Schema Markup and Structured Data Guide 2026: Boost SEO, AEO, and AI Visibility

TL;DR

• Google deprecates markup types, not schema: FAQ rich results stopped appearing in Google Search on May 7, 2026, and seven other types (Book Actions, Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, Vehicle Listing) were retired in June 2025. Yet John Mueller’s 2025 Reddit clarification still stands as the definition of this era: “Understand that markup types come and go, but a precious few you should hold on to, like title and meta robots.” Schema isn’t dying the checklist is.
• Structured data is optional for AI answers, still required for rich results: Google’s official generative AI optimization guide (updated July 10, 2026) states plainly: “Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add. However, it’s a good idea to continue using it as part of your overall SEO strategy, as it helps with being eligible for rich results on Google Search.”
• Don’t expect schema to move AI citations on pages AI already sees: Ahrefs tracked 1,885 pages that added JSON-LD between August 2025 and March 2026 against 4,000 control pages. AI Overviews citations fell 4.6% (small but statistically significant), AI Mode was +2.4% and ChatGPT +2.2% both indistinguishable from noise. “Adding schema produced no major uplift in citations on any platform.”
• The deploy-vs-valid gap is the biggest 2026 opportunity: A 5,000-site audit found 71% of sites deploy at least one schema type but only 22% pass Google’s Rich Results Test cleanly a 49-point gap. Valid schema correlates with AI citation frequency at +0.34 Pearson, and “three correctly-combined schemas beat five weakly-validated ones every single time.”
• Ship the combinations with evidence, not the whole vocabulary: Article + BreadcrumbList delivered +47% more AI Overview citations on informational queries; Product + Offer, +29% on commercial queries; Organization + WebSite with SearchAction, +18% on brand queries.
• AI Mode passed one billion monthly users (Google I/O, May 19, 2026), and the Search Console Generative AI performance reports launched June 3, 2026 you now have first-party data on impressions in AI Overviews and AI Mode. Schema work in 2026 is measured against that report, not against featured snippets alone.

What this guide covers

1. Why structured data still earns its bytes in 2026
2. How schema works under the hood: JSON-LD, graphs, and entities
3. The 2026 schema stack: which vocabularies matter most
4. The priority implementation matrix: schema type versus payoff
5. Real JSON-LD blocks for the four core builds
6. The implementation workflow: build, validate, deploy, monitor
7. Advanced schema: entity optimization and the LLM-visibility playbook
8. Schema mistakes and debugging: disaster-proofing your markup
9. Frequently asked questions
10. Sources and references

Why structured data still earns its bytes in 2026

Let’s be honest about what 2026 looks like from a search marketer’s chair. Between June 2025 and June 2026, Google retired a substantial set of rich result features and markup types. The FAQ rich result the most famously “free SERP real estate” in modern SEO stopped appearing entirely on May 7, 2026. The HowTo rich result had been gone for most sites since 2023. The June 2025 wave took out Book Actions, Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, and Vehicle Listing, with Search Console reporting dropped in September 2025 and API support ending in January 2026. Practice Problems tooling followed in January 2026. Then Google announced in November 2025 that it was “constantly working to simplify the search results page,” and a wave of panic posts asked whether structured data itself was being killed.

It wasn’t. Google’s own documentation still lists JSON-LD, Microdata, and RDFa as supported formats, and a note in the retirement announcement was blunt: “This update won’t affect how pages are ranked.” What died was the idea that schema is a decoration layer you bolt onto every page. What survived and grew is schema as a data layer that machines actually read. Here’s the 2026 case in three parts.

Rich results are fewer but still lucrative

There are roughly 14 structured data types Google still rewards with visible rich results in 2026, down from a much bigger list: Article (and NewsArticle, BlogPosting), Breadcrumb, Product (with Merchant Listing and Product Variants), Recipe, Event, LocalBusiness, JobPosting, Video, Organization, Speakable, Return Policy, Shipping Policy, Loyalty Program, and Carousel. The commercial ones carry real revenue. Product snippets can show ratings, price, availability, shipping, returns, pros-and-cons and price-drop badges; Recipe cards remain one of the highest-CTR formats on the web; the local pack and merchant knowledge panels run on the same markup.

Estimates of the payoff are consistently large: 58% of clicks on SERPs go to rich results versus 41% for standard blue links, and CTR benchmarks for the surviving formats run at +30–35% for product snippets and +40–50% for recipe cards. Google never guarantees any enhancement will display “Search result enhancements are shown at the discretion of each experience” but without structured data the eligibility question is settled before it’s ever asked.

Google is explicit about where schema helps AI

Google published its first official generative AI search optimization guide in May 2026 (updated July 10, 2026), and its position on structured data is unusually specific:

“Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add. However, it’s a good idea to continue using it as part of your overall SEO strategy, as it helps with being eligible for rich results on Google Search.”

Read that again. No special markup for AI but you should keep doing it, because AI features on Google Search are grounded in the same index, the same ranking systems, and the same entity understanding as classic Search. Google also says product listings and local business info can appear in generative AI responses when you feed them through Merchant Center feeds and Google Business Profiles, and it advises ignoring the “llms.txt and other special markup” trend entirely Google Search doesn’t use those files.

