Structured data for SEO in 2025–2026: what changed and what matters now

Google has aggressively pruned underperforming structured data types while doubling down on e-commerce markup and AI-readable entity signals. The structured data landscape has shifted fundamentally: FAQ and HowTo rich results are effectively gone for most sites, seven additional types were deprecated in June 2025, and structured data's primary value proposition is expanding from "win rich results" to "make your content legible to AI systems."

Core types like Product, Article, Organization, and Review remain fully supported and have been significantly enhanced. Schema.org reached version 30.0 in March 2026 with 823 types and 1,529 properties, while new protocols like NLWeb position structured data as the foundational layer of the emerging AI-agent web.

1. Google's supported structured data types: the full 2026 picture

Google's Search Gallery (last updated January 6, 2026) lists 26 active structured data types for rich results. Here is the current status of every major type:

New types added since early 2023

• Discussion Forum (Nov 2023) — for threaded user-generated content; Google notably recommends Microdata/RDFa here (not JSON-LD) to avoid duplicating large text blocks
• Profile Page (Nov 2023) — for pages about a single creator; feeds into Perspectives search feature
• Vacation Rental (Dec 2023) — requires Google Hotel Center connection
• Product Variants (Feb 2024) — ProductGroup schema for sizes, colors, and other variations
• Loyalty Program / MemberProgram (Jun 2025) — highlights exclusive pricing and membership rewards; available in 8 countries
• Structured Data Carousels (beta) — list-like rich results using ItemList; available only in EEA countries, Turkey, and South Africa

Types deprecated since 2023

The June 12, 2025 mass deprecation removed seven types in one sweep: Book Actions, Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, and Vehicle Listing. Google cited low usage and limited user value. Practice Problems followed in November 2025, and the Sitelinks Search Box was removed in November 2024.

John Mueller addressed community concern directly on Reddit: "Google is not killing schema" — these are routine retirements of underperforming features. Rankings are unaffected; only visual enhancements disappear.

2. The FAQ and HowTo story: a cautionary tale with a silver lining

The most consequential structured data change was Google's August 8, 2023 announcement restricting FAQ and HowTo rich results.

HowTo was hit hardest. Google first limited HowTo rich results to desktop only on August 8, then fully deprecated HowTo on September 13, 2023 — removed from all devices, documentation deleted, Search Console report eliminated. HowTo markup is now completely inert in Google Search.

FAQ was restricted, not removed. FAQPage rich results are now shown exclusively for well-known, authoritative government and health websites. For everyone else, the markup produces no visible rich results in Google. However, Google explicitly kept FAQ as a supported feature type during the June 2025 deprecation wave, signaling it retains internal value.

The emerging consensus is that FAQ schema is still worth implementing, but for entirely different reasons than before. Pages with FAQPage markup are reportedly more likely to appear in Google AI Overviews. FAQ schema continues to work for Bing rich results, feeds voice search answers, and structures content in ways AI systems find extractable.

The practical recommendation: keep FAQ markup on genuinely useful FAQ content, remove it from pages where it was purely a SERP-gaming tactic. For HowTo, the markup can remain without harm but delivers zero Google Search benefit.

3. Testing tools: two tools replaced one

The old Google Structured Data Testing Tool (SDTT) no longer exists. It was migrated into two distinct tools:

Google Rich Results Test (search.google.com/test/rich-results) is the primary tool for Google-specific validation. It tests whether your markup qualifies for Google rich results, provides previews of how results may appear, and supports URL and code-snippet testing. It only validates types Google actively supports.

Schema Markup Validator (validator.schema.org) validates all Schema.org types, not just Google-eligible ones. It's vendor-neutral, maintained by Google as a community service, and based on the original SDTT codebase. It supports JSON-LD 1.0, RDFa 1.1, and Microdata.

The recommended workflow: Use the Schema Markup Validator first to catch syntax errors, then the Rich Results Test to confirm Google eligibility. Search Console's Enhancement Reports provide ongoing monitoring at scale.

