LLM SEO, also called LLM optimization or LLMO, is the work of being findable and citable by ChatGPT, Gemini, Claude and Perplexity. The systems read differently, use different crawlers, and cite with different habits. The work is ordinary SEO with four additions: crawler access, entity consistency, third-party presence, and separate measurement per model.
What each system reads, what is sold under the name that does not work, the seven-step method, and llms.txt.
LLM SEO, also called LLM optimization or LLMO, is the work of making a company and its pages findable and citable by large language model systems such as ChatGPT, Gemini, Claude and Perplexity. It consists of allowing each system's crawler, writing passages a model can quote whole, stating specifics a model can attribute, keeping entity data consistent across the web so the model can place the company, and being present in the third-party sources each model reads. It is the same body of work as generative engine optimization under a different name.
The systems differ in what they read. ChatGPT composes from training data and a live search layer and leans on directories and roundups. Gemini and Google AI Mode draw on Google's index and cite what ranks. Claude searches and favors specific, sourced text. Perplexity runs a live search and cites prominently, with numbered sources beside the answer. A company that wants to be named by all of them has to be retrievable by all of them, which starts with four lines in robots.txt.
Prompt injection in page text, keyword density for "AI", hidden text addressed to the model, a separate "AI section" on the site, submission services, and blended visibility scores. None of these affects whether a model cites a page. Models cite passages that answer the question, from sources they can retrieve, about entities they can resolve. The work is the same discipline as good writing and clean data, measured per engine.
The tell for a vendor selling the right column is the absence of measurement. Ask how they count mentions, in which engines, with which queries, and whether separately. A vendor who answers with a single score has not measured anything a developer could act on.
Seven steps: allow every major AI crawler in robots.txt and confirm the server returns pages to each; publish an llms.txt describing the site; write each page to answer one question with the answer in the first paragraph; replace adjectives with figures, names and dates; add Organization, WebPage and FAQ structured data; make name, address and category identical everywhere on the web; then ask each model the buyer's question monthly and record who it names.
llms.txt is a proposed plain-text file at a site's root that tells language model systems what the site is, who it is for, and which pages matter, in a form a model can read in one pass. Adoption by the model providers is partial and unconfirmed; it costs nothing, it cannot hurt, and it is the most direct way to hand a model a map of the site. This site publishes one, along with a full-text version at llms-full.txt.
The file is deliberately simple: a title, a one-paragraph description, and a list of pages with a sentence each. Its value is not that any provider has committed to reading it. Its value is that writing it forces the site to state in plain words what it is for, which is the same discipline the pages need. The version on this site took an hour and reads as a summary a buyer could use.
Measured effect: none we can attribute yet. Published position of the providers: unconfirmed. Cost: one file. We publish it on that arithmetic.
It is ordinary SEO with four additions: crawler access for the AI bots specifically, entity consistency across the web rather than only on the site, presence in the third-party sources models read, and separate measurement per model. For Gemini and Perplexity, which cite what ranks, ordinary SEO is most of the job. For ChatGPT, which cites what it has encountered consistently, the additions are most of the job.
The GEO vs SEO page sets out the overlap by engine with the published figures. The ChatGPT and AI Overviews pages cover the two engines that differ most. The AI visibility audit is how the four are measured.
Figures described as measured are from Search Standing's own analyses using Semrush, Google Search Console and direct queries to each AI engine, on the dates stated. Published ranges move quarter to quarter; the patterns are what we rely on.
Send your company and your market. We will run the analysis and send back a summary naming your competitors, the terms they are winning, and where you sit against them.
No call required and no obligation. Read it, check it yourself, and decide. If the findings are worth discussing after that, we can talk then.
Prepared for your company against the competitors you name. Written by the person who would run the paid work.
Or email contact@searchstanding.com directly.