Schema.org for AI: The 5 Priority JSON-LD Schemas

Schema.org structured data is no longer just about earning rich snippets on Google. In 2026, it also helps generative engines read your site: when ChatGPT, Gemini or Perplexity analyze a page, JSON-LD markup tells them without ambiguity who is publishing, what type of content it is, and when it was published.

Let's be clear about what to expect: structured data is not decisive for being cited. Google states that no special markup is needed to appear in its AI Overviews and AI Mode (Google Search Central), and our study of 317 pages shows that what sets cited pages apart lies mostly in the visible content: short answers under headings, question headings, tables, a recent update date. Markup is a useful, low-cost complement, as long as it faithfully describes what is displayed.

This guide shows you exactly which schemas to implement, in what order, with ready-to-copy code examples.

Why Schema.org matters for GEO

AI engines work differently from Google. When Googlebot crawls your page, it interprets HTML, meta tags, and popularity signals to rank your page. When an AI search crawler (OAI-SearchBot, Claude-SearchBot, PerplexityBot) analyzes your page, it looks to understand entities: who is speaking, about what, with what authority, and when.

Structured data answers part of these questions (who is publishing, about what, on what date) in a format machines read without interpretation.

Specifically, Schema.org feeds 2 of the 6 Detekia GEO score criteria:

• Freshness (15 points): the dateModified of your Article schema is one of the dates Detekia reads to measure how recent the page is
• Entity clarity (5 points): an Organization named in your structured data tells engines who is speaking

Schemas also help Detekia recognize the page type (an Article or Product schema, for instance), which sets the content depth thresholds. They do not replace clear answers: in our September 30, 2026 study of 317 pages, the Organization schema alone, like sameAs links, did not set cited pages apart. The FAQPage schema remains useful because it goes with a question-and-answer format that AI engines extract easily.

The 5 priority schemas for GEO

You do not need to implement all 806 Schema.org types. For GEO, 5 schemas cover most needs. Here they are, in priority order. To understand how each criterion weighs in the GEO score, see the 6 Detekia score criteria explained.

1. Organization — your identity

This is the most fundamental schema. It tells engines: "This is who we are." It helps identify your business without ambiguity, even though our study did not find it setting cited pages apart from uncited pages on the same site.

Where to implement: on your homepage, in the <head> or before the closing </body>.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Company",
  "url": "https://www.your-website.com",
  "logo": "https://www.your-website.com/logo.png",
  "description": "Concise description of your business in 1-2 sentences.",
  "foundingDate": "2020",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "123 Main Street",
    "addressLocality": "New York",
    "addressRegion": "NY",
    "postalCode": "10001",
    "addressCountry": "US"
  },
  "contactPoint": {
    "@type": "ContactPoint",
    "email": "contact@your-website.com",
    "contactType": "customer service"
  },
  "sameAs": [
    "https://www.linkedin.com/company/your-company",
    "https://twitter.com/your-company"
  ]
}

Key points for GEO:

• The sameAs property lets engines connect your site to your LinkedIn, Twitter, Wikipedia, and Crunchbase profiles. Consistent links make your identity clearer; do not expect a direct effect on whether your pages are cited, though.
• The description should be factual and concise — it is often what AI will extract first to introduce you.
• The logo as an absolute URL removes any ambiguity about which file to use.

2. FAQPage — your questions and answers

This is the schema closest to how AI engines work, on a question-and-answer basis: a user asks a question, the AI looks for the best answer. But the visible FAQ is what matters first: headings phrased as questions followed by a short answer are among the traits that set cited pages apart in our study. The FAQPage schema repeats that FAQ, word for word, in a machine-readable format.

Important note: Google dropped FAQ rich results for all sites on May 7, 2026. The FAQPage schema remains a valid Schema.org type, useful for Bing and for AI engines, but it no longer produces any rich display in Google. Its value is therefore no longer display in Google: it is a reading aid for other engines.

Where to implement: on any page containing Q&A — dedicated FAQ page, bottom of product pages, help sections.

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "How can I improve my site's visibility in ChatGPT?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "To be cited by ChatGPT, put a short answer under each heading, phrase your headings as questions, verify your robots.txt allows OAI-SearchBot, and source your claims with verifiable data."
      }
    },
    {
      "@type": "Question",
      "name": "Which crawlers should you allow to appear in ChatGPT search?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "OAI-SearchBot, OpenAI's search crawler. GPTBot is used for model training: you can block it without disappearing from ChatGPT search."
      }
    }
  ]
}

Key points for GEO:

• Each answer must be self-contained: understandable without additional context. AI engines often extract the answer alone, without the question.
• Questions should be phrased in natural language, the way a user would ask ChatGPT. No jargon, no corporate phrasing.
• Limit yourself to 5–10 questions per page. Beyond that, information density drops and Google may ignore the markup.
• Each answer should contain verifiable facts (numbers, timeframes, steps) — not vague responses.

