llms.txt: what it is and how to write one
llms.txt is a short Markdown file that gives AI assistants a map of your most important content. Here is the format, an example and how to keep it useful.
GEO & AI search
BoostSEO Team
3 min read
What is GEO? Generative engine optimization explained
AI assistants now answer many questions before anyone clicks a link. GEO is how you make sure your content is the one they read, trust and cite.
BoostSEO Team
3 min read
On this page
Search is no longer only ten blue links. Google shows AI Overviews, Bing has Copilot, and millions of people ask ChatGPT, Gemini, Claude or Perplexity instead of searching. These systems read web pages, summarise them and — when you are lucky — cite the source. Generative engine optimization (GEO) is the work of making your pages the ones they choose.
How AI answer engines pick sources
Most AI assistants combine a language model with live retrieval. When a question needs fresh or specific information, the assistant runs searches (its own index or a partner search engine), fetches a handful of pages, extracts the passages that answer the question and writes a response. Three things decide whether your page is used:
- Retrievability — the page must be crawlable, indexed and ranking reasonably well for the underlying searches.
- Extractability — the answer must be easy to lift: a clear heading, a direct answer in the first sentences, lists or tables for steps and comparisons.
- Trust — the page should look authoritative: an identifiable author or brand, sources for claims, up-to-date dates and consistent information across the web.
GEO vs SEO: what changes and what doesn't
| SEO | GEO | |
|---|---|---|
| Goal | Rank and earn clicks | Be quoted, cited and recommended in AI answers |
| Unit | The page | The passage (a heading plus its answer) |
| Signals | Relevance, links, experience | The same, plus clarity, facts with sources and brand consistency |
| Technical base | Identical: crawlable HTML, fast pages, correct robots.txt, sitemaps, structured data | |
In practice GEO is not a separate discipline. It is SEO done with the reader — human or machine — who wants a precise answer.
A practical GEO checklist
- Let the right bots in. Check robots.txt for rules that block search and answer bots such as OAI-SearchBot, ChatGPT-User, Claude-SearchBot, PerplexityBot and Bingbot. You can still block training crawlers if you prefer.
- Serve content in HTML. Many AI crawlers do not execute JavaScript. If your text only appears after scripts run, they may see an empty page.
- Write question headings and answer first. Use the questions your audience asks as H2/H3 and answer each in 40–60 words before adding detail.
- Add data with sources. Specific numbers, dates and definitions with a citation are more quotable than general statements. Never invent figures.
- Show who is behind the content. Author names and bios, an About page, contact details, and Organization schema with
sameAslinks to your official profiles. - Keep it fresh. Show published and updated dates and revisit key pages regularly.
- Publish an llms.txt that summarises your site and links to your most important pages.
How to measure GEO
There is no “AI Search Console” yet for most assistants. The pragmatic approach is to define 20–50 questions that matter to your business, ask them regularly in the main assistants and record whether your brand is mentioned and whether your site is linked. BoostSEO Pro automates this with the AI visibility check, and the audit's GEO score tracks the on-site factors you control.
Where to start
Run a free audit and open the GEO tab. Fix blocked crawlers and JavaScript-only content first, then rewrite your top pages so every important question has a clear, sourced answer.
SEO & GEO guides from the team behind BoostSEO and ToolCMB.