Schema markup can help search systems interpret the entities, products, articles, and other facts on a page, but it is not a proven direct lever for earning AI citations. Current evidence suggests that pages with structured data are often cited because they also have strong content, authority, and search visibility, not necessarily because the markup itself caused selection. Use accurate, relevant JSON-LD that matches the visible page content, then prioritize crawlability, clear headings, and self-contained answers in plain text. The costly mistake is treating markup as a substitute for the information an AI system can actually retrieve and verify.
Current Evidence on Structured Data and AI Answer Citations
Citations, Retrieval, Mentions, and Referral Traffic
AI search visibility is not one metric. A page can be retrieved by an AI system but never shown to the user. It can be mentioned without a clickable link, cited as a supporting source, or receive a referral visit after a user opens that citation. These outcomes are related, but they are not interchangeable. Schema markup may help a system interpret page-level facts, such as a product’s price, an organization’s identity, or an article’s author. It does not make a URL automatically eligible for an AI answer citation. Google states that pages shown as supporting links in AI Overviews and AI Mode must first be indexed and eligible to appear with a Search snippet, with no separate AI-only schema requirement. Its guidance also stresses that structured data must match visible page content. Google’s AI features documentation is clear on this point. For measurement, separate citation visibility from business results. Track whether pages appear in AI experiences, then compare referral sessions, engagement, and conversions. A visible citation may build recognition even when it produces few clicks, while a less frequent citation on a high-intent query can be far more valuable.
What Controlled Studies Have Found
Public evidence does not currently prove that adding JSON-LD, by itself, causes more AI citations. Most published analyses are observational: they compare cited and non-cited pages, then attempt to account for variables such as organic rankings, domain authority, topic, and content format. That distinction matters because well-optimized pages often have both schema markup and stronger underlying SEO signals. Kurt Fischman’s study, Does Schema Markup Predict AI Citation? A Cross-Platform Empirical Study of Structured Data and Generative Engine Optimization, published February 22, 2026, analyzed 730 citations from ChatGPT and Gemini across 75 commercial queries and 1,006 unique pages. It found no statistically reliable independent relationship between general schema presence and citation likelihood after controlling for ranking-related factors; organic position was the strongest predictor. Product and Review markup with populated concrete attributes was associated with higher citation rates, but the observational result does not establish causation. This fits Microsoft’s practical guidance for Bing and Copilot: accurate structured data may support clearer grounding, but it does not guarantee visibility or grounding traffic. Bing’s Webmaster Guidelines place schema alongside crawlability, explicit facts, focused pages, and independently verifiable content.
Why Schema and AI Citation Correlations Can Mislead
Shared Signals Behind Cited Pages