SchemaValid
ENDE

AI Search Readiness Checker

Enter any URL and see how well that page is prepared for AI search engines like ChatGPT, Perplexity, Gemini and Google AI Overviews. You get a score out of 100 and a checklist of exactly what to fix — free, no account needed.

How this check works

The check runs entirely on our server: we fetch your page and its supporting files the same way an AI crawler would, then score eight categories that decide whether a model can reach, read and quote your content.

  1. Enter your URL. Any public page works — your homepage is the usual starting point.
  2. We fetch and analyse. Your page, robots.txt, sitemap.xml and llms.txt are read in parallel and checked against eight categories.
  3. You get a score and a fix list. Every category expands into concrete findings and a recommendation, ranked by how much it costs you.

Why AI search readiness matters

Search is shifting from a list of links to a synthesised answer. ChatGPT, Perplexity, Gemini and Google AI Overviews read pages, pick sources and cite a handful of them. If a model cannot fetch your page, cannot tell what it is about, or finds nothing substantial to quote, you are absent from that answer entirely — there is no page two to fall back on. The signals that decide this are unglamorous and technical: crawler permissions, structured data, honest meta tags and real content. That is exactly what this tool measures.

What gets checked, and why it's weighted this way

Not every signal carries the same weight. AI crawler access and structured data are worth 20 points each, because they decide whether a model can read your page at all and whether it understands what it is looking at. Meta tags and content structure are worth 15 each: they determine how quotable you are once you're in. HTTPS and canonical are worth 10, llms.txt another 10 — a promising standard, but not yet read by every provider. Sitemap and indexability are worth 5 each as hygiene checks. Within each category, a pass earns full points, a warning earns half, and a fail earns none.

What this tool does not do

This check looks at one page, not your whole site, and it measures technical readiness rather than actual visibility. It cannot tell you whether ChatGPT currently cites you — nobody can measure that reliably, and any tool claiming otherwise is guessing. What it does tell you is whether the technical groundwork is in place, which is the part you control. We fetch only publicly reachable pages, store nothing, and run every check live.

Missing an llms.txt is the single most common gap this check finds. Generate one for free in seconds

Frequently asked questions

Does a good score mean ChatGPT will cite my site?
No, and be sceptical of any tool that promises that. This score measures technical readiness — whether a model can reach your page, parse it and find something worth quoting. Whether it actually gets cited also depends on your authority, the specific question asked and the model's own ranking, none of which any external tool can see. Think of the score as removing the obstacles, not guaranteeing the outcome.
Which AI crawlers do you check for?
Eight: GPTBot and ChatGPT-User (OpenAI), ClaudeBot and anthropic-ai (Anthropic), PerplexityBot, Google-Extended, CCBot (Common Crawl, which feeds many models) and Bytespider (ByteDance). Only the four assistant crawlers count towards your score — see the next question.
I block CCBot and Google-Extended on purpose. Does that hurt my score?
No. Only GPTBot, ChatGPT-User, ClaudeBot and PerplexityBot affect the status, because those fetch pages to answer live questions — blocking them removes you from the answer. Google-Extended, CCBot and Bytespider are largely about training data, and opting out of training is a legitimate business decision, not a mistake. We show their status for completeness but don't penalise it.
Is llms.txt actually a real standard?
It's a proposal from llmstxt.org, not a ratified standard, and no major provider has publicly committed to reading it. That's exactly why it's weighted at 10 points rather than 30. The file costs you nothing to publish and gives assistants a clean summary of your site if they do look — a small bet with a decent payoff, not a must-have.
Why does my page score badly when it ranks fine on Google?
Classic search and AI search reward different things. Google can rank a page on backlinks and domain authority almost regardless of markup. An AI model working from the page itself has none of that context — it needs explicit structured data and readable text in the page it fetched. A page can rank first on Google and still be nearly unusable as a citation source.
Do you check the whole website or just one page?
One page, plus three domain-wide files: robots.txt, sitemap.xml and llms.txt. Crawling an entire site would take minutes and hammer your server. Start with your homepage for the overall picture, then check individual pages that matter — a product or article page can score very differently from the homepage.
What counts as "enough" content?
We measure readable text in your main content area, deliberately excluding navigation, header and footer — otherwise a page with a big menu and nothing to say would look substantial. Under 500 characters is thin enough that a model has nothing to quote. Above roughly 1,500 you have real substance. These are rules of thumb, not thresholds any provider publishes.
Is my URL stored or shared anywhere?
No. The check runs live and the result exists only in your browser. We don't keep a database of checked URLs, don't log them, and there's no account to create. Refresh the page and the report is gone.
Why did my page fail to load?
Usually one of four reasons: the page blocks automated requests (a firewall or bot protection), it took over five seconds to respond, it needs a login, or the content is rendered client-side so the raw HTML is nearly empty. That last one matters beyond this tool — if we see an empty page, so does an AI crawler.
Can I check a page on localhost or an internal address?
No. The tool only fetches publicly reachable addresses; local and private network addresses are rejected by design. That's a security measure — a server that fetches arbitrary internal addresses on request can be abused to reach things that shouldn't be reachable from the internet.
How often should I re-check?
After any change to your markup, meta tags or robots.txt, and otherwise every few months. The underlying signals change slowly. What changes faster is which crawlers exist — new AI assistants launch with new user agents, so a robots.txt written two years ago may be silently blocking something you'd want to allow.
What should I fix first?
Work top-down by weight. If a crawler is blocked or the page is noindex, fix that first — nothing else matters while a model can't read the page. Then structured data, since it's the difference between a model guessing and knowing. Meta tags and content depth come next. Sitemap and llms.txt are worth doing, but they won't rescue a page that fails the earlier checks.

Related tools

Schema Validator

Validate a live URL or JSON-LD code: SEO score, errors, warnings and rich-results readiness.

llms.txt Generator

Generate a ready-to-use llms.txt file for your website from any URL.

IBAN Validator

Check whether an IBAN is valid, instantly and for free.

FAQ Schema Generator

Generate FAQPage JSON-LD for AI search and structured Q&A data.

Video Schema Generator

Generate VideoObject JSON-LD for video rich results and AI search.

SERP Snippet Preview

Preview your Google snippet and check title & description pixel width.