Google’s own people have said this more bluntly. At Search Central Zurich in late 2025, Google told the audience that structured data “continues to be important. In particular, for data with authoritative meaning and regulatory information, like prices in shopping, it’s very important” (as reported by Aleyda Solis). Microsoft has been even more explicit: at SMX Munich in March 2025, Fabrice Canel, Principal Product Manager at Bing, confirmed that schema markup helps Microsoft’s LLMs understand content. That’s the machine-readable precision problem in one sentence: a model can infer “about $50” from prose, but it can’t reliably extract “$49.99, in stock, ships free” schema can.

The inconvenient evidence about citations

Now the part most 2025-era guides skip. There is a real possibility you’ve been told schema gets you cited by ChatGPT, Perplexity, and Gemini. The 2026 data says: it depends, and mostly not.

Ahrefs ran two studies back to back. First, they analyzed 6 million URLs and found AI-cited pages were almost three times more likely to have JSON-LD than non-cited pages 53% of AI-cited pages run schema. That’s the correlation that launched a thousand case studies. So they ran the experiment the industry needed: they tracked 1,885 pages that added JSON-LD between August 2025 and March 2026, matched them against 4,000 control pages with similar citation levels that never added schema, and ran a difference-in-differences analysis across Google AI Overviews, AI Mode, and ChatGPT. Result: AI Overviews −4.6% (statistically significant, but tiny and the trend was already declining for both groups), AI Mode +2.4%, ChatGPT +2.2% “Adding schema produced no major uplift in citations on any platform,” and “schema had no clear positive or negative effect.”

Two caveats frame the finding. Ahrefs studied pages that were already being cited heavily (100+ AI Overview citations) pages already inside the LLM consideration set. Their suggestion, echoed by a searchVIU experiment showing that ChatGPT, Claude, Perplexity, Gemini and Google AI Mode all ignored JSON-LD during live retrieval: schema may still matter for getting crawled, parsed, and indexed in the first place. Meanwhile, Search Engine Land’s own controlled test (September 2025) built three near-identical single-page sites good schema, bad schema, no schema and only the page with well-implemented schema appeared in an AI Overview and reached Position 3. Their authors were careful: “promising, but inconclusive.”

So the honest 2026 position is this: schema is a rich-results play, a precision-data play, and an entity-disambiguation play not an AI-citation cheat code. If you’re already ranking and already being cited, adding JSON-LD won’t lift you. If your content is invisible to machines, schema plus everything else (technical SEO, authority, quality) is what gets you in the room.

How schema works under the hood: JSON-LD, graphs, and entities

If you’re going to use schema in 2026, use it properly. That means understanding what the markup actually is.

Schema.org is the shared vocabulary maintained by Google, Microsoft, Yahoo, and Yandex a set of types (things) and properties (attributes) that describe entities. As of version 30.0, released March 19, 2026, the full vocabulary covers 800+ types, though only a small fraction map to Google rich results. Two new realities in 2026:

• Schema.org now publishes usage statistics. Announced June 4, 2026, the dataset measures term frequency across Google’s public web crawling infrastructure, aggregates it at domain level, buckets it into ranges (e.g., “100K–1M domains”), and updates monthly. Before you invest in a niche type, you can now check how many domains actually use it.
• A handful of types are used by 10 million or more domains each WebSite, WebPage, Organization, Person, and ImageObject among them. Overall, industry estimates suggest 45M+ domains use schema markup in 2026, up from roughly 10M in 2020.

JSON-LD is the language

Google supports three formats JSON-LD (recommended), Microdata, and RDFa but for new work there’s only one serious choice. JSON-LD is a<script type="application/ld+json"> block you can drop anywhere in the head or body. It’s cleanly separated from your HTML, it’s harder to get wrong, and the AI-era parsers are tuned for it. It’s also what Google’s guidance recommends and what most CMS plugins output.

A minimal block looks like this:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "@id": "https://www.example.com/#org",
  "name": "Example, Inc.",
  "url": "https://www.example.com/"
}

That’s three keywords you must internalize:

• @context points at the vocabulary; alwayshttps://schema.org.
• @type the entity type, e.g.Organization,Product,Article.
• @id the entity’s identifier. This is the 2026 skill. Give each real-world entity a stable@id (usually its canonical URL plus a fragment), then reference that ID from every page instead of re-describing the entity. When your product page, your homepage, and your sitemap all point athttps://www.example.com/#org, machines build one coherent graph instead of three conflicting descriptions of “Example, Inc.”

Using @graph for multiple entities per page

Most pages describe several entities: a blog post is authored by a person, published by an organization, belongs to a category, and sits in a breadcrumb trail. Instead of shipping five separate script blocks, ship one@graph array:

{
  "@context": "https://schema.org",
  "@graph": [
    { "@type": "Person", "@id": "https://www.example.com/#alex", "name": "Alex Dunmore" },
    { "@type": "Organization", "@id": "https://www.example.com/#org", "name": "Example, Inc." },
    { "@type": "Article", "@id": "https://www.example.com/blog/post#article",
      "headline": "Post title", "author": { "@id": "https://www.example.com/#alex" },
      "publisher": { "@id": "https://www.example.com/#org" } }
  ]
}

Inside a graph, entities reference each other by ID, and the whole page becomes a machine-readable statement: this article was written by this person and published by this organization. That’s the difference between markup that validates and markup that builds an entity graph. Search Engine Land’s agentic-web analysis calls this “thinking about your schema as a site-level graph” auditing where entity definitions conflict across URLs and keeping Organization and Person markup consistent site-wide.

