Is your page technically ready for AI search?
We run seven checks on whether AI tools such as ChatGPT, Claude, Perplexity and Copilot can reach, fetch, parse and attribute your page. It tells you whether the page is technically ready to be found and cited in AI search. It's a technical check only; your writing isn't scored.
What we check
Crawler access and permissions
Whether AI crawlers are allowed to reach this page at all. This is the check that can make every other one irrelevant.
Crawl path and fetch economics
Whether a crawler can find this page, and fetch it before giving up.
Renderable content
Whether the page's text is in the raw HTML, or appears only after JavaScript runs, which no AI crawler except Google's does.
Parseability and extraction structure
Whether the markup lets a machine separate content from furniture and lift a coherent passage. AI systems cite passages, not whole pages.
Structured data and entity graph
Whether the page labels itself for machines, and anchors its entities to identifiers AI systems can resolve.
Provenance and freshness signals
Whether the page answers who wrote this, when, and on whose authority — in a form a machine can read.
Indexation directives and citation surface
Whether anything blocks this page from being indexed, and how it gets titled and summarised when an AI system quotes it.
Why other GEO scores disagree
If you run the same page through several tools, you'll get wildly different numbers. We've seen scores of 46 and 88 out of 100 for the same site on the same day. That's usually because they measure different things under the same label, weight them by undisclosed rules, and sometimes tune the result to make a product look necessary.
This one scores 35 binary sub-checks against fixed, published pass conditions. Every result is a recorded observation you can re-check yourself, and it scores the same pages the same way every time.
Every check runs on publicly served information, using ordinary browser requests. We never impersonate a crawler, evade a block, or store personal data about the sites we scan.
Common questions
Generative engine optimisation is the work of getting your brand into AI-generated answers. Most of what is written about it concerns content strategy. This is the unsexy technical part: can an assistant reach your page, fetch it affordably, render it, parse it, and work out that it is yours? If any of those is a no, nothing else you do matters much.
A lot of it overlaps, because AI crawlers and search crawlers want similar things. The gaps between them are what matters. AI systems fetch far more conservatively than Googlebot, several ignore JavaScript completely, and they have to attribute a claim to a named entity rather than simply rank a URL. So a page can be technically ready for Google search and still never reach an assistant.
No. Be wary of any tool that suggests otherwise. This measures whether assistants can read you at all, which is a precondition, not a result. What they then say about you, and how you look beside your competitors, depends on everything written about you across the web. That is a different question, and the one our AI-visibility audit answers.
The named agents behind the major assistants: GPTBot and ChatGPT-User for OpenAI, ClaudeBot for Anthropic, PerplexityBot, Google-Extended and Applebot-Extended among them, plus the catch-all rules that apply when no agent is named. It reports what your robots.txt permits. It also flags where a CDN may be overriding that above your robots file, because those rules cannot be seen from outside and plenty of sites are turning crawlers away without knowing it.
No. Your copy is not scored, and nothing in the result is a verdict on how well it is written. The checker asks only whether the words you have already published can be reached and read by a machine. A page can be the best thing on the internet and still score zero, because a firewall turned the crawler away before it saw a single sentence.
We fetch it the way an ordinary browser would, using publicly served information only. We do not impersonate a crawler, we do not work around a block, and we do not store personal data about the sites we scan. We keep the results so we can see how sites perform against these checks over time. If you unlock the full report, your details are kept for twelve months and then deleted.
After anything that changes your site's structure, its templates or its CDN rules. Those are what move the result. Every score is stamped with the date it was run, because the ground shifts on its own: Cloudflare's three-tier system for AI crawlers arrives on 15 September 2026 and will change the defaults on millions of sites without anyone at those sites touching a setting.