Understanding Your AI-Friendliness Score

What we check in each parameter and why it matters for AI visibility

What the score measures: How well your profile is structured for AI-powered search and discovery when attorneys use AI tools to research expert witnesses. Each parameter targets a specific structural signal that AI retrieval systems rely on. Higher score = stronger AI-readiness signals = better positioned for AI-sourced discovery.

❓ About This Audit

How the audit works and what the scores actually represent

▶ Does this audit actually test ChatGPT, Claude, or Gemini directly?

No — and we want to be transparent about why. No public API exists that lets a third-party tool query whether your specific website would be cited by ChatGPT or Claude in response to a given prompt. Any tool claiming to directly measure that is making an unverifiable claim.

What our audit does is test whether your site meets the structured content, semantic markup, and metadata standards that AI retrieval systems are designed to favor. Think of it as testing whether your site speaks the same language as AI — not whether any specific AI has read it yet.

What the score actually represents: These checks — schema markup, heading hierarchy, content clarity, freshness signals — are the same signals that search engineers, RAG developers, and AI platform architects document as best practices for machine-readable content. A higher score means your site is better prepared for AI discovery. It does not guarantee citation by any specific model.

🎯 Section A — AI Standards & Discoverability

Weight: 20% of total score | Three parameters

▶ A1 — llms.txt File

40% of Section A

What we check: Whether your site has a file at /llms.txt that is non-empty. We make a direct HTTP request to that path and check the response.

Scoring: 100/100 if the file exists and contains content. 0/100 if it's missing or empty. No partial credit.

Why it matters: The llms.txt standard is an emerging convention that signals to AI crawlers how to interpret your site. Adoption is growing — implementing it now positions your site ahead of the standard becoming widely required. Without it, AI crawlers must infer your site's purpose and content priority without guidance.

▶ A2 — Sitemap Freshness

30% of Section A

What we check: We fetch your sitemap.xml and look for at least one <lastmod> date within the past 90 days. We parse the XML directly.

Condition	Score
At least one URL updated within 90 days	100
Sitemap exists but all dates are older than 90 days	60
No sitemap found or malformed XML	0

Why it matters: AI systems treat freshness as a proxy for current expertise. A sitemap with recent dates signals active maintenance and current knowledge. Stale or missing dates lead AI to deprioritize your content in favor of more recently updated sources.

▶ A3 — URL Semantics

30% of Section A

What we check: We analyze the path of the URL being audited by splitting it on hyphens and underscores and counting meaningful words. The homepage (/) automatically passes.

Condition	Score
Homepage path (`/`) or 2–5 meaningful words in path	100
1 word or more than 6 words in path	60
Numeric IDs only, no readable words	0

Why it matters: AI systems read URL paths as a context signal before processing page content. A URL like /expert-witness-medical-malpractice immediately establishes topic context, reducing the chance of misclassification. Numeric or symbol-only URLs provide no context signal at all.

🏗️ Section B — AI Content Architecture

Weight: 35% of total score — the highest-weighted section | Four parameters
Measures whether your site's content structure meets the formatting standards used by AI retrieval systems to locate and extract relevant expertise.

▶ B1 — Semantic Heading Hierarchy (H-Tags)

30% of Section B

What we check: We scan all heading tags (H1–H6) and evaluate: whether there is exactly one H1, whether levels are skipped (e.g. H1 jumping directly to H4), and whether any headings are empty.

Condition	Score
Single H1, logical nesting, no empty tags	100
Minor level skip (e.g. H1→H3)	60
Multiple H1s, major skips (H1→H4+), or excessive empty headings	20
No headings found	0

Why it matters: This creates parseable document structure for AI content extraction. A proper hierarchy allows retrieval systems to identify which credentials belong to which specialty and which case types relate to which expertise area. Broken hierarchy forces AI parsers to guess relationships between content blocks, increasing the risk of inaccurate citations.

▶ B2 — RAG Anchors (ID Attributes on Headings)

25% of Section B

What we check: We count all H2 and H3 tags and calculate what percentage have an id attribute. Only H2 and H3 are checked (section-level headings).

Condition	Score
More than 80% of H2/H3 tags have id attributes	100
20–80% of H2/H3 tags have id attributes	60
Fewer than 20% have id attributes	0

Why it matters: This enables section-level retrieval in AI-augmented research tools. Rather than referencing your page as a whole, retrieval systems can link directly to the specific section covering a particular expertise area — making citations more precise and credible to the attorney reading them.

