Technical Audit Criteria

Schema coverage ratio measures what percentage of your indexed pages have relevant JSON-LD. A site with Organization schema on the homepage but no Article schema on blog posts has low coverage. The audit crawls all pages and calculates the ratio — above 80% gets full marks, below 40% scores poorly because most pages are invisible to structured data consumers.

Fact density counts verifiable claims per 1,000 words — named statistics, specific numbers, dated references, attributable statements. The audit uses NLP to identify factual assertions versus opinions or filler. Pages with fact density above 5 claims per 1,000 words are significantly more likely to be cited by AI engines. It's one of the strongest predictors of citation-worthiness.

Content velocity tracks how frequently you publish new or substantially updated content over a rolling 90-day window. The audit measures this through sitemap lastmod timestamps, RSS feed dates, and crawl-detected changes. Sites publishing at least weekly maintain stronger freshness signals than sites publishing monthly or less.

Conditional criteria are checks that only apply to certain site types. Product/Offer schema applies only to e-commerce. Speakable schema only to content-heavy sites. Hreflang only to multilingual sites. ai.txt/TDM policy only to sites with substantial original content. These 4 conditional criteria supplement the 10 universal technical audit criteria.

The audit compares URLs in your sitemap.xml against pages discovered by crawling. It checks for missing URLs, stale lastmod timestamps that don't reflect actual changes, incorrect priority values, and orphaned entries pointing to deleted pages. A complete, accurate sitemap ensures AI crawlers discover every page you want indexed.

Definition pattern detection identifies sentence structures AI systems extract for direct-answer responses. Patterns like "AEO is the practice of..." or "Content velocity refers to..." are structures AI engines pull when answering "What is..." queries. Pages with clear definition patterns appear in direct-answer results at significantly higher rates.

When multiple URLs serve the same content — with/without www, HTTP/HTTPS, trailing slashes — AI crawlers split signals across duplicates. The audit checks every page for a rel=canonical link pointing to the correct URL. Missing or conflicting canonicals confuse crawlers about which version to index, diluting your AI visibility.

The audit verifies your site has a discoverable RSS or Atom feed with proper metadata. It checks feed existence at common paths, a link tag in the HTML head, adequate item count, and pubDate per item. RSS feeds let AI indexing systems detect new content automatically without waiting for a full crawl.

AI engines frequently restructure HTML tables and lists into their answers — but only when the HTML is properly formed. The audit checks that tables use semantic thead, tbody, and th elements, that lists use proper ol/ul markup, and that data tables have clear column headers. Poorly formatted tables get skipped by AI extraction.

The audit looks for machine-readable licensing metadata: CreativeWork license properties in JSON-LD, meta tags indicating reuse policy, copyright notices, ai.txt declarations, and TDM Reservation Protocol headers. Clear licensing tells AI systems whether they can quote or reference your content in their answers.