One principle governs everything else: schema is a statement about the page, not a wish about the SERP. Google’s general guidelines demand that the structured data describe content visible on the page. If the markup describes a performer, the HTML body must describe that same performer. Everything else validity, trust, citation follows from that.

The 2026 schema stack: which vocabularies matter most

Most guides list schema types alphabetically. That’s the 2018 approach, and 2026 has no patience for it. What matters is the stack: the layers every commercially sensible site should ship, matched to the page type and the job you need done.

Layer 1 Site-wide entity baseline (every site):Organization (or the most specific subtype Google explicitly recommendsOnlineStore for ecommerce and specificLocalBusiness subtypes for physical businesses),WebSite, andBreadcrumbList on every non-homepage page. Google’s Organization docs confirm there are no required properties “we recommend adding as many properties that are relevant” but the load-bearing ones arelogo,sameAs,name, andalternateName, withlegalName,contactPoint,iso6523, andnaics used behind the scenes for disambiguation. Logo matters: minimum 112×112px, crawlable, indexable, and designed to look right on a white background.

Layer 2 Content layer (publishers, blogs, SaaS):Article (withBlogPosting for opinion andNewsArticle for news), nestedPerson author withsameAs,datePublished/dateModified,image. Google reads these for top-stories carousels, the dateline under blue links, Discover (which leans on a 1200px-wide image), and AI citation bylines. In the 5,000-site audit, valid Article + BreadcrumbList was the strongest single combination in the data for AI Overview citations on informational queries.

Layer 3 Commercial layer (ecommerce, SaaS pricing):Product with a nestedOffer (price, priceCurrency, availability required for the price-and-availability rich result),brand,sku or GTIN, and genuineaggregateRating. AddProductGroup +hasVariant for size/color variations instead of one URL per SKU, and nest merchant policies (return policy, shipping, loyalty program) inside Organization or on the offer. Two 2026 specifics: the price-drop enhancement requiresOffer, notAggregateOffer; and product pages where customers can purchase need the merchant-listings class of markup (which also accepts apparel sizing, shipping details, and return policy), while review sites and editorial product reviews belong in the product-snippets class. Feeding Merchant Center can fill any gaps Google notes product snippets may pull pricing from your merchant feed if it’s not in the page markup.

Layer 4 Local layer (physical business):LocalBusiness with the most specific subtype you can find (Restaurant,Dentist,AutoRepair, and so on you can also use an array of types). Required:address andname. Recommended:geo (5 decimal places or more),openingHoursSpecification,telephone,url,priceRange (under 100 characters or Google won’t show it),department for big-box stores. Local schema matters more in 2026, not less: Google expands agentic booking to local experiences, and for select categories like home repair, beauty, and pet care, Google can call businesses on the user’s behalf. Machine-readable address, hours, and service data is precisely what those systems need.

Layer 5 Niche (ship only if it fits):VideoObject (thumbnail, duration, contentUrl or embedUrl, uploadDate transcripts improve eligibility),Event,Recipe,JobPosting,ProfilePage for author pages,QAPage for Q&A with multiple perspectives,Speakable for news-style content targeting Google Home (still a US-English beta that returns up to three articles per topical query),Dataset,DiscussionForum, andSoftwareApplication.

And the do-not-ship list for 2026. FAQ rich results are gone (May 7, 2026). HowTo rich results are gone for everyone since the 2023 changes. Book Actions, Course Info, Claim Review (except for approved fact-check publisher programs), Estimated Salary, Learning Video, Special Announcement, and Vehicle Listing were retired June 12, 2025. Sitelinks search box markup was sunset in 2024 Google generates sitelinks automatically now. Practice Problem tooling support ended January 2026. If your CMS plugin is emitting any of these by default, turn it off; you’re spending crawl time and bytes on formats that can no longer pay.

The priority implementation matrix: schema type versus payoff

Use this to decide what to ship. “Payoff” balances direct rich result value, AI-system value, effort, and current status. The validation and AI-citation figures come from the April 2026 Digital Applied 5,000-site audit, the Retired-format statuses from Google’s documentation and 2025–2026 announcements, and CTR estimates from 2026 rich-result benchmark analyses.

