Structured data audits often fail because markup describes pages, not entities, hindering AI retrieval systems. A common mistake is using a generic "WebPage" type instead of specific entity types like "Organization" or "SoftwareApplication" with clear identifiers. Missing or poorly maintained "sameAs" links also prevent disambiguation, as these links provide crucial context to LLMs about an organization's identity across the web.Client-side injection of JSON-LD, common in modern JavaScript frameworks, renders the markup invisible to most AI crawlers, which do not execute JavaScript. Similarly, content that only appears after JavaScript execution means crawlers receive an empty page, regardless of valid metadata. Markup values must be semantically correct, not just structurally, such as using ISO 8601 for dates and fully qualified URLs for images.Misusing "FAQPage" as a keyword dump or with malformed structures degrades its utility for retrieval systems, even if it no longer provides rich results. Conflicting entity data across pages due to multiple, inconsistent "Organization" blocks also confuses retrieval systems, requiring deduplication for consistent information. To audit effectively, verify body copy without JavaScript, count "application/ld+json" blocks in raw HTML, check "sameAs" URLs, confirm ISO 8601 "dateModified," and ensure a single, consistent "Organization" type. The Rich Results Test primarily checks syntax, not semantic accuracy or crawlability.