Schema Markup for AI Search: The JSON-LD Your Website Needs

Schema markup is one of the most impactful things you can add to your website for AI search visibility. Microsoft confirmed at SMX Munich in March 2025 that schema markup directly helps its LLMs understand web content. GPT-4 accuracy jumps from 16% to 54% when structured data is present — a 3.4x improvement. And BrightEdge found that sites with structured data and FAQ blocks see a 44% increase in AI search citations.

Since Bing powers both Copilot and ChatGPT’s web browsing, Microsoft’s confirmation has implications far beyond Bing. Schema markup is no longer just an SEO best practice. It’s the machine-readable layer that determines whether AI systems can accurately understand, extract, and cite your content.

Source: Fabrice Canel, SMX Munich, March 2025; Data World, 2025; BrightEdge, 2025

Why Schema Matters More for AI Than for Traditional Search

Schema markup’s impact on AI search is fundamentally different from its impact on traditional search. In traditional search, schema earned you rich snippets — star ratings, FAQ dropdowns, event dates. Useful, but optional. In AI search, schema determines whether your content is understood at all.

AI chatbots don’t process JSON-LD during live fetches when answering your query. SearchVIU confirmed this in October 2025. But schema is processed at index time by Google AI Overviews and Bing Copilot. The structured data flows through the retrieval pipeline — it shapes how your content is indexed, categorized, and retrieved before a user ever asks a question.

WPRiders found that schema boosts AI-generated summary inclusion by 36%. This isn’t because the AI reads your schema in real time. It’s because schema makes your content more retrievable and more accurately represented in the indexes AI systems query.

Source: SearchVIU, October 2025; WPRiders, 2025

Organization Schema: Your Entity’s Foundation

Organization schema is the single most important schema type for AI visibility. Without it, AI systems may not know what your business is, what it does, or how to distinguish it from similarly-named entities. Our own audits have seen this fail spectacularly — Perplexity confused Toro TMS (a transportation management software company) with Toro the lawn mower manufacturer because of missing Organization schema.

The critical element is the sameAs property. This array links your website to your LinkedIn, Google Business Profile, YouTube channel, and other verified profiles. These links build your entity graph — the web of connections AI systems use to confirm your business is who it claims to be. Without sameAs, your website is an isolated node. With it, you’re a verified entity.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Company Name",
  "url": "https://yoursite.com",
  "logo": "https://yoursite.com/logo.png",
  "description": "One sentence describing what your business does.",
  "sameAs": [
    "https://www.linkedin.com/company/yourcompany",
    "https://www.youtube.com/@yourcompany",
    "https://twitter.com/yourcompany"
  ]
}

Source: Citable.dev audit data, 2025

FAQPage Schema: Pre-Structured Answers for AI Extraction

FAQPage schema provides AI systems with pre-formatted question-and-answer pairs they can cite directly. BrightEdge’s finding that FAQ blocks drive a 44% increase in AI citations makes sense: FAQs are already in the exact format AI systems need to generate responses. The AI doesn’t have to parse your prose to find the answer — it’s already structured as a question with a direct answer.

The key is writing FAQ answers in answer-first format. Lead with the conclusion, then provide supporting context. This matches how AI systems extract citations — they grab the first sentence or two that directly answers the query. If your answer buries the conclusion after three sentences of context, the AI may cite a competitor who answers directly.

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [{
    "@type": "Question",
    "name": "What is generative engine optimization?",
    "acceptedAnswer": {
      "@type": "Answer",
      "text": "Generative engine optimization (GEO) is the practice of optimizing your website to be cited by AI search platforms like ChatGPT, Gemini, and Perplexity."
    }
  }]
}

Source: BrightEdge, 2025

Article and TechArticle Schema: Content Identity

Article schema tells AI systems that a page is a publishable piece of content with an author, a publication date, and a topic. Without it, AI systems have to infer this from HTML structure — and they frequently get it wrong, attributing content to the wrong author or misjudging publication dates.

For technical documentation, product guides, and how-to content, TechArticle is more specific and more useful than generic Article. It signals technical proficiency and expertise, which matters for E-E-A-T signals that Google AI Overviews weigh when selecting citation sources. Always include datePublished and dateModified — recency shifts AI reranking by up to 95 positions according to SIGIR-AP 2025 research, and visible dates are a prerequisite for that signal.

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Your Article Title",
  "author": {
    "@type": "Person",
    "name": "Author Name"
  },
  "datePublished": "2025-01-15",
  "dateModified": "2025-03-01",
  "publisher": {
    "@type": "Organization",
    "name": "Your Company"
  }
}

Source: Fang et al., SIGIR-AP 2025

BreadcrumbList Schema: Site Structure for AI Navigation

BreadcrumbList schema maps your site’s hierarchy in a way AI systems can traverse. When an AI platform indexes your site, breadcrumb schema helps it understand how your content is organized — which pages are top-level categories, which are subcategories, and how individual pages relate to the broader structure.

This matters because AI systems use site structure as a quality signal. A well-organized site with clear hierarchy suggests authoritative, curated content. A flat structure with no discernible organization suggests a content farm. BreadcrumbList schema makes your hierarchy explicit rather than forcing AI systems to infer it from URL patterns or internal links — which they often do poorly.

