Three schema types that matter
- Organization: brand identity, with sameAs links to Wikidata, LinkedIn and verified profiles.
- Person: author or founder identity, linked to the Organization through @id.
- FAQPage: questions organized as potential entry points for distinct AI queries.
The most common mistake
Duplicated or inconsistent schema across pages. A different @id on each page breaks the connection between entities. Use one canonical @id per entity, everywhere.
Another common mistake is declaring information that is invisible or different from the actual content. When the schema says one thing and the page shows another, you create friction. A machine should not have to guess which version is true.
What good JSON-LD should establish
Good JSON-LD connects entities. Person points to Organization. WebPage points to WebSite. Article points to its author and sources. FAQPage matches questions actually visible on the page. BreadcrumbList reflects real navigation.
Every @id must be stable. Every sameAs must be a reliable identifier, not a Google search. Every significant property needs a source. This consistency makes a page readable for AI search engines.
Schema.org and AI Search
AI systems look for structures, recurring patterns and reliability signals. Schema.org provides a shared grammar for naming entities, relationships and evidence.
Schema is not magic and cannot compensate for weak content. With solid content, however, it reduces ambiguity. In AI Search, that helps a source become understandable, reusable and citable.