---
title: "What the AI Ready Check measures: bot access, discoverability and content | Mettle GEO Blog"
description: "A plain-language guide to every check in the free AI Ready Check: whether AI bots can reach your site, find their way around it and read what is there, why each one matters, and what to fix when it fails."
url: "https://mettlegeo.com/blog/ai-ready-check-what-we-measure"
image: "https://mettlegeo.com/og-card.png"
---
1.  [Home](/)
2.  [Blog](/blog)
3.  What the AI Ready Check measures: bot access, discoverability and content

Reader guide

### Turn this article into action

**Understand** **Apply** **Ask AI**

Plain-English point

A plain-language guide to every check in the free AI Ready Check: whether AI bots can reach your site, find their way around it and read what is there, why each one matters, and what to fix when it fails.

Best use

Use this article when you are working on technical foundations and need a concrete GEO next step.

AI ready check is my site ready for AI search can AI bots crawl my site GPTBot ClaudeBot robots.txt Content Signals robots.txt

**TL;DR:** The [AI Ready Check](/ai-ready-check) asks three questions about your site. Can AI bots reach it? Can they find their way around it? Can they understand what they read? Each question is a group of checks. Every check either gives you a clear result or is marked **Unverified**, and Unverified checks never count toward your score. This post explains each group, why it matters and what to do when a check fails.

## Why a site can be invisible to AI answers

ChatGPT, Gemini and Claude build their answers from pages they can fetch and parse. If a bot is blocked, gets lost, or receives an empty page shell, your content does not exist for that answer, however good it is.

Most sites never decided to block AI. It happens by accident: a firewall rule copied from a template, a robots.txt line written years ago, or a page that only renders its text after JavaScript runs. The AI Ready Check is a quick way to spot those accidents.

Note:

The check looks at your site from the outside, the same way a bot does. It does not log in, it does not see your server logs, and it cannot certify that any specific AI provider is or is not using your content. When something cannot be confirmed from the outside, we say so instead of guessing.

## Bot access: can AI bots actually reach your site?

This is the first gate. If it is closed, nothing else on the page matters.

### AI bot rules in robots.txt

We read your robots.txt and look for rules that name the major AI crawlers, such as GPTBot (OpenAI) and ClaudeBot (Anthropic). Blocking some of them can be a deliberate choice. For example, you may allow search assistants but not model training. What we flag is a block nobody meant to set.

**What to do:** decide your policy for each bot, then write it down explicitly. A clear `Allow` or `Disallow` per bot is better than relying on a catch-all rule.

### Real access test

robots.txt is a request, not a lock. Firewalls, CDNs and bot protection can still refuse a bot that robots.txt allows. So we also request your home page while identifying as each AI bot and check whether the answer is a normal page or a block, challenge or error.

**What to do:** if the rules say "allowed" but the test gets refused, look at your CDN or firewall bot settings. When the two disagree, nobody on your team usually knows, because nothing visibly breaks for human visitors.

### Content Signals

[Content Signals](https://contentsignals.org/) is a newer way to state in robots.txt how your content may be used: for search, as input to AI answers, or for training. We check whether you declare them and whether the values are recognizable.

**What to do:** add a line such as `Content-Signal: search=yes, ai-input=yes, ai-train=no` that matches your actual policy. Declaring a preference does not force anyone to follow it, but it puts your intent on record in a machine-readable place.

### Web Bot Auth

Web Bot Auth lets bots sign their requests so a site can tell a real AI crawler from an impostor using its name. Checking the signatures happens inside your server or CDN, so an outside scan cannot see whether you validate them. We show this check for awareness and mark it **Unverified**. It never affects your score.

**What to do:** if you use a CDN with verified-bot features, check whether signed-request validation is available and switched on.

### WebMCP

WebMCP is an early browser proposal that lets a website offer tools, such as "search products" or "get pricing", directly to an AI agent running in the browser. We look for WebMCP code in the scripts your site serves. A match means the code is there. It does not prove the tools work, because we do not run your site in a browser.

**What to do:** nothing urgent for most sites. If agents booking, buying or searching on your site matters to you, it is worth following.

