Methodology

Readout diagnoses machine-legibility, not AI-ranking. Every check below maps to a published guideline, an observed failure pattern, or a clearly labeled heuristic. Nothing here predicts citations.

CheckSeverityBasis
Claim not visible on page (content-mismatch) high Google: do not add structured data for content not visible to users. source
Missing required properties (rich-result-blocked) high Google structured-data type requirements; absence blocks the named rich result. source
Broken sameAs identity link high A sameAs URL returning 404 or 410 is flagged as definitively dead. 403, 5xx, or unreachable (0) are not flagged because they are usually bot-blocks or transient failures, not truly broken links. sameAs only helps when it points to an authoritative representation of the same entity; it is not a trust switch. source
Identity conflict (same type, different name / dangling @id) medium Contradictory identity fields prevent machines from merging references into one entity. Observed-pattern heuristic.
Missing provenance (author/publisher/datePublished on Article) medium Google Article guidance recommends author, datePublished, publisher for attribution of factual claims. source
Thin entity (missing recommended properties) medium Valid but under-described; recommended properties per the type's Google documentation. source
Identity not anchored low Labeled heuristic: an entity declares sameAs links but none point to a closed allowlist of identity authorities (Wikipedia, Wikidata, LinkedIn, ...).

No single score is produced. Findings are a decomposable tally. Readout does not query AI engines, track mentions, or persist anything you submit.