AI Answer Readiness Check

Find out whether ChatGPT search, Perplexity, Claude, Gemini, Copilot and Google's AI answers can reach, read and understand your website, and get the exact fixes in plain words. Give it one website or up to 50; each one gets a score out of 100, a grade, what already works, and a numbered list of what to fix first.

What it checks

Area	Share of the score	What is checked
Access and indexing	30	AI crawlers can open the home page (no bot wall or robot check), robots.txt lets search and AI answer crawlers in (Googlebot, Bingbot, OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User, PerplexityBot, Perplexity-User), no leftover noindex, valid HTTPS
Page basics	20	Title, meta description, canonical link on your own domain
Readable content	20	Enough text in the HTML itself (many AI crawlers do not run scripts), clear h1 and h2 headings, questions answered on the page (FAQ)
Structured data	15	schema.org JSON-LD that parses and names what the business and page are
Discovery	15	A real sitemap (not a web page answering at /sitemap.xml) and the basics on up to 30 inner pages from it (or, without a sitemap, linked from the home page)

Also reported as advice, outside the score: /llms.txt, Open Graph tags for shared links, the page language, and whether crawlers that only collect AI training data are blocked (a valid choice that does not hide you from AI answers).

If something cannot be read (a timeout, a page longer than the read limit, a server error), that check is marked could not check and left out of the score instead of counted as a failure. A grade is given only when at least 70% of the score could be checked.

Passing these checks makes a site easy for AI tools to read; it does not guarantee they will cite it.

Input

{
  "urls": ["https://example.com", "another-site.com"],
  "maxPages": 10
}

• urls: home page addresses, 1 to 50. An address without https:// gets it added.
• maxPages: how many inner pages are also checked, taken from the site's sitemap or, without one, from the home page's links (0 to 30, default 10).

Only public websites are checked; private network addresses are refused.

Output

One dataset row per website, plus the full report as Markdown in the key-value store (REPORT-<site>-<n>):

{
  "url": "https://example.com/",
  "status": "checked",
  "score": 40,
  "grade": "Partly ready",
  "coverage": 0.94,
  "fixes": [
    { "id": "structured_data", "title": "Structured data (JSON-LD) describes the business",
      "evidence": "No JSON-LD structured data on the home page.",
      "fix": "Add a <script type=\"application/ld+json\"> block with schema.org data ..." }
  ],
  "checks": [{ "id": "ai_fetch", "state": "pass", "evidence": "A visitor named GPTBot gets the page (200)." }],
  "report": "# AI Answer Readiness: example.com ...",
  "billed": true
}

status is checked, unreachable (the site did not open, answered with an error or showed a robot check page), refused (not a public address) or invalid (not a website address).

Pricing

You pay once per website checked; a site that cannot be reached is free. A site where too little could be read for a grade is free too. When your spending limit for a run is reached, the run stops and lists the sites it did not check.

FAQ

Does a high score mean ChatGPT will recommend my business? No. It means AI tools can find, open and understand your pages. Whether they cite you also depends on what your pages say and on other sites that mention you.

Why does it test as GPTBot? Many sites block AI crawlers by accident through firewall or bot protection settings. Opening the home page under an AI crawler's name shows whether that happens.

Does it change anything on my site? No. It only reads public pages, the same way a crawler does.

Why is llms.txt only advice? It is a young proposal that some AI tools read and others ignore. It is cheap to add, so it is listed, but it does not lower your score.

Can someone fix the list for me? Yes. Each report ends with an offer from the maker of this tool to fix the items for a fixed price.