### MCP server

If you run a product or a service with an API, a public MCP server lets assistants like Claude or ChatGPT work with it directly. We look for one at the usual addresses and in the places sites commonly announce it. If your server requires sign-in and announces itself correctly, that counts as found.

**What to do:** if you have an MCP server, mention its address in your llms.txt and documentation so assistants and people can find it. If you do not run a product with an API, you can ignore this check.

## Discoverability: can AI find and navigate your content?

Once a bot can reach your site, it needs a map. These three files are that map.

### robots.txt

We check that a readable robots.txt exists and that it does not block your home page. A missing or broken file leaves every crawler guessing.

**What to do:** serve a plain-text robots.txt at the root of your domain, keep it short, and review it whenever you change bot policy.

### sitemap.xml

A sitemap lists the pages you want found. We look for one referenced in robots.txt or at the standard address and check that it is valid XML with real page addresses.

**What to do:** most CMS platforms generate a sitemap automatically. Make sure it is switched on, up to date, and listed in robots.txt with a `Sitemap:` line.

### llms.txt

llms.txt is a short Markdown file at the root of your site that describes what you do and links to your most important pages. It is a proposal, not a standard. We check that it exists and has the expected shape: a heading and a list of links.

Being honest about llms.txt

No major AI provider has publicly committed to reading llms.txt. We include it because it is cheap to add, it helps AI agents and coding assistants that do use it, and it forces you to decide which pages matter most. Do not expect it to change your visibility on its own.

**What to do:** write a short llms.txt with one paragraph about your business and links to your key pages, such as products, pricing, documentation and contact.

## Content: can AI understand what it reads?

Reaching and finding a page is not enough. The bot has to get the text and understand what it is about.

### Text readability without JavaScript

Many AI crawlers do not run JavaScript, or run very little of it. If your page sends an empty shell and builds its text in the browser, those bots may see almost nothing. We count how much meaningful text is in the HTML your server sends before any script runs.

**What to do:** use server-side rendering or prerendering for pages you want cited. Your main heading, introduction and key facts should be in the first HTML response.

### JSON-LD structured data

JSON-LD is a block of schema.org data that states plainly what a page is: an organization, a product, an article, an FAQ. We check that it is present and valid. This is a structural check, not a full rich-results validation. For a deeper guide, see [schema markup for AI](/blog/schema-markup-for-ai).

**What to do:** add `Organization` data on your home page and the matching type on your key pages. Only mark up what is actually visible on the page.

### Content freshness

AI assistants tend to prefer current information, and bots have no reliable way to know when a page changed unless you tell them. We look for machine-readable dates on your home page, such as `dateModified` in structured data or a `<time>` element.

**What to do:** show accurate published and updated dates, and update them only when the content really changes. A fake "updated today" stamp does more harm than good.

## How the score works

| Result | What it means |
| --- | --- |
| Pass | The check confirmed what it looks for. |
| Warning | Something is present but incomplete, or set up in a way worth reviewing. |
| Fail | What the check looks for is missing or blocked. |
| Unverified | It could not be confirmed from the outside. Shown, never scored. |

Your overall score and each category score only count checks we could actually confirm. A check we could not see into never lowers your score, and it never raises it either.

## What the check does not tell you

A perfect score means AI bots can reach, find and read your site. It does not mean AI answers will mention you. That depends on what you publish, who else covers your topic and which sources each assistant trusts.

The AI Ready Check covers the technical foundation. To see whether ChatGPT Search and Gemini actually name your brand for the questions your customers ask, and what to change when they do not, that is what [Mettle](/ai-search-visibility-tool) tracks.

[Run the free AI Ready Check on your site](/ai-ready-check), fix what it finds, then measure whether AI answers start to include you.

## Questions this article answers

Does a high AI Ready score mean ChatGPT will mention my brand?

No. The score means AI bots can reach, find and read your site. Whether answers mention you depends on your content, your competitors and which sources each assistant trusts. That needs prompt-level tracking.

Why are some checks marked Unverified?

Some things cannot be confirmed from outside your site, for example whether your CDN validates signed bot requests. We show those checks for awareness but never count them toward your score.

