How We Score Your AI Readiness: Full Methodology

Your AI Readiness Score is a composite metric (0–100) that measures how well your website is prepared to appear in AI-generated answers from ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews. Every point is computed by deterministic code — no AI guesswork, no black box.

1. How Your AI Readiness Score Is Calculated

Your overall score is a weighted average of six categories, plus a small E-E-A-T bonus (up to 5 points):

Content citability carries the highest weight because AI engines need quotable passages. Crawler access is the prerequisite — if bots cannot crawl your site, nothing else matters. Brand authority plays a supporting role since AI citation favors passage quality over domain reputation.

2. The 14 AI Crawlers We Check

We parse your robots.txt file (per RFC 9309) and check access rules for each of these crawlers individually:

Each crawler receives one of five statuses: ALLOWED (explicitly permitted), BLOCKED (explicitly disallowed), PARTIALLY_BLOCKED (some paths restricted), BLOCKED_BY_WILDCARD (caught by a blanket User-agent: * / Disallow: / rule), or NOT_MENTIONED (no specific rule, defaults to allowed).

What You Should Do

Allow at minimum GPTBot, ClaudeBot, PerplexityBot, and Google-Extended — these cover the four most influential AI search platforms. If you have a blanket Disallow: / under User-agent: *, add explicit Allow: / rules for the crawlers you want to reach.

3. Content Citability Scoring

This is the most distinctive part of our analysis. We break your page into content blocks (grouped by heading), then score each block on five metrics that predict whether an AI system will select that passage as a citation.

The Five Metrics

Answer Block Quality (max 30 points) — Does the passage directly answer a question? We detect definition patterns ("X is a..."), early placement of key facts in the first 60 words, question-answer heading pairs, clear sentence structure (5–25 words per sentence), and attribution signals ("according to," "research shows").

Self-Containment (max 25 points) — Can the passage stand alone without context from surrounding paragraphs? We measure passage length (134–167 words is optimal), pronoun density (lower is better — AI strips context), and proper noun density (named entities help AI attribute correctly).

Structural Readability (max 20 points) — Is the passage easy for both humans and machines to parse? We check average sentence length (10–20 words is ideal), transition markers ("first," "additionally," "moreover"), numbered lists, and paragraph breaks.

Statistical Density (max 15 points) — Does the passage contain concrete data? We count percentages, dollar amounts, named quantities ("1,460 marketers"), year references, and citations to recognized sources (Gartner, Forrester, Google, etc.).

Uniqueness Signals (max 10 points) — Does the passage contain original insight? We detect first-party research language ("our research found," "we analyzed"), case study references, and specific tool/methodology mentions.

The Optimal Passage Length: 134–167 Words

Passages in the 134–167 word range are cited most often by AI systems. This length contains a complete answer with evidence, but avoids truncation. Under 80 words rarely has enough substance; over 250 words gets split or summarized, losing attribution.

Grade Distribution

Your overall citability score is the average across all passages, with bonuses for having multiple A/B passages and for maintaining optimal passage lengths.

4. Schema Validation

We parse all <script type="application/ld+json"> blocks on your page and check for the schema types that matter most for AI entity recognition, following the schema.org vocabulary.

Organization — The foundation of entity identity. Must include name, url, logo, and sameAs (links to social profiles, Wikipedia, Wikidata). Missing sameAs is the most common gap — it is how AI connects your site to your broader online presence.

Article / BlogPosting / NewsArticle — Signals authored content with a publication date. Must include headline, datePublished, and author. The optional speakable property (per Google Search Central) marks sections suitable for AI reading.

FAQPage — Directly maps question-answer pairs for AI extraction. One of the highest-impact schema types for AI visibility.

BreadcrumbList — Helps AI understand site hierarchy and generate accurate source attributions.

knowsAbout — An underused Organization/Person property that declares your expertise areas. AI systems use this for topical authority.

WebSite — Enables sitelinks search box and helps AI treat your site as a unified entity.

5. Technical Appendix

This section documents the exact scoring formulas implemented in analyzer.py. All scores are deterministic — running the same page twice will produce identical results.

Overall Score Formula

overall = (crawler_score * 0.20) + (schema_score * 0.10) + (citability_score * 0.25)
        + (tech_score * 0.15) + (llm_score * 0.15) + (brand_score * 0.10)
        + min(experience_signals * 1.5 + expertise_signals * 1.0, 5)

Result is clamped to [0, 100].

