Introducing the AI Readability Score: Is Your Site Ready for LLMs?

Search is changing. Google is still important, but an increasing share of traffic now comes from AI systems — ChatGPT, Claude, Perplexity, and countless RAG pipelines pulling content from the web.

The question isn't just "can Google find me?" anymore. It's: can an LLM actually understand and cite my page?

Today we're launching the AI Readability Score — a new 0-100 metric that answers exactly that. It's available as a standalone API endpoint and included automatically in every SEO audit.

What It Measures

The AI Readability Score evaluates your page across four categories, each targeting a different aspect of how AI systems consume content:

Crawl Accessibility (25 points)

Can AI bots actually reach and read your page?

• Server-side rendering — Is content available without JavaScript execution? Most AI crawlers don't run JS.
• robots.txt AI bot access — Are GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and CCBot allowed?
• llms.txt — Do you have a /llms.txt file providing AI-friendly site navigation?
• HTTPS, response time, canonical URL — The basics that affect all crawlers.

Structural Markup (25 points)

Can AI systems parse your content structure?

• Schema.org JSON-LD — Structured data with valid @type gives LLMs explicit metadata about your content.
• Semantic HTML — <article>, <section>, <main> vs div-soup. AI systems use these signals to identify content boundaries.
• Heading hierarchy — Proper H1/H2/H3 nesting helps LLMs understand content organization.
• Lists, tables, image alt text, meta description — Structured content elements that aid extraction.

Content Extractability (30 points)

This is the biggest category because it's the hardest to get right.

• Answer-first format — Do your sections lead with the key facts? LLMs pulling snippets for RAG will grab the first 1-2 sentences of each section.
• Chunk-friendly sections — Sections between 50-200 words are ideal for RAG chunking. Too long and they get split mid-thought; too short and they lack context.
• Fact density — Numbers, percentages, dates, and proper nouns per 100 words. Dense factual content gets cited more.
• Flesch Reading Ease — Simpler language is easier for LLMs to parse and summarize accurately.
• Content-to-boilerplate ratio — How much of your page is actual content vs navigation, footers, and sidebars?
• Freshness signals — dateModified in JSON-LD, article:modified_time meta tags, Last-Modified headers.
• Source citations — Outbound links in your content area signal well-researched content.

AI Discoverability (20 points)

Can AI systems find and contextualize your page?

• Open Graph completeness — AI systems use OG tags for content previews and context.
• Semantic URLs — Readable path segments vs numeric IDs or hashes.
• Sitemap presence — Is this page listed in your sitemap.xml?
• Internal linking — More internal links help AI crawlers discover related content.
• Author and E-E-A-T signals — Author metadata in JSON-LD, meta tags, or byline elements.
• Multi-format availability — RSS/Atom/JSON feeds for alternative content access.

Real Example: aiopscrew.com

We ran the AI Readability Score against aiopscrew.com to see how a well-optimized technical site performs:

The SEO score is an excellent 95/100 (A+), but the AI Readability score tells a different story. Crawl accessibility is perfect — the site has an llms.txt file, allows all AI bots, uses HTTPS, and has server-rendered content with fast response times.

Where it loses points is content extractability. The Flesch Reading Ease score is 23 (quite technical), zero sections hit the ideal 50-200 word range for RAG chunking, and there are no outbound citation links. These are things that don't hurt traditional SEO but matter for AI consumption.

How to Use It

Standalone endpoint

curl "https://seoscoreapi.com/ai-readability?url=https://example.com" \
  -H "X-API-Key: YOUR_KEY"

Returns the full breakdown — score, grade, four categories with individual checks, and prioritized recommendations.

Included in audits

Every /audit and /demo-audit response now includes an ai_readability field alongside the existing SEO data:

{
  "score": 95.006,
  "grade": "A+",
  "audit": { ... },
  "ai_readability": {
    "ai_readability_score": 78,
    "ai_readability_grade": "B",
    "categories": { ... },
    "recommendations": [ ... ]
  }
}

Demo (no API key needed)

curl "https://seoscoreapi.com/demo-ai-readability?url=https://example.com"

Quick Wins to Improve Your Score

1. Add /llms.txt — A simple text file describing your site for AI crawlers. Free 4 points.
2. Don't block AI bots — Check your robots.txt for GPTBot, ClaudeBot, and CCBot disallow rules.
3. Add JSON-LD structured data — Article, WebPage, FAQPage schemas give LLMs explicit content metadata. Worth 8 points.
4. Use semantic HTML — Replace <div> soup with <article>, <section>, <main>. Helps AI systems find the actual content.
5. Lead with facts — Put the key takeaway in the first sentence of each section. This is what RAG systems extract.
6. Break content into 50-200 word sections — Use headings to create natural chunk boundaries for retrieval systems.
7. Add dateModified — In your JSON-LD or as article:modified_time meta tag. Freshness matters for AI citation.

SEO Score vs AI Readability Score

These are intentionally separate metrics. A page can score A+ on SEO and C on AI Readability (or vice versa). Traditional SEO optimizes for search engine ranking factors. AI Readability optimizes for content extraction and citation by LLM systems.

Both matter. We report both. Your existing SEO scores and weights are completely unchanged — AI Readability is additive, not a replacement.

Try it on your site and see where you stand.