▶ B3 — Text-to-Code Ratio

25% of Section B

What we check: We measure the ratio of readable text to total HTML size. Scripts, style blocks, SVGs, and noscript tags are removed before calculating the text content length against the raw page size.

Condition	Score
Text content is more than 15% of total page size	100
Text content is 5–15% of total page size	60
Text content is under 5%	20

Why it matters: This ensures content is accessible to crawlers and AI parsers. A page bloated with JavaScript, embedded styles, or inline SVGs uses up the parser's processing budget with code rather than content. The less space wasted on markup, the more of your actual expertise retrieval systems can read and index in a single pass.

▶ B4 — Structured Formatting (Lists & Tables)

20% of Section B

What we check: We count the number of <ul>, <ol>, and <table> elements in the main content area of the page.

Condition	Score
3 or more lists/tables found	100
1–2 lists/tables found	60
No lists or tables found	20

Why it matters: Structured content that retrieval systems can segment and cite produces far more accurate results than parsing prose. Presenting credentials as a bullet list or table significantly reduces the chance of a retrieval system misreading or omitting a qualification when surfacing your profile.

🆔 Section C — Entity Clarity & Trust

Weight: 30% of total score | Three parameters
AI systems need to confirm who you are before confidently recommending you.

▶ C1 — Person / Profile Schema

40% of Section C

What we check: We look for JSON-LD <script> blocks and inspect every schema object, including those nested inside @graph arrays. We check the @type of each object.

Condition	Score
`Person`, `Expert`, or `ProfilePage` schema with `name` + `jobTitle` or `description`	100
`LocalBusiness` or `Organization` schema with `name`	60
No qualifying schema found	0

Why it matters: Structured schema is your AI identity card. Without it, AI systems must infer who you are from unstructured page text — a process prone to errors and omissions. A Person schema with a job title tells AI-powered search and discovery tools exactly who you are and what you do, enabling confident, accurate surfacing of your profile without guessing.

▶ C2 — Knowledge Graph Linking (sameAs)

30% of Section C

What we check: We scan all schema objects (including @graph arrays) for a sameAs property containing at least one valid HTTP/HTTPS URL.

Condition	Score
`sameAs` found with one or more valid URLs	100
No `sameAs` property found in any schema	0

Why it matters: sameAs links tell AI systems "this entity is also described at these authoritative sources" — typically LinkedIn, a university profile, or a professional association page. This cross-reference allows AI to verify your identity against third-party sources before recommending you, significantly increasing citation confidence. Without it, your profile is self-asserted with no external confirmation.

▶ C3 — Content Date Signals

30% of Section C

What we check: We look for date signals in three places: datePublished or dateModified in any schema object (including @graph), an article:published_time meta tag, or a visible date pattern in page text matching "Published/Modified/Updated: [date]".

Condition	Score
datePublished/dateModified in schema or article:published_time meta tag	100
Visible date text matching "Updated: [date]" pattern on page	60
No date signals found anywhere	0

Why it matters: AI systems use date signals to assess whether information is current. A profile with no date could be from any year — AI may treat it as potentially outdated and downweight it in recommendations. Explicit date signals confirm your expertise is actively maintained, which directly affects how confidently AI systems cite your profile for recent case enquiries.

📖 Section D — Readability

Weight: 15% of total score | Two parameters
How efficiently AI can parse and extract information from your content.

▶ D1 — Paragraph Chunking

50% of Section D

What we check: We find all <p> tags with more than 50 characters of text and calculate the average character length across those paragraphs.

Condition	Score
Average paragraph under 400 characters (~3–4 sentences)	100
Average paragraph 400–800 characters	60
Average paragraph over 800 characters	20

Why it matters: AI systems split content into chunks when indexing it for retrieval. Long paragraphs that cover multiple ideas become a single chunk — AI either retrieves the whole block or none of it. Shorter, focused paragraphs each covering one idea are chunked cleanly, making specific facts easier to retrieve and cite accurately.

▶ D2 — Sentence Complexity

50% of Section D

What we check: We extract all visible text, split it into sentences by punctuation, filter to sentences with 3 or more words, and calculate the average number of words per sentence.

Condition	Score
Average sentence under 20 words	100
Average sentence 20–30 words	60
Average sentence over 30 words	20

Why it matters: Long, complex sentences with multiple clauses force AI to resolve ambiguity about what the subject is and which qualifier applies to which claim. Concise sentences reduce parsing errors and lower the chance of AI hallucinating or misattributing a credential. This also correlates with clearer communication for human readers.

See How Your Site Scores

Run the free audit on your website — results in under 60 seconds.

Run Your Free Audit