Schema and Entity Audits: Check Structured Data Against Page Facts
Audit structured data for valid syntax, consistent entities, current facts, correct URLs, and agreement with the visible page.
The short answer
Audit schema by checking syntax and relevant requirements, then comparing every important claim with the visible page and approved product facts. Review entity names, identifiers, URLs, dates, offers, and relationships. Use appropriate types for the actual content. A syntactically valid JSON-LD block can still describe the wrong thing or contain misleading information. Structured data does not guarantee a rich result or an AI citation.
Inventory what the page declares
Collect structured data from the initial response and, when necessary, rendered content. List the declared types and the template or plugin that produced them. A page may include duplicate organization or article entities from several sources. Preserve the existing output before recommending a change so the reviewer can see the exact difference.
Identify the main content and choose applicable types from current official guidance. An article should describe its actual headline, author, dates, and page. A product or offer needs accurate product and commercial information. Do not add an unrelated type solely because it sounds useful for AI search.
Validate syntax and then validate meaning
Run a JSON parser and the appropriate structured-data validation tools. Distinguish syntax errors, missing recommended fields, eligibility requirements, and factual contradictions. These are not the same severity. A parser passing confirms readable JSON, not that the entity describes the correct page.
Google's structured-data policies require markup to represent the page content appropriately. Compare prices, availability, authorship, review claims, and dates with what readers can actually see. Never manufacture ratings or update timestamps to make a page look more authoritative or recent. A valid-looking block containing unsupported facts is a correctness issue.
Reference: Google Search Central: structured data guidelines
Review entity identity and URL relationships
Use consistent company and product names with stable identifiers where the implementation calls for them. Check that page URLs, main-entity references, breadcrumb items, and logos point to the intended public destination. Avoid staging addresses or broken relative paths in production metadata. An entity audit should detect contradictions across templates, not merely grade each block in isolation.
Compare important pages that describe the same entity. If a product name or plan definition changed, update the approved facts and relevant templates together. Keep regional or language variants explicit when they materially change the content. A correction should preserve factual distinctions rather than collapse all variants into one oversimplified entity.
Verify the delivered schema and visible answer
In a hypothetical audit, an article's JSON-LD points to a staging URL and declares a recent modification date even though the text has not changed. The reviewed correction uses the public canonical URL and the truthful content date. The team also checks the displayed byline and timestamp so the metadata agrees with the page.
After delivery, re-fetch the page, parse the live output, and inspect the same fields. Confirm that a plugin did not restore the old block or leave a duplicate behind. Treat improved markup validity as a delivery outcome. Evaluate search presentation and citation observations separately; Google makes clear that supported Article markup can help describe an article without guaranteeing presentation.
| Layer | Review |
|---|---|
| Syntax | Can the structured data be parsed? |
| Applicability | Does the type fit the actual page? |
| Factual agreement | Do claims match visible, approved information? |
| Identity | Are names, IDs, and entity relationships consistent? |
| Delivery | Does the corrected block exist on the live page? |
Your next steps
- Collect initial and rendered structured data and identify its generating templates.
- Validate syntax separately from eligibility and factual correctness.
- Compare important claims with visible content and approved product facts.
- Check public URLs, entity identifiers, dates, and duplicate blocks.
- Re-fetch the delivered page and inspect the exact corrected fields.
Common questions
Is there a special schema type for Google AI features?
Google's AI-feature guidance does not require special schema markup for inclusion. Use relevant, accurate structured data for the actual page and follow ordinary technical and content guidance.
Should we update modification dates automatically?
Use dates that represent substantive content changes. A scheduled render or deployment alone should not imply that the article was materially revised. Keep displayed dates, structured data, and sitemap modification information consistent with the actual content history.
Sources & further reading
Official documentation supporting this guide. Worked examples are illustrative unless explicitly identified as measured results.