content in Gemini models; it does not change Search | Gemini training and grounding use | |
| Googlebot | Crawls for Search, which also feeds AI Overviews | Search and AI Overviews | |
| Bingbot | Microsoft | Crawls for Bing, which feeds Copilot | Bing and Copilot |
Block training crawlers and you can still be cited in live answers, because those use different agents. Block the search agents and you remove yourself from those answers. Decide per bot, deliberately.
Rather have it done for you?Talk to a person now, or ask for a fixed quote. We reply within one working day.
Steps to make a page citable
- Check accessRead robots.txt and test as each bot. Also check the firewall, CDN bot rules and rate limits. Many sites block AI crawlers by accident through a security setting.
- Serve the text in the HTMLFetch the page without running scripts. If the answer is not in the raw HTML, many crawlers will not see it. Server-side rendering fixes this on JavaScript sites.
- Write self-contained passagesModels lift paragraphs, not pages. Each section should make sense alone: state the subject, then the answer, in two or three sentences.
- State facts plainly with a sourceDates, numbers and definitions should be exact and attributed. Vague promotional text is rarely quoted.
- Name the entity clearlyUse the same business name, address details and descriptions on your site and profiles. Add Organization or LocalBusiness schema so the facts are explicit.
- Earn mentions elsewhereAI answers lean on sources they trust: industry publications, reviews, directories and well-known reference sites.
- Refresh dated pagesModels that retrieve live content prefer current facts. Update the date only when the content really changed.
What about llms.txt?
An llms.txt file is a plain text file at the root of a site that lists key pages in a simple format for language models. It is a proposal, not a standard. Google has said it does not use the file for Search, and the major assistants have not confirmed that they rely on it. Adding one is cheap and does no harm, but it is not a substitute for crawlable pages. Our llms.txt generator creates a draft.
How to test whether it is working
- Write twenty questions your customers ask. Run them in the main assistants each month.
- Record whether your brand is named, whether your page is a cited source, and whether the facts are correct.
- Check analytics for referral visits from assistants.
- Look at server logs for the bots above to confirm they fetch your pages.
- Compare with Search Console. Pages that gain impressions often gain citations later.
Answers vary between runs and users, so look for patterns across several runs, not a single result. See the AI visibility checker and our GEO checklist for a fuller routine.
Pitfalls
- Blocking every AI bot out of caution, then wondering why you never appear.
- Allowing all of them without thought about training use of your content.
- Hiding key facts in images, tabs or pop-ups.
- Keyword-stuffed pages written for machines. Models and search systems read for meaning.
- Believing a service that promises mentions in ChatGPT.
We cover this as part of generative engine optimization. You get a fixed quote after a short call on the contact page.
Updated 20 November 2026
Questions, answered
Something else on your mind? Ask us directly or WhatsApp us. We reply within one working day.
Can I make ChatGPT recommend my business?
Should I block GPTBot?
Does llms.txt help SEO?
Is SEO for LLMs different from normal SEO?
How do I know if an AI assistant visited my site?
Keep exploring
GEO checklist: earn citations in AI answers
Generative engines write an answer, then cite a handful of sources. Your job is to be a source they can find, understand and trust. This checklist is…
Free toolllms.txt generator
Add your site name, a short summary and your most useful pages with a note on each. The generator returns a formatted llms.txt file ready to upload…
ServiceGenerative engine optimization (GEO): be the name models recommend
When a model is asked to recommend a provider, it draws on what the web says about you. GEO is the work of making that picture accurate, consistent…
See how Google and AI see your site.
A free SEO and AI visibility check with the top fixes, within one working day. No obligation.








