The role of structured data in AI Search visibility
Learn more about structured data: what it is, and how to implement it properly so that your page content is easier to interpret – not just by traditional search engine algorithms, but also by LLM crawlers.
Key takeaways
- Short answer: schema helps search engines, not AI citations directly. Controlled studies found no meaningful lift in AI citations after adding structured data.
- Google says no special markup is needed for AI Overviews or AI Mode. Microsoft has said schema helps the LLMs behind Copilot understand content.
- ChatGPT and Perplexity read JSON-LD as plain text, not as structured data. What they cite is the content visible on the page.
- FAQ rich results are gone from Google since May 2026, and HowTo since 2023. Don’t add them expecting a search or AI boost.
- Still worth doing: Organization, Article, Product and Breadcrumb markup, kept accurate and consistent with the visible page, in the HTML the server sends.
AI search tools such as ChatGPT, Perplexity, Claude, Google AI Overviews and AI Mode now answer many questions directly, citing a few sources. That’s raised an obvious question for site owners: does structured data help you become one of those sources?
The honest answer is more nuanced than most guides suggest. Structured data still matters for Google, Bing and shopping results. But the evidence that it increases AI citations is weak, and several popular recommendations are out of date. This article explains what schema does, what it doesn’t, and what to implement.
Make your brand visible, from Google to AI chatbots.
What is structured data?
Structured data is a standard way of describing a page’s content to machines. It tells a search engine that a piece of text is, for example, a product name, a price, an author or a publication date. Google uses it to understand the content of the page and to enable rich results.
The most common vocabulary is schema.org. There are three formats:
- JSON-LD: a script block (<script type=”application/ld+json”>) separate from the visible layout. It’s the format Google recommends and the easiest to maintain.
- Microdata: attributes such as itemprop and itemscope added to HTML elements. Still supported, but harder to manage on large sites.
- RDFa: attributes used mainly in linked open data projects. Valid, but rare in SEO.
How do AI search engines choose what to cite?
AI search engines retrieve pages through search (Google’s index, Bing or their own crawlers), extract the passages that answer the question, and cite some of them. What they reward is content they can quote:
- a direct answer near the top of each section;
- logical, question-style headings, lists and tables;
- specific, sourced facts rather than promotional claims;
- pages that already rank and are mentioned elsewhere; see backlinks and mentions in GEO.
In an AirOps study of ChatGPT citations, precision and ranking mattered more than length. Structured data doesn’t appear in these findings as a selection factor.
Does structured data help AI search? What the evidence shows
| Source | Finding |
|---|---|
| Google, AI features guide (May 2026) | Structured data isn’t required for AI Overviews or AI Mode, and there’s no special schema.org markup to add for them. |
| Microsoft Bing, SMX Munich (Mar 2025) | Fabrice Canel said schema markup helps Microsoft’s LLMs understand content. |
| Ahrefs (May 2026) | 1,885 pages that added JSON-LD, compared with 4,000 control pages: no meaningful lift in citations in AI Overviews, AI Mode or ChatGPT. |
| Search Atlas (Dec 2025) | No link between schema coverage and how often domains were cited by OpenAI, Gemini or Perplexity. |
| Mark Williams-Cook test (2026) | ChatGPT and Perplexity returned an address hidden only in invalid, made-up JSON-LD, so they read it as text, not as structured data. |
| searchVIU (2025) | Assistants that fetched a page live extracted the visible content; data that existed only in JSON-LD was ignored. |
What does this mean?
- For Google and Bing, structured data helps their systems understand your entities, products and pages. Bing has said this extends to Copilot.
- For ChatGPT, Perplexity and Claude, there’s no evidence that schema is used as structured data. At most, JSON-LD is extra text on the page.
- For AI citations, no study so far shows that adding schema increases how often you’re cited. Content, rankings and reputation explain far more.
Which schema types are still worth implementing?
| Schema type | Why it matters | AI relevance |
|---|---|---|
| Organization (with logo, sameAs) | Connects your site to official profiles and brand facts | Helps Google and Bing understand the entity |
| Article / BlogPosting | Author, publication and update dates | Supports freshness and authorship signals in search |
| Product, Offer, Review, AggregateRating | Price, availability, ratings in Google results and Merchant listings | Product data also feeds shopping features; see product descriptions for AI search |
| BreadcrumbList | Shows site structure in results | Indirect |
| LocalBusiness | Address, hours, contact details | Helps consistent local data |
| FAQPage, HowTo | No longer generate rich results in Google | No evidence of AI benefit; put Q&A and steps in visible content |
Google also phased out several rich result types in June 2025, including Course Info, Claim Review, Estimated Salary, Learning Video and Vehicle Listing. For a full SEO view of schema types, see which Schema.org structured data is worth using.
