Definition
Schema markup is structured data, written with the Schema.org vocabulary, that labels what a page describes, such as a product with a price, an article with an author, or an organization with a logo and official profiles. It is most often added as a JSON-LD script in the page HTML, though Microdata and RDFa also work. Search engines use it to understand pages, show rich results and connect pages to entities in their knowledge graphs.
Schema.org was launched in 2011 by Google, Microsoft (Bing) and Yahoo, with Yandex joining later that year. The vocabulary is still actively maintained: version 30.1, released on September 16, 2026 according to the Schema.org release notes, added terms for EU Digital Product Passports.
What Google shows from markup has narrowed. In June 2025 Google phased out seven rich result types, including ClaimReview, Course Info, Estimated Salary, Learning Video, Special Announcement and Vehicle Listing. Practice problem results were removed in January 2026, and FAQ rich results stopped appearing on May 7, 2026, after being restricted to government and health sites since 2023. HowTo rich results were cut back in 2023. Types that still produce visible features include:
- Product and merchant listings, review snippets and shipping or return policies
- Article, VideoObject, Recipe, Event and JobPosting
- Organization (logo, contact details), BreadcrumbList and LocalBusiness
- Discussion Forum and Q&A pages, which gained new supported properties in March 2026
The AI search picture is more mixed than early GEO advice suggested. Google's May 2026 guide to generative AI features says structured data 'isn't required for generative AI search, and there's no special schema.org markup you need to add', while recommending it for rich result eligibility. Microsoft's Fabrice Canel said at SMX Munich in March 2025 that schema markup helps Microsoft's LLMs understand content. In an October 2025 searchVIU test, ChatGPT, Claude, Perplexity, Gemini and AI Mode did not extract information that existed only in JSON-LD when fetching a page live. And in a matched study published in May 2026, Ahrefs found that 1,885 pages adding JSON-LD saw no meaningful citation gain in AI Mode or ChatGPT and a small relative decline (4.6%) in AI Overviews, which Ahrefs could not attribute to schema.
The practical reading: markup helps at indexing time, for search engines and knowledge graphs that parse it, and for agent protocols such as NLWeb that return Schema.org data. It does little for an assistant reading the page live. Keep markup accurate and matching the visible text, put every important fact in the copy as well, and do not expect markup alone to win citations. Organization and Person markup with sameAs links remain the most useful types for entity SEO, and Product markup aligned with your Merchant Center feed matters for shopping surfaces.
Validate with Google's Rich Results Test and the Schema Markup Validator, and monitor Search Console's enhancement reports. The free AI SEO audit tool checks markup alongside AI crawler access. For the content side of structure, see structured content.
Examples of Schema Markup
- An online furniture store aligns the price, availability and shipping data in its Product markup with its Merchant Center feed, after Search Console flagged mismatches that kept products out of merchant listings.
- A publisher keeps its FAQPage markup after May 2026 because it causes no harm, but stops reporting on FAQ rich results and moves the effort to Article and VideoObject markup that still produces visible features.
- A SaaS company finds its pricing tiers exist only in JSON-LD on the pricing page. It adds a visible pricing table so assistants that read the page live can quote the tiers correctly.
- A recipe site validates its Recipe markup in the Rich Results Test after a template change dropped cook times from 3,000 pages, which Search Console's enhancement report showed as a sudden fall in valid items.