Schema type	What it gets you in 2026	Effort	When to use
Organization + sameAs	Knowledge panel and merchant knowledge panel eligibility; the entity anchor machines use to disambiguate your brand; logo control; +18% brand-query AI citations when paired with WebSite; 4–12 weeks typical knowledge panel lag	Low	Every site, one block, homepage or site-wide via template
WebSite	Site name in Search; powers the site entity that pairs with Organization	Trivial	Every site
BreadcrumbList	Breadcrumb path display in SERPs; clean site topology for AI extractors; strongest single citation combo with Article (+47%)	Low	Every page below the homepage
Article / BlogPosting / NewsArticle	Top Stories eligibility, Enhancements dateline, Discover reach, author bylines; the core citation combo (+47% with BreadcrumbList)	Low	Every editorial page
Person + ProfilePage	Author entity and E-E-A-T signals; binds bylines to real identities	Low	Editorial sites, blogs
Product + Offer	Product snippets, ratings, price, availability, shipping, returns, price-drop badge, merchant listings, Shopping eligibility; +29% commercial AI citations when Offer is valid	Medium	Every purchasable product page
ProductGroup / hasVariant	Variant support without one URL per SKU	Medium	Size/color/configuration catalogs
MerchantReturnPolicy / hasShippingService / hasMemberProgram	Trust signals on product cards and merchant knowledge panel; can also be set in Search Console since Nov 2025	Medium	Ecommerce sites
LocalBusiness (specific subtype)	Local pack, Maps, knowledge panel, AI answers for local queries, agentic-calling readiness	Medium	Any physical location
FAQPage	No rich results since May 7, 2026. Still valid markup that structures Q&A for LLM retrieval keep genuine FAQs, drop the SERP-play mindset	Low	Only genuine FAQ pages with definitive answers; use QAPage for multi-perspective Q&A
VideoObject	Video rich results (thumbnail, duration, upload date), stronger video snippet chances	Medium	Pages with meaningful video
AggregateRating / Review	Star ratings; 2026 rule: no self-serving reviews a review must be genuinely independent, withreviewedBy/sourceOrganization pointing at a real reviewer	Medium	Only where you truly have third-party reviews
Recipe / Event / JobPosting / Course	Vertical rich results (recipe cards are the highest-CTR format measured)	Medium	Matching business only
Speakable	Audio-distribution for Google Home news queries (US English beta)	Low	News publishers
Deprecated list	Book Actions, Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, Vehicle Listing, FAQ/HowTo rich result features, Sitelinks search box, Practice Problems		Don’t ship; existing harmless markup can stay

How to use this matrix. Week one: Organization + WebSite + BreadcrumbList as the baseline (or remove the ones your plugin duplicates). Week two: your content layer Article + Person on every post. Week three: the commercial or local layer wherever you sell or serve. That is the “three correctly-combined schemas” pattern; per the audit, only 8% of sites ship five or more types correctly, and that tiny Tier 1 cohort dominates rich results and AI citations.

Real JSON-LD blocks for the four core builds

Copy these, replace the values, and keep every value identical to what’s visible on the page Google’s guidelines and AI parsers both punish mismatch.

Build 1: Organization (+ WebSite) for the homepage

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "@id": "https://www.example.com/#org",
      "name": "Example, Inc.",
      "alternateName": "ExampleCo",
      "url": "https://www.example.com/",
      "logo": {
        "@type": "ImageObject",
        "url": "https://www.example.com/logo.png"
      },
      "sameAs": [
        "https://www.linkedin.com/company/example",
        "https://en.wikipedia.org/wiki/Example_Inc",
        "https://www.wikidata.org/entity/Q1234567",
        "https://x.com/example"
      ],
      "contactPoint": {
        "@type": "ContactPoint",
        "contactType": "customer service",
        "telephone": "+1-555-123-4567",
        "email": "help@example.com"
      }
    },
    {
      "@type": "WebSite",
      "@id": "https://www.example.com/#website",
      "url": "https://www.example.com/",
      "name": "Example, Inc.",
      "publisher": { "@id": "https://www.example.com/#org" }
    }
  ]
}
</script>

sameAs is the field that does the work see the advanced section below. If you’re an ecommerce store, use@type: "OnlineStore" (per Google’s recommendation) and nesthasMerchantReturnPolicy andhasShippingService here.

Build 2: Product + Offer for a product page

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "@id": "https://www.example.com/shop/coffee-maker#product",
  "name": "AeroPress Coffee Maker",
  "description": "A compact manual coffee press for home brewing.",
  "image": [
    "https://www.example.com/shop/coffee-maker/1x1.jpg",
    "https://www.example.com/shop/coffee-maker/16x9.jpg"
  ],
  "sku": "AP-COFFEE",
  "brand": { "@type": "Brand", "name": "Example Home" },
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.7",
    "reviewCount": "231"
  },
  "offers": {
    "@type": "Offer",
    "url": "https://www.example.com/shop/coffee-maker",
    "priceCurrency": "USD",
    "price": "49.99",
    "availability": "https://schema.org/InStock",
    "itemCondition": "https://schema.org/NewCondition",
    "priceValidUntil": "2026-12-31"
  }
}
</script>

Required for product snippets:name, plus at least one ofreview,aggregateRating, oroffers. UseOffer (notAggregateOffer) if you want price-drop eligibility, and keepprice/availability exactly in sync with the page and your Merchant Center feed.

Build 3: Article + author for a blog post

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "@id": "https://www.example.com/blog/2026-schema-guide#article",
  "headline": "Schema Markup and Structured Data Guide 2026",
  "description": "A practical vocabulary-by-vocabulary schema handbook for 2026.",
  "image": "https://www.example.com/blog/2026-schema-guide/hero.jpg",
  "datePublished": "2026-08-27",
  "dateModified": "2026-08-27",
  "author": {
    "@type": "Person",
    "@id": "https://www.example.com/#author-alex",
    "name": "Alex Dunmore",
    "url": "https://www.example.com/authors/alex-dunmore",
    "sameAs": [
      "https://www.linkedin.com/in/alexdunmore"
    ]
  },
  "publisher": { "@id": "https://www.example.com/#org" },
  "mainEntityOfPage": {
    "@type": "WebPage",
    "@id": "https://www.example.com/blog/2026-schema-guide"
  }
}
</script>

Use@id references instead of re-declaring the publisher that’s what connects this page to the homepage graph. UseNewsArticle on news sites,BlogPosting for opinion pieces.

