Adding schema markup will not make an AI cite your page. That is the conclusion of an Ahrefs study from May 2026, reported on October 28, 2026, which tracked 1,885 pages adding schema and found that AI citations barely moved. The study is a useful corrective to a popular belief in the AI search world: that structured data is a shortcut to being cited. The data says otherwise, and the reason is instructive about how AI systems actually choose their sources.
The research began with an analysis of approximately six million URLs, and found that pages cited in AI-generated answers were almost three times as likely to contain JSON-LD as pages that were not cited. That correlation is what launched the belief: cited pages have schema, so adding schema must cause citations. The comparison could not establish causation, because the pages containing schema may have differed from the others in content quality, domain authority, technical maturity, or other characteristics associated with citation selection.
To investigate whether implementing schema itself made a difference, Ahrefs conducted a follow-up analysis of pages that introduced JSON-LD. The result was clear: citations barely moved. The most defensible interpretation is that the study did not demonstrate a meaningful, reliable citation uplift from the introduction of JSON-LD on the tested pages. Citations declined in both the treatment and control groups during the study period, and the treated group experienced a somewhat larger decline under the preferred model. Only the AI Overviews estimate was statistically significant.
The mechanism is easy to see once it is stated. The sites that implement schema properly tend to be the sites that do everything else properly too: they have real content, real authors, clean technical setups, and genuine authority. Those qualities are what earn citations, and schema is a marker of them, not a cause. Adding schema to a thin page does not make the page authoritative. It makes a thin page with schema, and the AI system can tell the difference.
Google's own 2026 guidance confirms the point from the other side. Google states explicitly that no special Schema.org markup is required for its generative AI search features. A page must be indexed and eligible to show with a snippet, and it must be included in Search generative AI features in Search Console, but there is no markup gate to pass. Independent research has not established a reliable positive citation effect from adding JSON-LD alone, and Ahrefs' study is now the strongest evidence for that position.
Yes, but for the right reasons. Schema markup still helps with rich results in classic search, it still clarifies what a page is about for any system that reads it, and it still makes a site's data machine-readable in a way that helps across the board. What it does not do is buy a citation. A site should implement schema as part of good technical hygiene, not as an AI visibility strategy, and it should not expect the citation graph to move when the markup goes live.
The budget that would have gone into a schema-first AI programme is better spent on the things the evidence says do move citations: original data, clear answers, genuine expertise, and pages that deserve to be quoted. Schema is the label on the package. The content is the product, and no label will sell a product that is not worth buying. The study does not argue against schema. It argues against magical thinking, and it leaves the real work where it has always been: on the page itself, in the quality of what it says and the authority of who says it.
Sources: Ahrefs study, May 2026, reported October 28, 2026.
Linkbaits helps you create content that earns citations and backlinks. Start with the free tools and build your first linkable asset this week.
Start free trial