Schema Markup Won’t Win You AI Citations: We Checked 138 Pages in Google’s AI Overviews

By Blogging Titan Team

Last Updated on July 22, 2026

Add Blogging Titan as a trusted source on Google

Quick answer: Schema markup does not decide which pages Google’s AI Overviews cite. In our study of 138 pages pulled from 18 live AI Overviews, the pages that got cited were no more likely to use structured data than the pages that got ignored. Cited pages carried article schema 60 percent of the time; uncited pages carried it 65 percent of the time. Schema is table stakes for eligibility, not a lever you pull to earn citations. We call this the schema plateau: structured data is now so common that its presence no longer separates the cited from the uncited.

Ask around any SEO community and you will hear the same advice on repeat: add FAQ schema, add Article schema, add HowTo schema, and the AI engines will start citing you. It is tidy, it is technical, and it feels like something you can control. So we tested it.

We collected every source Google’s AI Overview cited across a batch of real searches, then checked the structured data on each page and compared it against the organic results that were not cited. If schema were a citation lever, cited pages should be visibly richer in structured data than their uncited neighbours. They were not.

Table of Contents show

What we found

Across 20 Google searches (informational, commercial, and transactional intent), an AI Overview appeared on 18 of 20 queries, a 90 percent trigger rate. We then analysed the structured data on 138 independent publisher pages drawn from those results. Here are the numbers that matter.

Cited pages were no more likely to use schema than uncited pages. 84 percent of AI-Overview-cited publisher pages carried some structured data, versus 89 percent of the uncited organic pages. If anything, the pages Google ignored had slightly more schema.
Article or blog schema showed up on 60 percent of cited pages and 65 percent of uncited pages. The gap runs the wrong way for the “add schema to get cited” theory.
40 percent of cited pages had no article schema at all, and 16 percent had no structured data whatsoever. Google’s AI Overviews happily quote pages with weak or missing markup.
FAQ schema barely moved the needle: 21 percent of cited pages versus 17 percent of uncited pages. On a sample this size, that difference is well within noise.
Cited and uncited pages carried an almost identical amount of markup: a median of roughly eight schema types each (8.2 cited, 8.0 uncited).
Nearly one in five AI Overview citations (19 percent) went to sources that cannot use article schema at all because it is not their page: Reddit and forums took 9 percent, YouTube 7 percent, and Medium or Substack 4 percent.

Why doesn’t schema predict AI citations?

Because almost everyone already has it. Structured data used to be a differentiator when only sophisticated sites bothered. Today it ships by default with every serious WordPress SEO plugin, every Webflow template, every enterprise CMS. When 84 to 89 percent of the pages in a result set carry schema, its presence cannot explain why one page beats another. That is the schema plateau: a signal so widely adopted that it has flattened into a baseline.

This lines up with what we have found in our other AI Overview studies. In our citation-rank disconnect study, roughly 68 percent of AI Overview citations went to pages that do not even rank in the organic top 10. In our first-person language study, the writing signal everyone assumed mattered turned out not to predict citation either. The pattern keeps repeating: the levers people obsess over rarely turn out to be the ones the models actually pull.

So what does schema actually do?

Plenty, just not this. Structured data still earns you rich results in classic search, still powers eligibility for features like FAQ snippets and review stars, and still gives machines a cleaner read of your entities and relationships. Those are real reasons to keep your schema tidy. What our data says is narrower and more useful: do not expect adding schema to move your AI citation rate, because the pages you are competing against already have it.

The one place structured data and citation genuinely overlap is not the markup itself but what good markup tends to sit on top of: a clear, extractable, answer-first page. A page that deserves FAQ schema usually has real questions answered in plain language near the top. That answer is what gets quoted. The schema is just the wrapper around it, and what the model reads for meaning is the answer inside.

