AI Strategy & GEO

Schema for AI search: accurate descriptions, no citation guarantee

Fernando Angulo
Senior Market Research Manager at Semrush, an Adobe company
11 Min Read
Jun 9, 2026 · Updated Aug 4, 2026

Structured data describes a page’s entities and content in a machine-readable vocabulary. Use supported types accurately. It is not a substitute for useful visible content or a special requirement for Google’s generative AI features.


Start with visible facts

The markup should match what a person can verify on the page. Correct the author, organization, dates and URLs before adding properties. A schema-valid statement can still be factually wrong.

Choose a type for its actual purpose

Person, Organization and Article can describe their respective entities. Follow current feature-specific documentation for eligibility. FAQPage, QAPage and speakable have different purposes and are not interchangeable AI-citation switches.

Preserve identity without exact-word rules

Keep employer, role and dates consistent across pages. Use canonical identifiers and relevant sameAs links where appropriate. Natural paraphrases and translations do not by themselves create conflicting people.

Validate syntax and meaning

Check JSON syntax, identifiers, visible-content alignment and current platform guidance. A valid document is not proof that a search engine will index, display or cite it. Do not publish ratings or credentials in schema without supporting evidence.

Measure separately

Google’s Generative AI performance report provides AI-feature impressions; it is not a causal test of schema. If evaluating a markup change, record what changed and compare outcomes while considering other changes and ordinary variability.

Sources and scope

Practitioner analysis by Fernando Angulo. Proposed examples are not measured outcomes.

Google: current AI search guidance

Google: Generative AI performance report

Recommended Reading

Latest Insights

View all articles