3. Article / BlogPosting — your editorial content

This schema identifies your articles as structured editorial content, with an author, a date, and editorial context. Its most useful property for GEO is the update date: a recent date was one of the traits of cited pages in our study.

Where to implement: on every blog post, guide, case study, or editorial content page.

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "GEO: the complete guide to getting cited by AI in 2026",
  "description": "GEO definition, 7 citation-worthiness criteria, 7 concrete actions to optimize your site.",
  "author": {
    "@type": "Person",
    "name": "John Smith",
    "url": "https://www.linkedin.com/in/johnsmith",
    "jobTitle": "Founder",
    "worksFor": {
      "@type": "Organization",
      "name": "Your Company"
    }
  },
  "publisher": {
    "@type": "Organization",
    "name": "Detekia",
    "logo": {
      "@type": "ImageObject",
      "url": "https://www.detekia.fr/logo.png"
    }
  },
  "datePublished": "2026-03-27",
  "dateModified": "2026-03-27",
  "image": "https://www.detekia.fr/images/geo-guide-og.jpg",
  "mainEntityOfPage": "https://www.detekia.fr/blog/geo-guide-complet-2026"
}

Key points for GEO:

• The author object with a url pointing to LinkedIn or a bio page is useful to readers, and expected in regulated sectors (health, finance, law). Our study did not, however, find an identified author setting cited pages apart, and Google reminds us that E-E-A-T is not itself a specific ranking factor.
• dateModified is even more useful than datePublished. A recent modification date, consistent with the date shown on the page, is a positive signal.
• The description should summarize the content in a factual, extractable way.

4. WebSite + SearchAction — your site as an entity

This schema helps AI engines understand your site as a whole, not just page by page, and declares your site's official name. Google, however, removed the sitelinks search box from its results in November 2024: the SearchAction property is no longer used there, but keeping it causes no problem.

Where to implement: on the homepage only.

{
  "@context": "https://schema.org",
  "@type": "WebSite",
  "name": "Detekia",
  "url": "https://www.detekia.fr",
  "description": "Free GEO score to measure how likely your site is to be cited by AI.",
  "potentialAction": {
    "@type": "SearchAction",
    "target": "https://www.detekia.fr/results?url={search_term_string}",
    "query-input": "required name=search_term_string"
  }
}

→For more on FAQPage, see FAQ and FAQPage schema: answers AI engines cite.

5. BreadcrumbList — your architecture

Breadcrumb markup helps AI engines understand your site's hierarchy: which page is the parent of which, how content is organized. It is a structural signal, useful mainly for how engines read the site.

Where to implement: on all internal pages.

{
  "@context": "https://schema.org",
  "@type": "BreadcrumbList",
  "itemListElement": [
    {
      "@type": "ListItem",
      "position": 1,
      "name": "Home",
      "item": "https://www.detekia.fr"
    },
    {
      "@type": "ListItem",
      "position": 2,
      "name": "Blog",
      "item": "https://www.detekia.fr/blog"
    },
    {
      "@type": "ListItem",
      "position": 3,
      "name": "GEO: the complete 2026 guide",
      "item": "https://www.detekia.fr/blog/geo-guide-complet-2026"
    }
  ]
}

Secondary schemas by activity type

Beyond the 5 priority schemas, certain specific markups strengthen your visibility depending on your business type.

E-commerce

• Product — name, price, availability, reviews. AI engines use these for product comparisons.
• AggregateRating — average rating and number of reviews. Strong trust signal.
• Offer — price, currency, conditions. Lets AI engines cite precise pricing. Incomplete Product markup is one of the common GEO mistakes in e-commerce.

→For a complete approach, read E-commerce: how to appear in AI product recommendations.

Services and professional firms

• LocalBusiness — address, hours, geographic area. Essential for local queries ("best [profession] in [city]").
• Service — description of your offerings with pricing and coverage area.

SaaS and tech

• SoftwareApplication — name, operating system, category, price. AI engines cite this info in software comparisons.
• HowTo — steps for using your tool. A format that is highly extractable by AI engines.

→Before implementing schemas, make sure your site is crawlable by AI bots: llms.txt, robots.txt and AI accessibility.

Technical implementation

The format: JSON-LD only

There are three formats for structured data: JSON-LD, Microdata, and RDFa. In 2026, the choice is clear: use JSON-LD in 100% of cases.

Why:

• Google officially recommends it
• It is added as a <script> block without touching the visible HTML
• It is easier to maintain and debug
• AI crawlers parse it natively

Where to place the code

The JSON-LD block goes in the <head> of your page or just before </body>. Both work. The most common convention is in the <head>.