Is blocking AI training crawlers bad for my score?

Not necessarily. Blocking training can be a deliberate policy. The check flags blocks that look unintentional and asks you to state your rules explicitly for each AI bot.

Do AI providers read llms.txt?

No major AI provider has publicly committed to it. It is cheap to add and useful for agents and coding assistants that do read it, but it will not change your visibility on its own.

## Keep reading

[

Technical Foundations 6 min read

### Should I block GPTBot, ClaudeBot and PerplexityBot in robots.txt?

](/blog/should-i-block-ai-crawlers-robots-txt)[

Technical Foundations 8 min read

### The schema markup checklist every B2B SaaS site needs in 2026

](/blog/schema-markup-for-ai)[

GEO Basics 7 min read

### Why do ChatGPT answers about my brand change every time?

](/blog/why-chatgpt-answers-change)

## See what AI engines say about your site

Run a Mettle audit to check whether ChatGPT, Gemini and Claude can read your pages, and where competitors get cited instead.

[Run an audit](/tools/website-readability)

## Structured Data
```json
[
  {
    "@context": "https://schema.org",
    "@type": "Organization",
    "name": "Mettle",
    "alternateName": [
      "MettleGEO",
      "Mettle AI",
      "Mettle GEO"
    ],
    "url": "https://mettlegeo.com",
    "logo": "https://mettlegeo.com/logo-only-white.png",
    "description": "Mettle helps teams monitor AI search visibility across ChatGPT Search and Gemini, track citations, and improve GEO performance.",
    "address": {
      "@type": "PostalAddress",
      "addressLocality": "Sofia",
      "addressCountry": "BG"
    },
    "contactPoint": {
      "@type": "ContactPoint",
      "email": "solutions@mettlegeo.com",
      "contactType": "customer support"
    },
    "sameAs": [
      "https://www.linkedin.com/company/mettle-geo",
      "https://www.facebook.com/profile.php?id=61592024245195",
      "https://www.instagram.com/mettlegeo/"
    ]
  },
  {
    "@context": "https://schema.org",
    "@type": "BlogPosting",
    "headline": "What the AI Ready Check measures: bot access, discoverability and content",
    "description": "A plain-language guide to every check in the free AI Ready Check: whether AI bots can reach your site, find their way around it and read what is there, why each one matters, and what to fix when it fails.",
    "image": [
      "https://mettlegeo.com/logo-only-white.png"
    ],
    "datePublished": "2026-10-28T10:00:00Z",
    "dateModified": "2026-10-28T10:00:00Z",
    "author": {
      "@type": "Organization",
      "name": "Mettle Editorial"
    },
    "publisher": {
      "@type": "Organization",
      "name": "Mettle",
      "url": "https://mettlegeo.com",
      "logo": {
        "@type": "ImageObject",
        "url": "https://mettlegeo.com/logo-only-white.png"
      }
    },
    "mainEntityOfPage": {
      "@type": "WebPage",
      "@id": "https://mettlegeo.com/blog/ai-ready-check-what-we-measure"
    },
    "@id": "https://mettlegeo.com/blog/ai-ready-check-what-we-measure#article",
    "articleSection": "Guide",
    "keywords": [
      "AI ready check",
      "is my site ready for AI search",
      "can AI bots crawl my site",
      "GPTBot ClaudeBot robots.txt",
      "Content Signals robots.txt",
      "llms.txt",
      "WebMCP",
      "MCP server discovery",
      "JSON-LD for AI",
      "JavaScript rendering AI crawlers"
    ],
    "about": [
      {
        "@type": "Thing",
        "name": "AI ready check"
      },
      {
        "@type": "Thing",
        "name": "is my site ready for AI search"
      },
      {
        "@type": "Thing",
        "name": "can AI bots crawl my site"
      },
      {
        "@type": "Thing",
        "name": "GPTBot ClaudeBot robots.txt"
      },
      {
        "@type": "Thing",
        "name": "Content Signals robots.txt"
      },
      {
        "@type": "Thing",
        "name": "llms.txt"
      },
      {
        "@type": "Thing",
        "name": "WebMCP"
      },
      {
        "@type": "Thing",
        "name": "MCP server discovery"
      },
