Why Structured Data Matters for AI
HTML is readable for humans, but interpretable for machines. An <h2> element signals "this is a second-order heading" — but it doesn't say what that heading means. Schema.org closes this gap.
Schema.org is a vocabulary for semantic metadata. Embedded as JSON-LD in HTML pages, it explains the meaning of content to machines: "this text is an article," "this person is the author," "this organization is the publisher." AI models that process web pages use these signals to assess relevance, authority, and context.
Anyone who doesn't use Schema.org relies on an AI model to derive the context from the running text. That works — but unreliably. Schema.org makes meaning explicit and thereby machine-readably reliable.
Article — Mandatory for Every Blog Post
The Article schema is the most important type for content pages. It defines an article as such and connects it with author, publisher, and metadata. For GEO, the following fields are critical:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "Title of the article (max. 110 characters)",
"description": "Meta description (max. 155 characters)",
"datePublished": "2026-05-16T08:00:00+02:00",
"dateModified": "2026-05-16T08:00:00+02:00",
"author": {
"@type": "Person",
"name": "Thomas Leonhardt",
"url": "https://forgeproject.eu/about/"
},
"publisher": {
"@type": "Organization",
"name": "FORGE",
"url": "https://forgeproject.eu"
},
"wordCount": 1700,
"articleSection": "GEO Technology",
"keywords": ["Schema.org", "JSON-LD", "GEO"],
"inLanguage": "en-US",
"mainEntityOfPage": {
"@type": "WebPage",
"@id": "https://forgeproject.eu/blog/slug/"
}
}
</script>Why wordCount matters: AI models use wordCount as an indicator of depth. Articles with more than 1200 words signal: this was covered thoroughly, not just superficially. That increases the likelihood of a citation.
dateModified and datePublished: Both fields together signal currency. Only datePublished without dateModified looks like an outdated article that was never updated.
Person — Defining Authorship Globally
The Person schema belongs on the About page and defines the author as a machine-readable entity. It is the anchor for all Article schemas on the domain:
{
"@context": "https://schema.org",
"@type": "Person",
"name": "Thomas Leonhardt",
"url": "https://forgeproject.eu/about/",
"jobTitle": "Founder & AI Orchestrator",
"knowsAbout": [
"Generative Engine Optimization",
"AI agent orchestration",
"GEO strategy"
],
"sameAs": [
"https://www.linkedin.com/in/thomas-leonhardt-forge/"
],
"worksFor": {
"@type": "Organization",
"name": "FORGE",
"url": "https://forgeproject.eu"
}
}Organization — Your Company as an Entity
The Organization schema belongs on the homepage. It defines FORGE as an independent machine-readable entity — important so that AI models can distinguish between "Thomas Leonhardt" and "FORGE" and recognize both as belonging together:
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "FORGE",
"url": "https://forgeproject.eu",
"logo": "https://forgeproject.eu/og-image.png",
"founder": {
"@type": "Person",
"name": "Thomas Leonhardt"
},
"foundingDate": "2020",
"areaServed": "DE",
"description": "FORGE builds autonomous AI agent systems for companies. Founded in 2020, 40+ completed projects, specializing in GEO and multi-agent orchestration.",
"contactPoint": {
"@type": "ContactPoint",
"contactType": "Business Inquiries",
"email": "kontakt@forgeproject.eu"
}
}FAQPage — Enabling Direct AI Answers
FAQPage is the most powerful schema type for GEO. It gives AI models exactly the format they need for direct answers: question + answer in structured form. Pages with FAQPage schema are cited significantly more often in Perplexity and other AI search engines:
Weiterlesen — kostenlos
Den vollständigen Inhalt freischalten
Trag deine E-Mail-Adresse ein und bestätige sie: Du abonnierst den Signal-Forge-Newsletter von FORGE und erhältst sofort Zugang zu diesem und allen weiteren registrierungspflichtigen Inhalten. Die Abmeldung ist jederzeit möglich.
