HomeInsightsSchema Markup and AI…

Schema Markup and AI Citations: What a 1,885-Page Study Found

SEO vendors are selling schema markup as the fix for getting cited by ChatGPT and AI Overviews. A controlled study of 1,885 pages, Google's own guidance, and a direct test of what AI platforms can even read say otherwise.

By Alex RiveraPublished October 12, 2026

Ahrefs tracked 1,885 pages that added JSON-LD schema markup between August 2025 and March 2026, matched against 4,000 similar pages that didn't, and measured what happened next. Google AI Overviews citations fell 4.6%. AI Mode and ChatGPT moved less than 3% either direction — statistical noise (Ahrefs, 2026). **Adding schema markup to a page does not get it cited more by AI search; the tag helps a machine confirm facts about your business, not win it a citation.**

That's worth sitting with, because schema markup is being sold right now as a 2026 AI-visibility essential — the same pitch llms.txt got a year ago. If a vendor is quoting a Kalispell or Missoula business owner a line-item fee specifically to add "AI schema" and calling it the thing that gets their business named by ChatGPT, the controlled data below is the reason to ask what else is actually in that invoice.

Can ChatGPT and Perplexity Even Read Schema Markup?

Mostly not, and a separate experiment shows exactly why. Otterly.ai ran a three-part test across seven AI platforms between December 2025 and March 2026, directly asking each one to fetch and interpret the raw JSON-LD schema on a page. Six of the seven — ChatGPT, Google AI Overviews, Perplexity, Microsoft Copilot, Claude, and Google AI Mode — could not correctly retrieve or interpret the markup when asked outright. Google AI Mode went further and hallucinated schema that wasn't on the page at all. Only Gemini read it correctly (Otterly.ai, 2026). Most large language models tokenize JSON-LD as plain text alongside the rest of the page, rather than parsing it as structured data the way a search engine's indexer does.

PlatformCould it read your JSON-LD when asked directly?What Ahrefs measured after schema was added
Google AI OverviewsNo — not reliably parsed as structured data (Otterly.ai, 2026)Citations fell 4.6%, a small but statistically significant drop (Ahrefs, 2026)
Google AI ModeNo — hallucinated schema not present on the page (Otterly.ai, 2026)+2.4%, indistinguishable from zero (Ahrefs, 2026)
ChatGPTNo — tokenized as raw text, not semantically parsed (Otterly.ai, 2026)+2.2%, indistinguishable from zero (Ahrefs, 2026)
GeminiYes — the one platform that retrieved it correctly (Otterly.ai, 2026)Not isolated separately in the Ahrefs study

What Does Google Itself Say About Schema and AI Search?

It says skip it. Google's own Search Central guide to generative AI features, last updated July 10, 2026, states flatly: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." The guide lists "overfocusing on structured data" by name inside its mythbusting section — grouped with other things site owners are told they need but don't (Google, 2026). Google's AI Overviews and AI Mode run on the same core Search ranking and quality systems as regular results; they don't have a separate schema requirement layered on top.

So Is Schema Markup a Waste of Time in 2026?

Honest answer: no, but not for the reason it's being sold. Schema markup still earns rich results — star ratings, FAQ dropdowns, event and product details — inside classic Google search, and it still helps Google's Knowledge Graph confirm who your business is, what it does, and where it's located. That entity clarity feeds your Google Business Profile and local search presence. What it does not do, per Ahrefs' controlled study, is move your citation count inside ChatGPT, AI Overviews, or AI Mode. Keep schema for the job it's actually good at. Don't pay extra for it as an "AI visibility" line item.

Ahrefs' own caveat matters here too: the study only tracked pages that already had 100+ AI Overview citations before adding schema, over a 30-day window, with all schema types pooled together. It can't say whether schema helps a page get noticed for the first time, or whether FAQPage specifically behaves differently from Organization or LocalBusiness. What it can say, cleanly, is that adding schema to a page already being cited didn't increase how often it kept getting cited — and for most small businesses paying someone to retrofit markup onto an existing site, that's the exact scenario being sold to them.

What Actually Moves AI Citations, If Not Schema?

  • A direct, liftable answer in the first 40-60 words of a page, stated as a conclusion rather than a lead-up.
  • Real statistics with a named, checkable source — the kind of content both Ahrefs and Google point toward instead of markup.
  • Clean, crawlable HTML that AI bots can actually render without JavaScript.
  • Consistent business facts — name, services, location — across your site and your Google Business Profile, which is what actually feeds entity confidence.
Skyline AI runs AI SEO and AEO work for Northwest businesses that's built around what the data actually shows moves AI citations — not what's easiest to sell as a line item. Book a free AI visibility audit to see where your site stands today.

Sources

  1. Ahrefs (2026)
  2. Google (2026)
  3. Otterly.ai (2026)
[ 05 ]Questions

Related questions

Clear answers to the questions operators ask most. Still not sure if AI fits your business? Talk to us — no pitch, just a straight read on where it pays off.

Does schema markup help you get cited by ChatGPT or AI Overviews?

No, according to Ahrefs' controlled study of 1,885 pages: adding schema markup produced no statistically meaningful lift in AI Mode or ChatGPT citations, and Google AI Overviews citations actually fell 4.6% (Ahrefs, 2026). Google's own guidance agrees, listing structured data as unnecessary for generative AI search.

Can AI platforms like ChatGPT actually read JSON-LD schema markup?

Mostly not. Otterly.ai directly tested seven AI platforms and found six — including ChatGPT, Perplexity, and Google AI Overviews — could not correctly retrieve or interpret JSON-LD when asked. Only Gemini parsed it correctly; most others tokenize it as plain text (Otterly.ai, 2026).

Should I still add schema markup to my website in 2026?

Yes, for classic Google search and entity clarity — schema still earns rich results and helps confirm business facts to Google's Knowledge Graph. Just don't pay for it expecting a lift in ChatGPT or AI Overview citations; the controlled data says it won't move that number.

What should I do instead of schema markup to get cited by AI?

Lead with a direct, 40-60-word answer, back claims with named and checkable sources, keep your site crawlable without JavaScript, and keep your business facts consistent across your site and Google Business Profile. Those are the factors Ahrefs and Google both point to.

Related questions

Related services

More from Insights

No Pitch, No Obligation

See exactly where AI pays off in your business

Book a free AI audit. We'll map your biggest leak — missed calls, slow follow-up, manual admin — and show you the system that fixes it. No pitch, no obligation.

Free · no obligation~30 minutesYou own everything