LLM SEO: How ChatGPT, Claude and Perplexity Name Your Company

LLM SEO makes sure that large language models such as ChatGPT, Claude, Gemini or Perplexity name your company when someone asks for a provider. For that, your website has to answer your customers' questions so clearly that a model can take over the answer. For you, that means you make the shortlist before a prospect has even visited a website.

What is LLM SEO?

LLM stands for large language model. Classic SEO targets Google's results list. LLM SEO targets the answer a model writes. In that answer there are no ten results, just a handful of named providers.

Several names are in circulation for the same topic: LLM SEO, LLM optimization, GEO or AEO. At their core they mean the same. Where the term GEO comes from is explained in our guide Generative Engine Optimization.

What this means for your company: The shortlist gets narrower. If you are missing from the answer, you are not compared in that search.

How do language models choose providers?

A language model knows your company in two ways.

From what it has learned. Models learn from large amounts of public text. What is often and consistently written about you there, the model remembers. This knowledge has a cutoff date. A new service or a new name is therefore missing until a new model version is released.

From a search during the answer. ChatGPT with search, Perplexity, Claude with web search and Google's AI overviews fetch current pages, read them and summarize them. What counts here is whether your page is found and whether it answers the question directly.

What this means for your company: You can influence the second route within weeks, the first only over months. Both start with a website that says clearly what you do, for whom and where.

What to change on your website

Most improvements are not programming work but better answers. Check your most important pages for these points:

• The answer comes first. The first paragraph answers the question in two to three sentences. A model likes to take over exactly that paragraph.
• One page per service and per frequent question. A model is more likely to recommend a page that matches the question exactly than a catch-all page.
• Clear statements instead of advertising phrases. Process, region, languages, who you work for. No model quotes “We offer tailor-made solutions”.
• Consistent details. Company name, services and region identical everywhere, on your website and in directories.
• Readable for the AI providers. Your pages must be accessible to their crawlers and describe services unambiguously (structured data and a file called llms.txt exist for this).

What this means for your company: Good answers help twice. They earn mentions in AI answers and also convince the people who read your page.

What others write about you

Models trust statements more when they appear in several places. Check what industry directories, review portals, partner pages and trade media write about you. Outdated details there would otherwise end up in the answer.

What this means for your company: A mention in a trade article can be worth more than another blog post on your own site.

How to check whether LLM SEO works

There is no Google ranking for ChatGPT. You can still track the effect:

1. Create a question list. Write down 20 to 30 questions your customers ask. For example: “Which providers of [service] are there in [region]?”
2. Query it every month. Ask these questions in ChatGPT, Claude, Perplexity and Gemini. Note who is named.
3. Count visits. Your web analytics shows visits coming from chatgpt.com, perplexity.ai and similar addresses.
4. Ask in the first conversation. “How did you find us?” is often the most reliable source.

Answers from language models fluctuate. A single query says little; the development over several months shows the trend.

What this means for your company: In the end, measure by enquiries, not by mentions. A mention nobody clicks does not pay an invoice.

How does LLM SEO differ from classic SEO?

The foundation is the same: pages that get found and answer real questions. The difference lies in the result. At Google, your position in the list counts. With a language model, what counts is whether you are named and how the model describes you.

Many AI assistants look for sources through classic search engines. What Google does not find, the model often does not find either. You can read more about the distinction in SEO or GEO, and concrete steps for ChatGPT in Getting recommended in ChatGPT.

What this means for your company: You do not have to start over. If you have good SEO, you build on it.

Frequently asked questions