Build 4: LocalBusiness for a single location

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Restaurant",
  "name": "Example Bistro",
  "image": "https://www.example.com/bistro.jpg",
  "url": "https://www.example.com/",
  "telephone": "+1-555-123-4567",
  "priceRange": "$$",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "148 W 51st St",
    "addressLocality": "New York",
    "addressRegion": "NY",
    "postalCode": "10019",
    "addressCountry": "US"
  },
  "geo": {
    "@type": "GeoCoordinates",
    "latitude": "40.7608047",
    "longitude": "-73.9843667"
  },
  "openingHoursSpecification": {
    "@type": "OpeningHoursSpecification",
    "dayOfWeek": ["Tuesday", "Wednesday", "Thursday", "Friday", "Saturday"],
    "opens": "17:00",
    "closes": "23:00"
  }
}
</script>

Correct the@type to your specific subtype. Keep NAP (name, address, phone) identical to your Google Business Profile and local directories when schema contradicts GBP data, citations, or reviews, Google “discounts the markup and often ignores the information altogether,” and multiple compounding conflicts can cost visibility site-wide.

The implementation workflow: build, validate, deploy, monitor

Follow the same pipeline you’d use for a code release because this is code. The Digital Applied audit found almost no team had continuous validation beyond a one-off Rich Results Test on launch day, and five error patterns account for over 90% of all schema failures. Validation is not a pre-launch checkbox; it’s a build step.

1. Build

Three ways to generate markup, in order of reliability:

• CMS plugins and automation-first platforms the default for most sites. Adoption by CMS in the audit: WordPress 78% (Yoast/Rank Math/Schema Pro ecosystem), Shopify 89% for Product schema via theme defaults, Webflow 41%, Framer 27%, custom HTML 19%. The catch: defaults are thin only 31% of Shopify stores pair Product with Organization, and only 19% include theOffer object Google requires for price-and-availability.
• AI-assisted generation useful for scale, but treat the output as a draft. Complete property lists, verify against Google’s docs, and validate before it ships.
• Manual JSON-LD lowest adoption, highest validity per instance. Best for the four core builds above and anything custom.

Whatever the method, “completeness over coverage” is the 2026 rule: fully populated markup on your most important pages beats thin markup everywhere. An incomplete product schema signals uncertainty to AI agents; a complete one signals reliability.

2. Validate

Every block, before it touches production:

1. Rich Results Test (search.google.com/test/rich-results) past the URL or code. Note: FAQ support was removed from this tool in June 2026; expect no FAQ feedback.
2. URL Inspection in Search Console validates what Google’s crawler actually saw, which is the ground truth.
3. Schema.org validator (validator.schema.org) validates against the full vocabulary, not just Google’s features.
4. Structured data report in Search Console page-level errors and warnings across the site (the FAQ enhancement report was removed there too).
5. Assertions for anything your stack renders dynamically a tiny JSON-LD validator in your CI pipeline catches missing-required-property regressions before they ship.

Validation and deployment checklist

• Every@type has all required properties populated from visible page content. -price,availability, and dates match the page exactly, anddateModified changes when content changes.
• Values use the correct types ISO 8601 dates (2026-08-27), full URLs,schema.org range values likehttps://schema.org/InStock.
• One@id per real-world entity; the same@id used on every page that mentions it. -@graph contains the full set of entities for the page instead of duplicated script blocks.
• Logo is ≥112×112px, crawlable, and indexable; geo coordinates use 5+ decimal places.
• Schema is server-rendered (or follows Google’s JavaScript-generation guidance) schema hidden in<noscript> is not parsed, and AI crawlers treat JS-injected markup differently from server-rendered markup.
• No deprecated types in the output (FAQ/HowTo rich result formats, the seven June 2025 retirements, Practice Problems, Sitelinks search box).
• Zero errors, and warnings triaged: missing recommended fields are legitimate backlog; blocking warnings are release blockers.

3. Deploy

Ship it as part of templates, not one-off edits. Schema belongs in your CMS’s head injection for each page type one rule per template, which keeps coverage at 100% by construction. If you change a template, re-validate. Pair the markup with crawlable HTML, because content that can’t be crawled can’t be an entity: Google’s AI guidance says generative AI features use publicly accessible, crawlable content, and a page must be indexed and eligible for a normal snippet to be eligible for AI features.

4. Monitor

Monitoring is where most teams fail. Run a monthly rhythm:

• Search Console structured data report watch error counts per type.
• Search Console Generative AI performance report launched June 3, 2026; dedicated views of impressions in AI Overviews and AI Mode (plus generative AI features in Discover), by page, country, device, and time granularity (hourly to monthly). It’s rolling out to a subset of sites first, and Google is still deciding which additional metrics to add.
• Doctoring against deprecations before the 2026 cutoffs, if you’d been tracking FAQ rich result performance, that reporting is now gone (June 2026) and the API field returns null (August 2026). Export historical data you need before a type loses support.
• Quarterly content-drift checks schema goes stale when content changes. Google Search Central’s update log changed structured-data eligibility multiple times across 2023, 2025, and 2026 and shows no sign of stopping.