4. JSON-LD remains Google's clear recommendation

Google's position has not changed: JSON-LD is the recommended format. Their documentation states: "Google recommends using JSON-LD for structured data if your site's setup allows it, as it's the easiest solution for website owners to implement and maintain at scale."

JSON-LD's advantages are practical: separation from HTML means no interleaving with page content, it can be placed anywhere in the document, it supports dynamic injection via JavaScript, and it's less error-prone at scale. The one notable exception is Discussion Forum structured data, where Google actually recommends Microdata or RDFa to avoid duplicating large text blocks in a separate JSON-LD script.

A basic example:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Company",
  "url": "https://example.com",
  "logo": "https://example.com/logo.png",
  "sameAs": [
    "https://www.linkedin.com/company/your-company",
    "https://twitter.com/yourcompany"
  ]
}

RDFa remains relevant for combining multiple vocabularies beyond Schema.org, and Microdata persists in some CMS default implementations. Neither is deprecated, but neither is preferred.

5. Structured data's expanding role in AI search

This is the most significant strategic shift. Structured data has evolved from "earn rich results" to "make your content machine-legible for AI systems."

Confirmed by platforms:

Google published a May 2025 blog post explicitly stating structured data is "useful for sharing information about your content in a machine-readable way that our systems consider" for AI experiences in Search. At SMX Munich in March 2025, Fabrice Canel, Microsoft's Principal Product Manager for Bing, confirmed that "schema markup helps Microsoft's LLMs understand your content." ChatGPT has confirmed using structured data for product results.

What's less certain:

A rigorous March 2026 analysis by Aimee Jurenka provided essential nuance: "The leap from 'LLMs can process structured data' to 'web schema markup improves AI search visibility' requires assumptions we can't verify for most platforms." For Google AI Overviews and Bing Copilot, the connection is confirmed. For ChatGPT, Perplexity, and Claude, no official implementation details exist.

Testing has revealed that during live page fetches, most AI chatbots ignore JSON-LD entirely and extract only visible HTML content. Schema's AI value likely comes during the indexing phase, not real-time fetching.

The NLWeb protocol changes everything. Announced at Microsoft Build 2025 by R.V. Guha (creator of Schema.org himself), NLWeb turns any website into a conversational interface for AI agents. It is built directly on Schema.org — JSON-LD markup is loaded into NLWeb's MCP (Model Context Protocol) server, which then exposes REST API and MCP endpoints for natural-language queries. Early adopters include Shopify, Allrecipes, and Tripadvisor.

Practical implication: Schema markup is becoming the semantic infrastructure layer that AI systems — whether search engines, chatbots, or autonomous agents — use to understand web content. This is a fundamentally different value proposition than rich result eligibility alone.

6. New search features built on structured data

Several significant features launched since 2023:

Merchant listing enhancements represent Google's most active development area. Product structured data now supports variants (ProductGroup), loyalty programs (MemberProgram), certification markup, 3D model references, and organization-level return and shipping policies. The returnPolicyCountry field became mandatory in March 2025.

Google also warns that dynamically-generated Product markup makes Shopping crawls less frequent and less reliable — server-side rendering is critical for fast-changing prices and availability.

Discussion Forum and Profile Page types (both November 2023) feed Google's Perspectives feature, surfacing user-generated content and creator-focused results.

Structured Data Carousels (beta, 2024) combine ItemList with types like Hotel, VacationRental, LocalBusiness, Restaurant, Event, and Product to create visually rich list-based results. Currently limited to EEA countries, Turkey, and South Africa — worth monitoring for potential global expansion.

Organization schema received its biggest update ever in November 2023, replacing the old Logo structured data type entirely. It now supports legal name, description, address, contact information, company identifiers, employee count, founding date, and social profile links via sameAs.

7. Schema.org in 2026: version 30.0 and the interoperability push

Schema.org reached version 30.0 on March 19, 2026, comprising 823 types, 1,529 properties, 96 enumerations, and 535 enumeration members. Over 45 million web domains now use Schema.org markup, generating more than 450 billion Schema.org objects.

