Improving your AI-readability score

AI-readability is a score we calculate for every audited page that estimates how easily LLMs can extract and cite the page. Different from SEO score – different signals. This page documents each factor and how to move it.

The 8 factors

1. Plain-text content density (weight: 15%)

Ratio of readable text to non-text elements (images, JS widgets, interactive components). LLMs can't read interactive widgets. Pages over 80% interactive content score low. Quick win: surround widgets with plain-text summaries.

2. Heading hierarchy (weight: 15%)

One H1, descriptive H2s for major sections, H3s under those. LLMs use this hierarchy to chunk content. Skipped levels (H1 → H4) lose points. Quick win: audit headings; insert missing H2s.

3. Answer-shaped paragraphs (weight: 15%)

First sentence of each paragraph directly answers an implicit question. We grade this via NLP analysis. Quick win: rewrite section openings to lead with the answer ("X is the technique that..."), not the setup.

4. Schema markup (weight: 10%)

Valid JSON-LD for the page type. Article, FAQ, HowTo, Product, BreadcrumbList all count. Quick win: add FAQ schema to pillar pages with FAQ-shape content (we generate it as a one-click fix).

5. Canonical and metadata clarity (weight: 10%)

Clear title, description, canonical URL, OG tags. Duplicate canonicals or missing meta description hurt. Quick win: run the SEO audit; fix all the on-page metadata issues.

6. Internal linking (weight: 10%)

Pages with 3-7 internal links to related pages with descriptive anchor text score higher than island pages. Quick win: add a "Related articles" section to pillar pages with hand-picked links.

7. Reading level and sentence length (weight: 10%)

Sentences over 30 words and Flesch-Kincaid grade levels over 14 score lower. We don't penalize technical content per se – just unnecessarily complex prose. Quick win: split long sentences. Use a tool like Hemingway to spot the worst.

8. Citation-worthy specifics (weight: 15%)

Specific numbers, dates, named sources, definitions. LLMs prefer to cite concrete claims over vague ones. Quick win: add stats and dated sources to support every important claim ("the 2024 Stanford study (n=1,247) found X reduced Y by 38%").

Per-page detail view

Every audited page has an AI-readability tab next to the SEO tab. The tab shows the overall score, the breakdown by factor with current scores, and the top 5 specific recommendations for that page (e.g. "Insert H2 between line 23 H1 and line 47 H3").

When to expect movement

Score changes are immediate (next audit). Actual LLM citation rates respond slowly – expect 4-8 weeks before consistently-cited new pages show up in AI Visibility prompts.