Advanced schema: entity optimization and the LLM-visibility playbook

If the core stack is the floor, this section is the ceiling. It’s where schema stops being SEO plumbing and starts being the machine-readable identity system for your brand.

sameAs: the field that does the most work

Organization schema filled in withname,url, andlogo but nosameAs is a declaration with no verification.sameAs is how Google’s knowledge graph, and AI assistants, confirm that “Example, Inc.” in one source is the same entity as “Example” in another. Every AI system that mentions a brand resolves identity by cross-referencing sources; asameAs array pointing at Wikipedia, Wikidata, and LinkedIn gives them direct disambiguation cues.

The minimum viable array for a business: your Wikipedia article (or create one), a Wikidata item (more achievable for smaller businesses, and per practitioner guidance worth creating even if Wikipedia is a stretch), LinkedIn company page, Crunchbase, GitHub, and your verified social profiles. Google’s quality evaluations reward organizations verifiable through multiple independent sources completesameAs coverage is how you show, rather than tell. The manual check is simple: ask ChatGPT or Perplexity “What is [company name]?” If the answer doesn’t match what your owndescription says, your entity coverage isn’t complete.

The site-level graph and custom vocabularies

Audit your markup at scale, not per page: which page types have markup and which don’t, where entity definitions conflict across URLs, and whether Organization/Person markup is consistent. Then apply what Search Engine Land’s agentic-web guide recommends JSON-LD only, automation for the baseline, AI only to scale implementation, and completeness before coverage.

Custom vocabulary (extending schema.org with site-specific types) is technically possible but low-value: Google doesn’t enrich unknown types, and AI systems treat them as context at best. Invest in the standard types first; the only “extension” worth your time in 2026 is monitoring the emerging agent protocols NLWeb, Microsoft’s open-source initiative built on Schema.org by R.V. Guha (the creator of Schema.org itself he joined Microsoft as CVP and technical fellow), which lets AI agents query a website’s structured data directly in natural language, and the Universal Commerce Protocol (UCP) that Google says will let Search agents do more with commerce data.

The LLM-visibility playbook

Tactic	What it does	Effort	Evidence
Complete Organization + sameAs	Entity identity resolution across Google KG and AI assistants	Low	Knowledge panels; practitioner-reported 4–12 week lag; +18% brand citations with WebSite (Digital Applied, Apr 2026)
Product + Offer with all precision fields (price, availability, condition)	Makes machine-extractable facts out of prose; the one type Google’s own advice singles out (“prices in shopping”)	Medium	+29% commercial AI citations (Digital Applied); Google Zurich 2025 statement; Mueller, Jan 2026
Article + BreadcrumbList on every post	Gives AI systems clean provenance and site topology	Low	+47% AI Overview citations on informational queries (Digital Applied)
Genuine FAQ content (and FAQPage only where real)	Q&A structure is highly valuable to LLM retrieval even though the rich result is gone	Low	Quattr, Jul 2026: citation benefit unconfirmed, but content value remains; “keep the FAQ sections that answer real questions”
Server-rendered JSON-LD only	AI crawlers appear to ignore JS-injected and hidden markup	Low	searchVIU live-fetch test: ChatGPT, Claude, Perplexity, Gemini, AI Mode extracted only visible HTML
No stale or conflicting entities	Agents distrust schema when it contradicts the page or appears sporadically	Medium	Search Engine Land, Jun 2026: “If your schema says a product costs one price and your page displays another, agents will distrust both signals”
Measure with the new Generative AI report	First-party impressions data in AI Overviews and AI Mode	Low	Google Search Central, Jun 3, 2026
ItemList for enumerable things	Gives AI systems a clean list structure for top-N answers and rankings	Low	Schema.org docs; LLM retrieval practice

One more idea worth stealing: whether you need it or not, keep a small “schema policy” note for your team which types are standard, which are banned, and when to revisit (quarterly). The 2023–2026 cycle of retirements shows that markup decisions have shelf lives, and “treating markup as a static task can create future problems,” as the Jan 2026 Mueller guidance put it.

Schema mistakes and debugging: disaster-proofing your markup

The Digital Applied audit found that across the “deployed but broken” segment (49% of sites), five error patterns account for more than 90% of failures. Here they are, with the 2026 fixes.

Mistake 1: Missing required properties (38% of error pages). The single largest failure mode Article missingheadline,datePublished, orauthor; Product missingoffers oraggregateRating; Organization missinglogo orsameAs. Fix: checklist validation in CI, or template-level defaults.

Mistake 2: Schema hidden where parsers can’t see it (5%). Inside<noscript>, or injected client-side after DOM commit. Google’s own guidance covers JavaScript-generated structured data, but AI crawlers treat it differently, and<noscript> placement is a known dead spot. Fix: server-render or follow Google’s JS guidance precisely.

Mistake 3: Conflicting entities with overlapping@id values (5%). Two schema blocks describing the same entity with different facts on the same page price in one, different price in the other; brand name with and without “Inc.”; dates that disagree with the visible content. Google doesn’t reconcile conflicts. It discounts the markup. Fix: one@id per entity; identical facts everywhere.

Mistake 4: Markup for content that isn’t genuinely on the page. The oldest rule, still the costliest. Google’s general guidelines: “Don’t mark up content that is not visible to readers of the page.” Self-serving reviews (organizations reviewing themselves, sellers rating their own products) have been stripped from rich results since 2024, and marking up review or FAQ content that doesn’t exist on the page has historically risked a manual action. Fix: mark up what’s on the page, period.

