AI SEO · GEO · AEO
AI visibility: the complete guide
Traditional SEO gets your page ranked on Google. AI visibility gets your content cited by ChatGPT, Perplexity and Claude. This guide covers both — llms.txt, AI crawlers, schema for LLMs, and how to measure it all.
Guides
AI SEO guides
The scoring
What does an AI visibility score actually measure?
Every report this site produces carries a GEO score next to the classic SEO score. It is not a vibe check: it is six weighted factors, computed from the page and the domain it sits on. The weights below are imported directly from the engine that runs your audit, so what you read here is exactly what the checker applies.
- llms.txt — 22%: Whether you serve an llms.txt at the root of the domain and whether it is well formed.
- AI crawler access — 18%: Whether robots.txt lets the AI crawlers through (GPTBot, ClaudeBot, Claude-Web, PerplexityBot, Google-Extended, CCBot, Bytespider).
- Structured data — 18%: Organization and WebSite schema on the homepage; Article, FAQ, Breadcrumb or HowTo schema on the page itself.
- Answer structure — 18%: Heading density, question-style headings, a summary near the top, and lists or tables an answer engine can lift verbatim.
- Entity clarity — 13%: sameAs links, consistent naming across title, schema and visible text, a named author, and about or contact pages.
- Freshness — 11%: Published and modified dates, both in schema and in visible text.
Two things fall out of that list. First, more than half the score is decided before a single word of your copy is read — crawler access, llms.txt and schema are infrastructure, and they are the cheapest points on the board. Second, the content half rewards a specific shape: a question, a direct answer under it, then the detail. Answer engines lift passages, not pages.
Where should you start?
Most AI-SEO advice starts with crawler access. Our own audits say that is the step you will most likely pass. Across the 82 third-party sites this checker has audited (newest report per site, measured October 30, 2026), 6 block any of the AI crawlers we test in robots.txt. 52 write their headings as labels instead of the questions a reader would type, and 36 ship no FAQPage or HowTo schema. Almost nobody is locked out. Most sites simply give an answer engine nothing it can quote.
So spend the effort where the failures are:
- Check crawler access, but expect it to pass. It takes seconds with the AI crawler checker, and a block there is fatal, so it goes first. It is almost never where the work is.
- Rewrite your H2s as questions, each followed by a one- or two-sentence answer before the detail. This is the most common failure in the sample by a wide margin.
- Add FAQPage or HowTo schema where the page already has that shape, plus Organization and WebSite schema on the homepage.
- Publish an llms.txt that describes every link it lists. Most files we found get the format right and leave the descriptions out.
The full failure table, with method and caveats, is in how to get cited by AI. These are counts within a small, self-selected sample, not rates for the web.
Frequently asked questions
What is AI visibility?
AI visibility is how likely an answer engine — ChatGPT, Perplexity, Claude, Google AI Overviews — is to fetch your page, understand it and cite it in an answer. It depends on crawler access, machine-readable structure and clear entity signals, not on classic ranking factors like backlinks alone.
How is GEO different from SEO?
SEO optimises for a ranked list of ten blue links. GEO (Generative Engine Optimization) optimises for being quoted inside a generated answer, where there is no list and often a single source per claim. The technical foundations overlap — crawlability, speed, structure — but GEO adds AI-crawler access, llms.txt and answer-shaped content.
Do I need an llms.txt file?
It is not required and no answer engine guarantees it will read one. It is a low-cost, curated pointer to the pages you most want cited, and it is the single heaviest factor in our GEO score (22%) because it is also a proxy for whether a site has thought about AI access at all.
Does blocking GPTBot hurt my traffic?
Not by itself. GPTBot collects training data; the agents that fetch a page while an answer is being written are ChatGPT-User and OAI-SearchBot, and they obey their own robots.txt groups. Blocking GPTBot keeps you out of future training sets without removing you from live ChatGPT answers — and that is exactly the line sites draw in practice: of the 52 sites whose robots.txt we surveyed, 4 block an answer-engine crawler outright; every one of them blocks the training collectors GPTBot and ClaudeBot, and none of them blocks ChatGPT-User, OAI-SearchBot or PerplexityBot. Google-Extended is likewise separate from Googlebot, so classic rankings are unaffected either way.
How do I check my own AI visibility?
Run the free checker on any URL. Every report includes a GEO score built from the six weighted factors below, alongside the 61 classic SEO checks — no signup, no email, and the full report is visible immediately.
Free tools
Test your AI visibility for free
Start with your own site