What schema.org markup really does for AI crawlers, and where SEO folklore has gotten ahead of the evidence.
Every "how to get cited by AI" guide tells you to add schema markup. Some of that advice is grounded. A lot of it is a Google-era SEO habit copy-pasted onto a new problem without anyone checking whether it still applies. Here's what's actually documented, what's genuinely uncertain, and what's been tested and found wanting.
Structured data — usually written as JSON-LD, the format schema.org recommends — is a machine-readable description of a page embedded in its HTML. Instead of making a crawler infer "this page is about a product called X, made by company Y, priced at Z" from prose, you state it directly: Product, Organization, FAQPage, Review, and dozens of other types let you label entities and relationships explicitly. Mentioned's own homepage uses WebApplication, Organization, WebSite, and FAQPage markup for exactly this reason — it's a direct, unambiguous statement of what the product is and what its FAQ says, rather than something a crawler has to guess at from formatting.
Two things are well established, not speculative:
The best empirical look at schema's effect on AI citations specifically comes from Ahrefs, which tracked 1,885 pages that added JSON-LD schema between August 2025 and March 2026, compared against 4,000 control pages, and measured citation changes across Google AI Overviews, AI Mode, and ChatGPT. The result: adding schema produced no meaningful uplift. AI Overviews citations actually declined slightly (a statistically notable −4.6%), while AI Mode and ChatGPT moved a small amount in either direction — too small to distinguish from noise. One important caveat in that study: every page in the dataset already had 100+ AI Overview citations before schema was added. In other words, the test measured "does schema help pages that are already being cited get cited more," not "does schema help an invisible page get discovered at all" — those could have different answers, and the study says so explicitly.
A separate December 2024 analysis from Search/Atlas, cited in Search Engine Land's write-up on schema and AI search, found no correlation between how much schema coverage a site had and how often it got cited by AI systems — sites with comprehensive markup didn't consistently beat sites with little or none. That same piece notes there are, as of today, no peer-reviewed studies and no controlled experiments proving schema causes AI citations to increase.
The Ahrefs study also found that pages cited by AI were about three times more likely to have JSON-LD than pages that weren't cited. That sounds like an argument for schema — until you notice that sites which bother implementing structured data also tend to invest in the things that actually seem to matter: original content, real third-party coverage, clean crawlable pages, and topical depth. Schema correlates with citation-worthy sites because it correlates with sites that take their web presence seriously in general, not necessarily because the markup itself is doing the work.
Add schema markup. It's cheap, it's a real machine-readability signal that platforms like Bing Copilot and Google AI Overviews have said they use to understand entities, and it costs you nothing to be unambiguous about who you are and what you sell. What you shouldn't do is treat it as a lever that moves your AI citation rate — the evidence so far says it doesn't, at least not on its own, and not for pages that are already visible. If you're not showing up in AI answers at all, the more likely fix is upstream of schema: genuinely useful content, real coverage other sites link to, and a site that's crawlable in the first place. Structured data is the last 5%, not the first 50%.
If you want to know where you actually stand before spending time on markup, that's a five-minute question, not a guess.