Mistake 5: Chasing deprecated formats. The audit found 11% of sites misusing Review schema for editorial content and a cohort still shipping HowTo on content that lost eligibility years ago. In 2026, the sharper subset of this mistake is keeping FAQPage in the budget for SERP appearance that rich result is dead, and agency checklists that still say “add FAQ schema to every page” are outdated. Fix: 12-of-14 story keep the four or five types that earn their bytes; delete the rest, and watch whether anything changes (it usually doesn’t).

Debugging sequence. When a page stops getting a rich result it used to get, work this ladder: (1) Search Console URL Inspection is the data detected? (2) Rich Results Test what exactly fails, and what warnings remain? (3) validator.schema.org is it valid against the full vocabulary? (4) Search Console structured data report is this an isolated page or a template-wide regression? (5) Compare the page to a known-good version most post-update “schema breaking” is a CMS update or markup drift. And keep an eye on changes that could shift eligibility again Google’s own documentation is the schedule, so re-read the structured data gallery quarterly.

Finally, the reset mindset 2026 demands: schema is not a push-button, and “it doesn’t move the needle” is most often a symptom of invalid markup, not a verdict on the tactic. The audit’s framing is worth pinning to the wall “Schema is the only on-page SEO lever in 2026 where the gap between deployed and valid is bigger than the gap between deployed and missing.” The middle of the 2026 structured-data landscape is crowded with silently broken markup; the richest opportunity is to be the site whose markup actually validates.

Frequently asked questions

Is FAQPage schema dead? Should I delete it?

The rich result is dead, the markup is not. FAQ rich results stopped appearing on May 7, 2026; Search Console’s FAQ report and Rich Results Test support went in June 2026, and Search Console API support in August 2026. FAQPage remains a valid Schema.org type, and Google has repeatedly said unused structured data doesn’t cause problems. So: keep it on genuine FAQ pages that answer real questions, cut the FAQ sections that only existed to win a dropdown, and remember that QAPage is the right type when a question has multiple legitimate perspectives. If the only reason you added FAQ markup was SERP real estate, remove the content before you polish the markup.

Did the March 2026 core update “demote” Review and FAQ schema?

That specific claim is unverified. Core updates are broad ranking changes, and attributing structured-data-specific display effects to a particular update is not verifiable from outside Google. What remains well established is the long-standing rule: schema is not a ranking factor, and marking up content that isn’t genuinely present on the page FAQ or Review markup bolted onto pages that aren’t really about that has always risked a manual action, independent of any core update.

Should I remove HowTo schema?

No need to remove it; no reason to add it for display. HowTo rich results have been gone from mobile since 2023 and from desktop since late 2023, and HowTo remains valid schema that AI systems can still use for grounding step-by-step queries. If a page genuinely is a how-to, the markup is low-cost; if you’re stretching to fit a template, skip it.

Does schema improve AI citations for pages that are already visible?

The best available evidence says no, at least not in 30 days. Ahrefs’ matched difference-in-differences study of 1,885 pages found no major uplift on AI Overviews, AI Mode, or ChatGPT (with a small, statistically significant AI Overview decline in the treated group). Their caveat: all pages in the dataset were already heavily cited, so schema may still help pages that aren’t being seen at all. Run it yourself if you care: pick 5–10 test pages and 5–10 controls with similar citation baselines, add schema only to the test group, change nothing else, and compare after 30 days. That’s the only evidence that matters for your site.

Which validation tools should I actually use in 2026?

Google’s Rich Results Test (noting FAQ support has been removed), Search Console’s URL Inspection for what Google’s crawler genuinely saw, validator.schema.org for full-vocabulary validation, Search Console’s structured data report for site-wide error tracking, and a CI-side JSON-LD validator for anything in a rebuild pipeline. Tools claiming to show “all your rich results” are limited to schema that powers rich results use the schema.org validator to check the rest.

How long does it take for Organization schema and sameAs to change my knowledge panel?

Practitioners typically observe knowledge panel changes 4–12 weeks after adding accuratesameAs links, since Google’s knowledge graph update cycle isn’t public. AI model retrieval is faster for live-web systems (Perplexity-style) and slower for training-data-driven systems often months. And be honest about the ceiling: John Mueller, in January 2026, said schema won’t make a product rank higher or be named the “best” by AI systems; that comes from authority and real-world references. Schema supplies clarity; it doesn’t substitute for credibility.

