Skip to content
Digital Trust Audit

What is LLM SEO?

Published 20 September 2026. Last checked 20 September 2026.LLM SEO is optimising a website so that large language models can read it, quote it and name the business behind it when a user asks a relevant question. It is the same practice as answer engine optimisation under a different name, and the difference between the two terms is vocabulary rather than method.

Why there are three names for it

Answer engine optimisation, generative engine optimisation and LLM SEO all describe the work of being named in an AI-generated answer. The category is new enough that no term has settled. “LLM SEO” is the one people search for; “answer engine optimisation” is the one that describes the thing most precisely → answer engine optimisation

What an LLM can and cannot see

A model answering a live question either retrieves pages at the moment it is asked, or draws on what it was trained on, or both. That has three consequences worth knowing:

  1. Content behind JavaScript may not exist. Some retrieval systems do not execute scripts. If the copy is not in the server-rendered HTML, it may not be read at all.

  2. Blocked crawlers mean no citation. A system that cannot fetch a page cannot cite it — and most blocks are accidental, in a firewall or CDN rule.

  3. Training data is old and retrieval is live. What a model “knows” about a business may be a year or more out of date, while what it retrieves is current. A wrong fact repeated across the web is harder to correct than a wrong fact on your own site.

Whether to allow AI crawlers

Blocking them removes a site from what can be read and cited. It also removes it from what can be trained on, which is why some publishers block deliberately.

For a business that wants to be recommended, blocking is self-defeating. Check what is actually happening rather than what was intended: most businesses that block these agents did not decide to.

How to measure it

Not by inspecting the site. By asking the systems your buyers’ questions, under fixed conditions, more than once, and recording who gets named → the AI Visibility Audit

What the term means in practice is AI visibility, and the machine-readability half is covered by the free website health check.

Questions

Digital Trust AuditVerified from the outside in.

The work is real — crawlability, extractable structure, entity clarity, corroboration. The name is not settled.

The seven systems that answer buyers' questions rather than return links: Google AI Overview, Google AI Mode, ChatGPT, Copilot, Claude, Perplexity and Gemini.

Not directly, and nobody can sell you that. What is influenceable is what a system retrieves and cites when asked live.

No format is required. Content in the server-rendered HTML, clear headings, direct answers and valid structured data cover almost all of it.

Measure it by asking the systems.

Fixed questions, fixed conditions, more than one run, and a record of who gets named.

Request an AI Visibility Audit