      {
        "@type": "Thing",
        "name": "JSON-LD for AI"
      },
      {
        "@type": "Thing",
        "name": "JavaScript rendering AI crawlers"
      }
    ],
    "mentions": [
      {
        "@type": "Thing",
        "name": "AI ready check"
      },
      {
        "@type": "Thing",
        "name": "is my site ready for AI search"
      },
      {
        "@type": "Thing",
        "name": "can AI bots crawl my site"
      },
      {
        "@type": "Thing",
        "name": "GPTBot ClaudeBot robots.txt"
      },
      {
        "@type": "Thing",
        "name": "Content Signals robots.txt"
      },
      {
        "@type": "Thing",
        "name": "llms.txt"
      },
      {
        "@type": "Thing",
        "name": "WebMCP"
      },
      {
        "@type": "Thing",
        "name": "MCP server discovery"
      }
    ],
    "audience": {
      "@type": "Audience",
      "audienceType": "Marketing, SEO, revenue, and founder teams working on AI search visibility"
    },
    "learningResourceType": "Guide",
    "isAccessibleForFree": true,
    "potentialAction": {
      "@type": "ReadAction",
      "target": "https://mettlegeo.com/blog/ai-ready-check-what-we-measure"
    }
  },
  {
    "@context": "https://schema.org",
    "@type": "ItemList",
    "name": "Related Technical Foundations articles",
    "itemListElement": [
      {
        "@type": "ListItem",
        "position": 1,
        "name": "Should I block GPTBot, ClaudeBot and PerplexityBot in robots.txt?",
        "url": "https://mettlegeo.com/blog/should-i-block-ai-crawlers-robots-txt"
      },
      {
        "@type": "ListItem",
        "position": 2,
        "name": "The schema markup checklist every B2B SaaS site needs in 2026",
        "url": "https://mettlegeo.com/blog/schema-markup-for-ai"
      }
    ]
  },
  {
    "@context": "https://schema.org",
    "@type": "FAQPage",
    "mainEntity": [
      {
        "@type": "Question",
        "name": "Does a high AI Ready score mean ChatGPT will mention my brand?",
        "acceptedAnswer": {
          "@type": "Answer",
          "text": "No. The score means AI bots can reach, find and read your site. Whether answers mention you depends on your content, your competitors and which sources each assistant trusts. That needs prompt-level tracking."
        }
      },
      {
        "@type": "Question",
        "name": "Why are some checks marked Unverified?",
        "acceptedAnswer": {
          "@type": "Answer",
          "text": "Some things cannot be confirmed from outside your site, for example whether your CDN validates signed bot requests. We show those checks for awareness but never count them toward your score."
        }
      },
      {
        "@type": "Question",
        "name": "Is blocking AI training crawlers bad for my score?",
        "acceptedAnswer": {
          "@type": "Answer",
          "text": "Not necessarily. Blocking training can be a deliberate policy. The check flags blocks that look unintentional and asks you to state your rules explicitly for each AI bot."
        }
      },
      {
        "@type": "Question",
        "name": "Do AI providers read llms.txt?",
        "acceptedAnswer": {
          "@type": "Answer",
          "text": "No major AI provider has publicly committed to it. It is cheap to add and useful for agents and coding assistants that do read it, but it will not change your visibility on its own."
        }
      }
    ]
  },
  {
    "@context": "https://schema.org",
    "@type": "BreadcrumbList",
    "itemListElement": [
      {
        "@type": "ListItem",
        "position": 1,
        "name": "Home",
        "item": "https://mettlegeo.com"
      },
      {
        "@type": "ListItem",
        "position": 2,
        "name": "Blog",
        "item": "https://mettlegeo.com/blog"
      },
      {
        "@type": "ListItem",
        "position": 3,
        "name": "What the AI Ready Check measures: bot access, discoverability and content",
        "item": "https://mettlegeo.com/blog/ai-ready-check-what-we-measure"
      }
    ]
  }
]
```