Sources and references

1. Introducing Search Generative AI performance reports in Search Console Google Search Central Blog, June 3, 2026. https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports
2. A guide to optimizing for generative AI features on Google Search Google Search Central documentation, updated July 10, 2026. https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
3. Introduction to Product structured data Google Search Central documentation. https://developers.google.com/search/docs/appearance/structured-data/product
4. Product snippet structured data Google Search Central documentation. https://developers.google.com/search/docs/appearance/structured-data/product-snippet
5. Organization structured data Google Search Central documentation. https://developers.google.com/search/docs/appearance/structured-data/organization
6. Local Business structured data Google Search Central documentation. https://developers.google.com/search/docs/appearance/structured-data/local-business
7. Speakable (BETA) schema markup Google Search Central documentation. https://developers.google.com/search/docs/appearance/structured-data/speakable
8. General structured data guidelines Google Search Central documentation. https://developers.google.com/search/docs/appearance/structured-data/sd-policies
9. Google Search at I/O 2026: Search agents and more Google (The Keyword), May 19, 2026. https://blog.google/products-and-platforms/products/search/search-io-2026/
10. Full release summary, version 30.0 Schema.org, March 19, 2026. https://schema.org/version/latest/
11. Schema.org now shows you how many sites are using each schema type Barry Schwartz, Search Engine Land, June 10, 2026. https://searchengineland.com/schema-org-now-shows-you-how-many-sites-are-using-each-schema-type-479843
12. Google Drops FAQ Rich Results From Search Matt G. Southern, Search Engine Journal, May 10, 2026. https://www.searchenginejournal.com/google-drops-faq-rich-results-from-search/574429/
13. Adding schema didn’t boost AI citations Ahrefs Blog, May 11, 2026. https://ahrefs.com/blog/schema-ai-citations/
14. Schema Markup: What It Is & How to Implement It Ahrefs Blog. https://ahrefs.com/blog/schema-markup/
15. How to use schema markup to optimize for the agentic web Einat Hoobian-Seybold, Search Engine Land, June 1, 2026. https://searchengineland.com/schema-markup-optimize-agentic-web-479080
16. Schema and AI Overviews: Does structured data improve visibility? Molly Nogami and Ben Tannenbaum, Search Engine Land, September 23, 2025. https://searchengineland.com/schema-ai-overviews-structured-data-visibility-462353
17. The rise and fall of FAQ schema and what it means for SEO Search Engine Land, October 31, 2025. https://searchengineland.com/faq-schema-rise-fall-seo-today-463993
18. How schema markup fits into AI search without the hype Search Engine Land, March 25, 2026. https://searchengineland.com/schema-markup-ai-search-no-hype-472339
19. How structured data supports local visibility across Google and AI Search Engine Land, March 6, 2026. https://searchengineland.com/schema-local-visibility-google-ai-470906
20. Study: Adding Schema Did Not Improve AI Citations On Google, ChatGPT & More Barry Schwartz, Search Engine Roundtable, May 13, 2026. https://www.seroundtable.com/study-schema-citations-study-41311.html
21. Structured data does not help with visibility in AI search Search Engine Roundtable, September 15, 2025. https://www.seroundtable.com/structured-data-schema-ai-search-visibility-40099.html
22. Schema Markup Adoption: 5,000-Site Audit and Findings Digital Applied, April 26, 2026. https://www.digitalapplied.com/blog/schema-markup-adoption-5k-site-audit-2026
23. A 2026 Schema.org Cheatsheet for Google Rich Results and AI Citations SEO Juice, May 17, 2026. https://seojuice.com/blog/schema-org-2026-what-google-reads/
24. FAQ Schema in 2026: What’s Confirmed, What’s Not & What to Do Quattr, July 7, 2026. https://www.quattr.com/blog/faq-schema-in-2026
25. Every structured data type Google still rewards in 2026 (and the 8 that aren’t) Space and Story, May 26, 2026. https://spaceandstory.co/blog/structured-data-types-google-2026
26. John Mueller Clarifies Schema Changes Coming in 2026 Stan Ventures, November 11, 2025. https://www.stanventures.com/news/google-john-mueller-schema-update-2026-5719/
27. John Mueller Explains Where Schema Really Helps Stan Ventures, January 2, 2026. https://www.stanventures.com/news/john-mueller-explains-where-schema-really-helps-6540/
28. Google Structured Data 2026: What Changed (FAQ, How-To Removed) WebWise, July 20, 2026. https://webwise.digital/blog/google-structured-data-guidelines-2026-update
29. Google Structured Data Changes: 2026 Running Log AISchemaGen, July 26, 2026. https://www.aischemagen.com/blog/google-structured-data-changes-2026
30. Google Confirms Schema Still Matters in 2026 W3era, November 12, 2025. https://www.w3era.com/news/seo/google-confirms-schema-still-matters/
31. Schema Markup for AI Citations: What Changed in 2026 Innflows, April 9, 2026. https://www.innflows.com/blog/technology/schema-markup-ai-citations-2026
32. Google Rich Results Types: Requirements, Schema & CTR Benchmarks SchemaValidator.org, February 25, 2026. https://schemavalidator.org/guides/google-rich-results
33. Organization Schema in 2026: sameAs, Knowledge Graph, and the Fields That Actually Matter Yatna SEO Academy, February 18, 2026. https://seo.yatna.ai/seo-academy/organization-schema-2026/
34. AI Search Statistics 2026: 35+ Data Points Nico Digital, August 27, 2026. https://www.nicodigital.com/ai-search-statistics-2026/
35. Structured Data in 2026: Your AI Search Strategy Search Answer Lab, June 6, 2026. https://searchanswerlab.com/2026-structured-data-is-your-ai-search-lifeline/
36. Schema Markup Statistics 2026: 60+ Data Points & Key Findings SEOScaleUp, August 4, 2026. https://seoscaleup.com/blog/schema-markup-statistics-2026/
37. Schema Markup Didn’t Move AI Citations in Ahrefs Test Search Engine Journal, May 11, 2026. https://www.searchenginejournal.com/schema-markup-didnt-move-ai-citations-in-ahrefs-test/574568/
38. Structured Data: Google Confirms Its Importance Amid 2026 Changes and Evolution Business Tech Weekly, November 2025. https://www.businesstechweekly.com/technology-news/structured-data-google-confirms-its-importance-amid-2026-changes-and-evolution/