As AI Overviews, AI Mode, ChatGPT search, Perplexity and Copilot have grown, many teams ask whether Schema.org markup helps them get cited. The honest answer is mixed. Google says no special markup is needed for its AI features and that structured data should match visible content. Other AI vendors have published little about how, or whether, they use it.
What is documented
Google's guidance on AI features says the same SEO best practices apply and that there are no extra requirements, including no special schema, to appear in AI Overviews or AI Mode. It lists structured data that matches visible content as good practice. Since Google's AI features draw on its index and Knowledge Graph, markup that helps Google understand entities may indirectly help, but Google has not described a direct link.
Microsoft representatives have said publicly that structured data helps Bing understand content, including for its AI experiences, without publishing details of how it is used.
What is not known
OpenAI, Anthropic and Perplexity have not documented whether their crawlers or retrieval systems parse JSON-LD. Many AI crawlers fetch raw HTML; whether markup in that HTML is kept, stripped or weighted is not public. Claims that a specific schema type raises AI citation rates are not backed by vendor documentation, and published experiments tend to be small and hard to generalize.
A sensible position
- Keep the core types accurate: Organization, WebSite, Article, Person, product types, BreadcrumbList
- Make entities unambiguous with sameAs and stable @id values
- Put every important fact in visible text, which all systems can read
- Avoid building markup for AI alone where it has no other use
Example
Weighing a vendor pitch
Juniperhold, a SaaS company, is pitched a tool that promises AI citations by injecting dozens of schema types. Its SEO lead checks Google's AI features guidance, finds no special markup requirement, and instead fixes inconsistent Organization data and gets author pages marked up properly.
Common mistakes
Treating markup as a citation switch
No vendor documents a direct schema-to-citation effect. Content and entity clarity still do the work.
Hiding facts in markup only
If a fact is only in JSON-LD, systems that ignore markup never see it, and Google may see it as a mismatch.
Questions
No. They are separate, and llms.txt is a proposal with limited vendor adoption. Structured data is an established standard used by search engines.