What about FAQ schema?
Google removed FAQ rich results in May 2026. A visible FAQ section is still useful: question-and-answer pairs are easy for AI to quote. The markup itself is now optional. If you keep it, make sure it matches the visible text exactly.
How do you implement structured data correctly?
- Use JSON-LD in the HTML the server sends. Crawlers that don’t render JavaScript, including those from OpenAI and Anthropic, won’t see markup injected by Google Tag Manager or client-side scripts.
- Mark up only what’s visible. Every fact in structured data should also appear on the page. Mismatches can lead to manual actions in Google and confuse AI systems.
- Keep facts consistent. Prices, availability, dates and authors must match across the page, the markup and your product feed.
- Validate. Use Google’s Rich Results Test for supported features and the Schema.org validator for everything else.
- Notify search engines. Use IndexNow for Bing and other participating engines, and request indexing in Search Console after major changes.
Example of Article markup with an author and organization:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "BlogPosting",
"headline": "Does structured data help AI search visibility?",
"datePublished": "2025-07-25",
"dateModified": "[YYYY-MM-DD]",
"author": {
"@type": "Person",
"name": "[Name Surname]",
"jobTitle": "[Role]",
"url": "https://insightland.org/[author-page]/",
"sameAs": ["https://www.linkedin.com/in/[profile]/"]
},
"publisher": {
"@type": "Organization",
"name": "Insightland",
"url": "https://insightland.org/",
"logo": "https://insightland.org/[logo].png"
}
}
</script>
Boost sales with SEO + GEO.
So is structured data worth implementing?
Yes, but for the right reasons:
- it enables rich results in Google, such as product prices, ratings and breadcrumbs;
- it helps Google and Bing understand your brand, products and authors, and Bing says it helps Copilot;
- it feeds shopping features and third-party services that read product data;
- it forces you to keep key facts consistent, which also helps AI describe you correctly.
What it won’t do is get you cited by AI on its own. For that, focus on clear, specific content, rankings, brand mentions and trust signals. See E-E-A-T in AI search and 10 AI search myths.
Conclusion
Structured data is part of a solid technical foundation, not a shortcut to AI visibility. Keep Organization, Article, Product and Breadcrumb markup accurate and server-rendered, drop expectations for FAQ and HowTo, and put the information you want AI to quote in the visible content. Then measure what matters: whether AI tools cite and describe you correctly. See metrics that matter in AI search.
Not sure if your structured data is implemented correctly, or how visible you are in AI answers?
FAQ: structured data and AI search
Does schema markup help with AI citations?
Not measurably. Ahrefs tracked 1,885 pages that added JSON-LD and found no meaningful lift in citations in AI Overviews, AI Mode or ChatGPT. Search Atlas found no link between schema coverage and citations.
Do LLMs read JSON-LD?
They can read it as text when it’s in the page HTML, but tests show ChatGPT and Perplexity don’t treat it as structured data. They read invalid, made-up markup the same way.
Does Google use structured data in AI Overviews?
Google says structured data isn’t required for its AI features and there’s no special markup for them. It still uses structured data for rich results and to understand page content.
Is FAQ schema still worth adding?
It no longer generates rich results in Google (since May 2026). A visible FAQ section is still valuable for users and AI; the markup is optional.
Which schema types matter most in 2026?
Organization, Article or BlogPosting, Product with Offer and reviews, BreadcrumbList and LocalBusiness. Keep them accurate and consistent with the visible page.
Can AI crawlers see schema added with Google Tag Manager?
Usually not. Most AI crawlers don’t run JavaScript, so markup injected by GTM or client-side scripts is invisible to them. Add JSON-LD to the server-rendered HTML.
Sources
- Google Search Central: Optimizing your website for generative AI features (May 2026)
- Google Search Central: Introduction to structured data
- Google Search Central: Simplifying the search results page (Jun 2025)
- Search Engine Journal: Google drops FAQ rich results from Search (May 2026)
- Search Engine Land: Microsoft Bing/Copilot use schema for its LLMs (Mar 2025)
- Ahrefs: We tracked 1,885 pages adding schema (May 2026)
- Search Atlas: The limits of schema markup for AI search (Dec 2025)
- Search Engine Journal: Schema, LLMs and the low bar for evidence in GEO (Mark Williams-Cook) (2026)
- searchVIU: Schema markup and AI, what ChatGPT, Claude, Perplexity and Gemini really see (2025)
- Vercel: The rise of the AI crawler (Dec 2024)
About the author: This article was prepared by the Insightland SEO and GEO team. Insightland is a Wrocław-based agency, part of the Hexe Capital Tech Hub, that connects business, tech and data: SEO and GEO, performance marketing, UX and analytics, and SEO and AI training. Contact us.