<head>
  <!-- ... your meta tags ... -->
  <script type="application/ld+json">
  {
    "@context": "https://schema.org",
    "@type": "Organization",
    "name": "Your Company"
  }
  </script>
</head>

Multiple schemas on a single page

You can (and often should) have multiple JSON-LD blocks on a single page. For example, your homepage can contain an Organization schema, a WebSite schema, and an FAQPage schema. Each schema gets its own <script type="application/ld+json"> block.

Next.js implementation

If your site is built with Next.js, use the Head component:

import Head from 'next/head';

export default function HomePage() {
  const schema = {
    "@context": "https://schema.org",
    "@type": "Organization",
    "name": "Your Company",
    "url": "https://www.your-website.com"
  };

  return (
    <>
      <Head>
        <script
          type="application/ld+json"
          dangerouslySetInnerHTML={{ __html: JSON.stringify(schema) }}
        />
      </Head>
      {/* ... your content ... */}
    </>
  );
}

WordPress implementation

Two options:

• Plugin — Yoast SEO and Rank Math automatically add certain schemas (Organization, Article, BreadcrumbList). Check the settings and complete manually if needed (FAQPage is not always automatic). For more, see WordPress and GEO: plugins and optimizations.
• Manual — paste the JSON-LD block in your page's HTML editor, or use a plugin like "Insert Headers and Footers" to add it globally.

Testing and validation

Before publishing, always test your markup.

Testing tools

• Google Rich Results Test (search.google.com/test/rich-results) — verifies whether your markup is valid and eligible for Google rich results (Organization, Article, Product, BreadcrumbList…). It no longer supports the FAQ type since FAQ rich results were dropped: validate FAQPage with the Schema Markup Validator.
• Schema Markup Validator (validator.schema.org) — validates your markup syntax against the complete Schema.org vocabulary. Stricter than the Google test.
• Your Detekia score — run the free score before and after implementation to measure the impact on your score (structured data feeds the Freshness and Entity clarity criteria in particular). To then check whether ChatGPT, Gemini, Claude or Perplexity actually cite your pages, an AI visibility audit of your brand measures your citations on your key queries.

Common errors to avoid

• Invisible markup — the content described in the schema must be visible on the page. A FAQPage schema whose questions are not visually displayed violates Google guidelines and may be ignored.
• Inconsistent information — the company name in the Organization schema must match exactly what is displayed on your site, your Google Business Profile, and your social media. Any inconsistency blurs the identification of your brand.
• Missing or false dates — datePublished and dateModified must reflect reality. Do not set today's date as dateModified if you have not actually modified the content.
• Minimal schema — an Organization schema with only the name is nearly useless. Fill in all relevant properties: logo, address, contact, sameAs.
• Forgetting to test after deployment — a template change, plugin update, or migration can silently break your markup.

Implementation checklist

Frequently asked questions

Is structured data enough to be cited by AI?

No. Google states that no special markup is needed for its AI features, and our study did not find Organization markup setting cited pages apart. In the Detekia score, structured data contributes to the Freshness and Entity clarity criteria. But without clear answers under headings, without evidence, without access for AI search crawlers, and without external presence, markup alone is not enough. It is a complement, not a decisive lever.

How long before seeing the impact?

Effects on Google rich results (except FAQ rich results, which are no longer displayed) appear within a few days to 2 weeks. On AI citations, no direct effect of markup has been demonstrated: expect mainly a more reliable reading of your dates and identity.

Can I use an automatic schema generator?

Yes, to get started. Tools like Google's Structured Data Markup Helper or the built-in features of Yoast/Rank Math work well for basic schemas. But for careful markup (a reliable dateModified, a FAQPage identical to the visible FAQ, complete sameAs), manual customization is often necessary.

Can too many schemas hurt?

No, as long as each schema is valid and describes content that is actually present on the page. Google and AI engines have no limit on the number of schemas per page. The only rule: only mark up what is visible and truthful.

Go further

Structured data is only one building block of technical optimization for AI. The other guides in this series:

• The advanced schemas almost nobody implements (HowTo, Person, DefinedTerm, ItemList, Service)
• FAQ and FAQPage schema: answers AI engines cite
• llms.txt, robots.txt and AI accessibility, the guide to AI bot access
• Sitemap and robots.txt for AI bots
• Why your site blocks AI bots without you knowing (firewalls, Cloudflare, JavaScript rendering)
• Internal linking and site structure for AI
• Meta descriptions for SEO and GEO
• Website redesign: preserving your AI visibility
• WordPress and GEO: plugins and optimizations

Measure the impact

Have you implemented your schemas? Measure the impact right away.

Test your page with Detekia's free GEO score: it checks that your dates and your organization are read correctly (Freshness and Entity clarity criteria). It's free.