Schon registriert? Der Link aus deiner Bestätigungs-Mail schaltet dieses Gerät wieder frei.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "What is Generative Engine Optimization (GEO)?",
"acceptedAnswer": {
"@type": "Answer",
"text": "GEO (Generative Engine Optimization) is the optimization of website content for AI search engines like Claude, ChatGPT, and Perplexity. Unlike SEO, which targets search-result rankings, GEO optimizes for being cited as a trustworthy source in AI-generated answers."
}
},
{
"@type": "Question",
"name": "How does GEO differ from SEO?",
"acceptedAnswer": {
"@type": "Answer",
"text": "SEO optimizes for algorithms that rank pages by backlinks and keywords. GEO optimizes for language models that generate answers — with a focus on authorship, semantic clarity, and citation-worthiness instead of link building."
}
}
]
}When to use FAQPage: On pages that address frequently asked questions, on service pages, on the About page. Not on every blog post — that looks contrived. On pillar-content pages with 3–8 genuine questions, however, FAQPage is very effective.
HowTo — Making Instructions Recognizable to AI
When you write step-by-step instructions, HowTo schema signals: "This is a process with clear steps." AI models preferentially use this format for direct answers to "How do I do X?" queries:
{
"@context": "https://schema.org",
"@type": "HowTo",
"name": "How to implement Schema.org on your website",
"step": [
{
"@type": "HowToStep",
"position": 1,
"name": "Create a JSON-LD snippet",
"text": "Create a script tag with type='application/ld+json' in the head section of your HTML page."
},
{
"@type": "HowToStep",
"position": 2,
"name": "Choose the schema type",
"text": "Choose the appropriate schema type: Article for blog posts, Person for authors, Organization for your company."
},
{
"@type": "HowToStep",
"position": 3,
"name": "Validate",
"text": "Test the markup with the Google Rich Results Test at search.google.com/test/rich-results."
}
]
}BreadcrumbList — Context for Crawling
BreadcrumbList helps AI models understand the structure of your website: where does this page sit in the hierarchy? Is it part of a topic cluster? For GEO, a correct breadcrumb signals: "This article belongs to a structured content area, not to an isolated page":
{
"@context": "https://schema.org",
"@type": "BreadcrumbList",
"itemListElement": [
{
"@type": "ListItem",
"position": 1,
"name": "FORGE",
"item": "https://forgeproject.eu/"
},
{
"@type": "ListItem",
"position": 2,
"name": "Blog",
"item": "https://forgeproject.eu/blog/"
},
{
"@type": "ListItem",
"position": 3,
"name": "Schema.org Markup for AI Crawling",
"item": "https://forgeproject.eu/blog/schema-org-markup-fuer-ki-crawling/"
}
]
}Schema Types at a Glance: What AI Models See
| Schema type | What AI models recognize | GEO benefit |
|---|---|---|
Article |
Author, date, topic area, word count | Communicate authorship + currency |
Person |
Name, expertise, external profiles, employer | Define authority in a machine-readable way |
Organization |
Company name, founder, area of service | Anchor the brand as a trustworthy entity |
FAQPage |
Question-answer pairs directly structured | Direct citation format for AI answers |
HowTo |
Process steps as a structured list | Mark instructions as directly citable |
BreadcrumbList |
Page hierarchy + topic cluster | Signal content structure and context |
Validation — Three Tools
Before Schema.org markup goes live, it should be validated. Faulty JSON-LD snippets are worse than no schema at all — they generate inconsistent signals:
- Google Rich Results Test:
search.google.com/test/rich-results— shows whether Google recognizes the schema correctly and can use it for rich results - Schema.org Validator:
validator.schema.org— checks pure schema conformance independently of Google - Browser console: test JSON-LD snippets in the browser console with
JSON.parse()— syntactic errors are immediately visible
Schema.org is not a one-time setup — it is maintenance. New articles need Article schema, new features need new schema types. Whoever establishes it as part of the publishing process has the greatest GEO advantage in the long run.