Category 1: AI Crawler Access (weight 20%)

• No robots.txt exists: base score = 60 (accessible but not ideal).
• robots.txt exists: allowed_ratio = 1 - (blocked + partial * 0.5) / total_crawlers. Base = allowed_ratio * 70.
• Bonuses: robots.txt exists (+10), sitemap referenced in robots (+10), llms.txt exists (+10), llms.txt valid format (+5).
• Penalty: -10 for each blocked key crawler (GPTBot, ClaudeBot, PerplexityBot, Google-Extended).
• llms.txt validation follows the llmstxt.org spec: must have # Title, > Description, ## Section headings, and - [Link](url) entries.

Category 2: Structured Data & Schema (weight 10%)

Points are additive: Organization (+15), sameAs links 5+ (+15) / 2+ (+8) / 1+ (+4), Article (+10), FAQ (+8), Product (+8), BreadcrumbList (+5), WebSite (+5), speakable (+5), any JSON-LD present (+5). If no JSON-LD exists at all, score is forced to 0. Penalties: -5 per schema issue (up to -10 for 5+ issues), -5 if Organization exists but lacks knowsAbout.

Category 3: Content AI-Citability (weight 25%)

Base = average passage score across all content blocks. Bonuses: 3+ optimal-length passages (+10) or 1+ (+5); 3+ grade A/B passages (+10) or 1+ (+5); 3+ question headings (+5) or 1+ (+2). Penalty: word count under 300 (-20) or under 500 (-10).

Passage scoring (0–100 per block):

Category 4: Technical SEO Foundations (weight 15%)

Points: title (+8), meta description (+8), canonical (+5), viewport (+5), complete OG tags (+8), sitemap with lastmod (+8), semantic HTML (+5), <main> tag (+3), single H1 (+5), valid hierarchy (+3), image alt ratio (up to +5), SSR detected (+10), security headers (up to +8), lang (+2). Security: HTTPS (+4), HSTS (+2), CSP/X-Content-Type/X-Frame-Options/Referrer-Policy (+1 each).

Category 5: LLM Discoverability (weight 15%)

Points: llms.txt exists (+20), valid format (+15), llms-full.txt (+5), FAQ schema (+10), 5+ question headings (+10) / 2+ (+5), 3+ high answer-quality passages (+10) / 1+ (+5), Organization + meta description (+10), sitemap (+5), 20+ internal links (+5) / 10+ (+3), 1500+ words (+5) / 800+ (+3).

Category 6: Brand & Authority Signals (weight 10%)

Social platforms (capped at 40 total): YouTube (+15), Reddit (+12), LinkedIn (+10), Twitter/X (+8), GitHub (+8), Facebook (+5), Instagram (+5), TikTok (+3). YouTube carries the highest weight per the Ahrefs brand authority study (r = 0.737). Additional: Wikipedia (+15), Wikidata (+10), About/Contact pages (+5 each), author info (+5), sameAs 5+ (+10) / 2+ (+5), testimonials (+5), privacy policy (+3), terms (+2).

Edge Cases

• SPA / client-side rendering: Near-empty #app, #root, __next, or __nuxt containers (< 50 chars) flag SSR failure. AI crawlers generally do not execute JavaScript.
• Wildcard robots.txt blocks: User-agent: * / Disallow: / blocks all crawlers including AI bots, flagged separately from per-crawler blocks.
• No robots.txt: Scored as permissive (60) per RFC 9309 Section 2.3 — absence means all crawlers allowed.
• Schema in @graph: CMS platforms (WordPress/Yoast) nest schema in @graph arrays. We parse these recursively.
• Minimal content pages: Under 300 words receives a -20 citability penalty.

References

• Google Search Central: Structured Data — Schema markup guidelines and supported types.
• web.dev: Technical SEO — Semantic HTML and meta tag best practices.
• schema.org — Full vocabulary reference for Organization, Article, FAQPage, speakable, knowsAbout.
• llmstxt.org — Specification for the llms.txt standard.
• RFC 9309: Robots Exclusion Protocol — The authoritative standard for robots.txt parsing.
• GEO: Generative Engine Optimization — Princeton/Georgia Tech/IIT Delhi research on optimizing content for AI search engines.