The short answer
Make structured data describe the article readers can actually see. Use the real title, actual publication history and legitimate author entity. Validate syntax and inspect the rendered page together; passing a JSON parser does not establish that the markup is accurate or eligible for a search feature.
A worked example
A fictional brand publishes an organizational guide, not a named expert's report. Its Article markup should not invent a person, credentials or a review score. If the visible page discloses AI assistance and a pending owner review, the structured data must not imply a separate expert approval that never occurred.
A practical checklist
- Compare headline, author, dates and image information with the visible article. Use an organizational author when that is the true attribution rather than creating a fictional staff biography.
- Check JSON-LD syntax and basic fields locally. Then use the applicable official testing tool when preparing a public release; different validators check different things.
- Remove properties unsupported by the page or evidence. Confirm the canonical URL and ensure any linked author information identifies the actual author entity.
What to avoid
Avoid using structured data as hidden advertising or adding fabricated aggregate ratings. This rebuild's local schema checker is intentionally a structural aid, not a full vocabulary or rich-results certification. Search engines can decline special presentation even for valid markup.
A useful follow-up
Can an organization be an article author?
Yes. Use the legitimate organizational identity and an identifying URL when that matches the visible byline.