{
  "@context": "https://schema.org",
  "@type": "BreadcrumbList",
  "itemListElement": [
    {"@type": "ListItem", "position": 1, "name": "Home", "item": "https://yoursite.com/"},
    {"@type": "ListItem", "position": 2, "name": "Blog", "item": "https://yoursite.com/blog/"},
    {"@type": "ListItem", "position": 3, "name": "Article Title"}
  ]
}

Source: Search Engine Land, 2025

Product and Service Schema: Making Offerings Citable

Product and Service schema tells AI systems exactly what you sell, at what price, and with what features. Without it, AI systems have to scrape your pricing page and hope they interpret the layout correctly. With it, they have machine-readable data they can cite with confidence in recommendation queries.

This matters most for queries like “best [product category] for [use case]” — the exact queries where AI citations drive purchasing decisions. If your competitor has Product schema with clear pricing and feature descriptions and you don’t, the AI system can describe their offering precisely while only vaguely referencing yours. For service businesses, Service schema with serviceType, areaServed, and provider properties serves the same function.

If you’re not sure how your offerings currently appear to AI systems, a Citable Score audit will show you exactly what AI platforms can and can’t extract from your site.

Source: Schema.org documentation; BrightEdge, 2025

LocalBusiness Schema: The Local AI Advantage

Gemini cites brand websites at 52.15% for local queries when Google Business Profile is optimized. LocalBusiness schema is what connects your website entity to your GBP entity — without it, AI systems may treat them as two separate businesses.

Include precise geo coordinates, not just an address. AI systems use coordinates for entity resolution in local queries, matching your website to the correct physical location. Include openingHoursSpecification for each day, telephone, priceRange, and a sameAs link to your Google Business Profile URL. The more complete your LocalBusiness schema, the more confidently AI systems can cite you for “near me” and location-based queries.

{
  "@context": "https://schema.org",
  "@type": "LocalBusiness",
  "name": "Your Business",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "123 Main St",
    "addressLocality": "City",
    "addressRegion": "ST",
    "postalCode": "12345"
  },
  "geo": {
    "@type": "GeoCoordinates",
    "latitude": 40.7128,
    "longitude": -74.0060
  },
  "sameAs": ["https://www.google.com/maps?cid=YOUR_CID"]
}

Source: Yext, October 2025

Speakable Schema: Flagging Content for Voice and AI Extraction

Speakable schema is the most underused and most forward-looking schema type for AI visibility. It explicitly marks sections of your page as suitable for text-to-speech and AI extraction. Google’s documentation describes it as identifying content “best suited for audio playback using text-to-speech” — but the implications extend to any AI system looking for extractable, citable content.

Use cssSelector to point to the specific HTML elements containing your most citable content — typically your opening summary paragraph and key answer sections. This gives AI systems a machine-readable signal for which parts of your page to prioritize when generating responses. Speakable is currently supported for news content in the US, but the underlying signal — “this is the most important extractable content on this page” — is valuable for all AI retrieval systems.

{
  "@context": "https://schema.org",
  "@type": "WebPage",
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": [".article-summary", ".key-takeaway"]
  }
}

Source: Google Developers documentation, 2025

Implementation Priority: Where to Start

Don’t try to implement all seven schema types at once. Start with the highest-impact types based on your business model and build from there. Here’s the priority order based on citation impact data:

Every website: Organization with sameAs (entity foundation), Article on blog posts (content identity), BreadcrumbList (site structure).

Businesses answering questions: Add FAQPage on pages with Q&A content. The 44% citation increase from BrightEdge makes this the highest-ROI addition after Organization.

E-commerce and service businesses: Add Product or Service schema to offering pages. Critical for recommendation queries.

Local businesses: Add LocalBusiness with geo coordinates and GBP sameAs. This connects your web presence to Gemini’s local citation pipeline.

Forward-looking optimization: Add Speakable to your most important content sections. Low effort, growing impact.

To see how your current schema implementation scores across all AI platforms, get your Citable Score report.

Frequently Asked Questions

Does schema markup directly affect AI citations?

Yes, but indirectly. AI chatbots don’t read your JSON-LD in real time during a query. Schema is processed at index time by Google and Bing, which shapes how your content is stored and retrieved. When AI systems query those indexes, schema-enhanced content is more accurately represented and more likely to be selected for citation. GPT-4 accuracy improves from 16% to 54% with structured data present.

Which schema type has the biggest impact on AI citations?

Organization schema with sameAs links has the broadest impact because it establishes your entity identity across all AI platforms. For citation volume specifically, FAQPage schema drives the highest measurable increase — BrightEdge found a 44% boost in AI citations for sites with structured data and FAQ blocks. Start with Organization, then add FAQPage.

Do I need a developer to implement JSON-LD?

No. JSON-LD is added as a <script> tag in your page’s <head> section. Most CMS platforms (WordPress, Shopify, Squarespace) have plugins or built-in options for adding schema without touching code. For WordPress, plugins like Yoast SEO, Rank Math, or Schema Pro can generate JSON-LD automatically. The key is validating the output with Google’s Rich Results Test to ensure accuracy.

Is schema markup enough to get cited by AI search?

No. Schema is one dimension of AI visibility. The Citable Score methodology evaluates six dimensions: content structure, schema markup, authority signals, technical accessibility, entity presence, and platform-specific optimization. Schema makes your content machine-readable, but you still need strong content, clear authority signals, and proper technical setup. Think of schema as the foundation — necessary but not sufficient.

Should I add schema to every page on my site?

No. Prioritize pages that answer questions your target audience asks AI systems. Your homepage needs Organization schema. Blog posts need Article schema. FAQ and service pages need FAQPage and Service schema. But don’t add FAQPage schema to a page with no questions, or Product schema to a page with no products. Inaccurate schema is worse than no schema — it confuses AI systems and can undermine your credibility in both traditional and AI search.