The most significant trend across 2024–2025 releases is interoperability with other vocabularies. Version 30.0 added equivalence annotations and exports for GS1, Dublin Core, and Open Graph — a major step toward making Schema.org a bridge vocabulary across standards.

Notable recent additions:

• Version 29.4 (December 2025) — ConferenceEvent, PerformingArtsEvent, OnlineMarketplace, InstantaneousEvent, and authentication actions (LoginAction, ResetPasswordAction)
• Version 29.2 (May 2025) — Organization properties for company registration, legal address, and legal representative

The broader trajectory is clear: Schema.org is evolving from a search engine optimization vocabulary into a universal machine-readable language for the AI-agent ecosystem.

8. Google Search Console structured data reports

Search Console currently provides Enhancement Reports for all active structured data types. Available reports include: Article, Breadcrumb, Discussion Forum, Education Q&A, Employer Aggregate Rating, Event, FAQ, Image Metadata, Job Posting, Local Business, Math Solver, Merchant Listings, Organization, Product Snippets, Profile Page, Q&A, Recipe, Review Snippet, Software App, Speakable, Vacation Rental, and Video.

The Unparsable Structured Data report aggregates syntax errors that prevent Google from identifying feature type — an often-overlooked diagnostic tool.

Reports removed on September 9, 2025 (following the June deprecation): Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, and Vehicle Listing. Practice Problem was removed in January 2026.

9. Implementation best practices have shifted toward server-side rendering

Server-side rendered JSON-LD is the gold standard for 2026. Google's documentation states that "dynamic rendering was a workaround and not a long-term solution" and recommends server-side rendering, static rendering, or hydration.

Google Tag Manager remains officially supported for structured data deployment. However, GTM adds JavaScript-rendering dependency, introduces mismatch risk between page content and injected schema, and makes Shopping crawls "less frequent and less reliable."

The expert consensus: use GTM for supplementary or non-critical schema; use hardcoded server-side JSON-LD for Product, Organization, and LocalBusiness markup.

The @graph and @id patterns

Using @graph wraps multiple entities into a single JSON-LD block, while @id creates unique identifiers enabling cross-page entity referencing:

{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "@id": "https://example.com/#organization",
      "name": "Your Company",
      "url": "https://example.com"
    },
    {
      "@type": "WebPage",
      "@id": "https://example.com/about/#webpage",
      "url": "https://example.com/about/",
      "isPartOf": { "@id": "https://example.com/#organization" }
    }
  ]
}

This pattern builds what practitioners call a "Content Knowledge Graph" — interconnected entities that AI systems can traverse.

Entity SEO

The recommended implementation hierarchy:

• Entity-level schema (Organization, Person, LocalBusiness) — "who and what this is"
• Content-level schema (Article, Product, FAQPage, Recipe) — "what this page does"
• Relationship schema (sameAs, mainEntityOfPage, about, isPartOf) — connections between entities and to external knowledge bases like Wikipedia and Wikidata

Google's Knowledge Graph now contains over 500 billion facts about 5 billion entities. Using sameAs to link your Organization to Wikipedia, Wikidata, and social profiles is how you connect your structured data to that graph.

10. Review stars and aggregate ratings

The rules governing review stars haven't changed substantially since the September 2019 self-serving review policy, but enforcement remains strict. LocalBusiness and Organization types are ineligible for star review rich results when the entity controls the reviews on its own website — this includes third-party review widgets embedded on your own site.

Only independent third-party review sites (Yelp, TripAdvisor, Trustpilot) can display review stars for businesses they review.

Products and services offered by a business can still carry legitimate review markup on the business's own site. The distinction is between reviewing the business entity (restricted) versus reviewing a specific product the business sells (permitted).

Eligible types for review snippets remain: Book, Course, CreativeWorkSeason/Series, Episode, Event, Game, HowTo, LocalBusiness (third-party only), MediaObject, Movie, MusicPlaylist, MusicRecording, Organization (third-party only), Product, Recipe, and SoftwareApplication.