The transactional blind spot

One more finding worth flagging. The only two queries in our set with no AI Overview at all were both transactional pricing searches, “ahrefs pricing” and “mailchimp pricing,” where a single brand-owned page dominates the answer. AI Overviews cluster on informational and comparison intent, where there are many partial answers to stitch together, and thin out where one authoritative source already settles the question. If your money pages target that kind of query, the AI Overview may not be your battleground at all.

What to do instead of chasing schema

Keep your structured data clean, because it costs little and helps elsewhere, then stop treating it as your AI-citation strategy. Put the effort into the things that actually correlate with getting quoted: leading with a direct answer in the first 40 to 60 words, writing question-shaped headings, packing in verifiable statistics, and citing primary sources. If you want a fast read on how quotable a given post is, run it through our AI Citation Grader, which scores a page across eight weighted GEO signals and tells you which ones to fix first. Note that schema is only one of those eight, and a lightly weighted one, for exactly the reason this study demonstrates.

How we ran the study

We ran 20 Google searches in a clean, US-localised, logged-in Chrome session, spread across informational intent (how-to and definitional queries), commercial intent (“best” and comparison queries), and transactional intent (pricing and single-product queries). For each search we isolated the AI Overview block and recorded every external source it cited, then recorded the organic top-10 results for the same query. We visited 138 unique independent publisher pages from those two groups and parsed the JSON-LD and microdata on each, recording every schema type present. Pages that Google’s AI Overview cannot control the markup on (Reddit, YouTube, Quora, Medium, Substack) were counted for citation share but excluded from the schema comparison, since a site owner cannot add article schema to someone else’s platform. A small number of heavy single-page-app homepages that would not expose their markup to inspection were excluded. This is a correlational snapshot of one moment in AI Overview behaviour, not a controlled experiment, and AI Overviews are personalised and change often, so treat the exact percentages as directional.

Cite this data

Blogging Titan, Schema Markup and AI Overview Citations: A 138-Page Study (2026). Structured-data presence on AI-Overview-cited vs uncited pages across 18 Google AI Overviews.

Metric	AI-Overview-cited pages	Uncited organic pages
Any structured data	84 percent	89 percent
Article or blog schema	60 percent	65 percent
FAQ schema	21 percent	17 percent
No structured data at all	16 percent	11 percent
Median schema types per page	8.2	8.0

Sample: 138 unique publisher pages across 20 Google searches, 18 with AI Overviews, July 2026. Data by Blogging Titan, released under CC BY 4.0. BibTeX key: bloggingtitan2026schema.

Frequently asked questions

Does schema markup help you get cited in Google’s AI Overviews?

Not on its own. In our study of 138 pages, AI-Overview-cited pages were no more likely to carry schema than uncited pages (84 percent versus 89 percent for any schema, 60 percent versus 65 percent for article schema). Schema is worth keeping clean for rich results and eligibility, but its presence does not distinguish cited pages from uncited ones because nearly every competing page already has it.

Should I still add schema markup to my blog?

Yes, but for the right reasons. Structured data supports rich results in classic search, powers features like FAQ and review snippets, and helps machines parse your content. Just do not expect it to raise your AI citation rate. Treat it as hygiene.

What actually gets a page cited in AI Overviews?

Our broader research points to answer-first structure, question-shaped headings, verifiable statistics, and primary sourcing far more than markup. The page has to contain a clear, extractable answer that the model can lift. Schema wraps that answer; the answer itself is what earns the citation.

Which schema type is best for AI search?

None of them reliably outperformed the others in our data. FAQ schema showed a tiny, statistically insignificant edge on cited pages (21 percent versus 17 percent). If you are choosing what to implement, prioritise the schema that earns you classic rich results for your content type, and put your real effort into the answer itself.

How many AI Overview citations go to Reddit and YouTube?

In our sample, 19 percent of all AI Overview citations went to sources where the site owner cannot add article schema: 9 percent to Reddit and forums, 7 percent to YouTube, and 4 percent to Medium or Substack. That alone shows structured data is not a precondition for being cited.

Want this done for you?

We audit your pages for the signals that actually earn AI citations, then fix them. See how our GEO citation audits and done-for-you service work.

View our services