Technique· 9 min de lecture

JSON-LD structured data: the secret weapon of GEO

What is JSON-LD and why it matters for GEO

JSON-LD (JavaScript Object Notation for Linked Data) is a structured-data format recommended by Google and Schema.org. It allows entities (companies, services, articles, people) to be described unambiguously in a format that machines, and LLMs, can interpret directly.

For GEO (Generative Engine Optimization), structured data is decisive: it allows the RAG (Retrieval-Augmented Generation) systems that power ChatGPT, Perplexity and Gemini to understand who you are, what you do and why you are authoritative in your field.

The essential schemas for GEO: complete examples

Each schema below is a ready-to-use JSON-LD block. Adapt the values to your business and integrate them in the <head> of your pages via a <script type="application/ld+json"> tag.

Organization: the company's identity

The Organization schema is the foundation of your digital identity. It must appear on every page of your site.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "MCVA Consulting SA",
  "legalName": "MCVA Consulting SA",
  "description": "Swiss strategy consultancy specialising in generative AI and GEO",
  "url": "https://mcva.ch",
  "logo": "https://mcva.ch/images/logo-mcva.png",
  "foundingDate": "2024",
  "founder": {
    "@type": "Person",
    "name": "Jérôme Deshaie"
  },
  "sameAs": [
    "https://www.linkedin.com/company/mcva-consulting",
    "https://twitter.com/mcva_consulting"
  ],
  "areaServed": ["CH", "FR", "EU"],
  "knowsAbout": ["GEO", "Generative Engine Optimization", "AI Strategy", "SEO", "Structured data"]
}

Key points:

  • knowsAbout indicates your areas of expertise. LLMs use it to assess your authority.
  • areaServed specifies your geographic area of operation.
  • sameAs links your entity to your official profiles on other platforms, which strengthens disambiguation.

LocalBusiness: local presence

If your business has a physical address and serves a local or regional clientele, the LocalBusiness schema complements Organization with geolocated information.

{
  "@context": "https://schema.org",
  "@type": "LocalBusiness",
  "name": "MCVA Consulting SA",
  "image": "https://mcva.ch/images/bureau-mcva.jpg",
  "telephone": "+41 79 612 38 79",
  "email": "contact@mcva.ch",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "Chemin des Crêtes, 7",
    "addressLocality": "Haute-Nendaz",
    "addressRegion": "Valais",
    "postalCode": "1997",
    "addressCountry": "CH"
  },
  "geo": {
    "@type": "GeoCoordinates",
    "latitude": 46.1884,
    "longitude": 7.3066
  },
  "openingHoursSpecification": {
    "@type": "OpeningHoursSpecification",
    "dayOfWeek": ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday"],
    "opens": "08:00",
    "closes": "18:00"
  },
  "priceRange": "$$"
}

Key point: the geo coordinates and structured address allow LLMs to recommend your business in response to local queries such as "GEO consultant in French-speaking Switzerland".

Service: your offers

Each offer should have its own Service schema. This is what allows AI to recommend you when a user looks for a specific service.

{
  "@context": "https://schema.org",
  "@type": "Service",
  "name": "AI Visibility Audit",
  "provider": {
    "@type": "Organization",
    "name": "MCVA Consulting SA"
  },
  "serviceType": "Generative Engine Optimization Audit",
  "areaServed": "Europe",
  "description": "Complete analysis of a company's visibility on 5 major LLMs with competitive benchmark and prioritised action plan.",
  "offers": {
    "@type": "Offer",
    "priceCurrency": "CHF",
    "availability": "https://schema.org/InStock"
  }
}

Key point: be as precise as possible in serviceType. "Consulting" is too vague. "Generative Engine Optimization Audit" allows AI to associate you with targeted queries.

FAQPage: frequently asked questions

The FAQPage schema is doubly useful: it generates rich results in Google and provides question/answer pairs directly usable by LLMs.

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is GEO?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "GEO (Generative Engine Optimization) is the set of practices aimed at optimising a brand's visibility in responses generated by LLMs such as ChatGPT, Perplexity and Gemini."
      }
    },
    {
      "@type": "Question",
      "name": "Why is structured data important for GEO?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "JSON-LD structured data lets LLMs unambiguously understand the identity, services and expertise of a company, increasing the likelihood of being cited in AI responses."
      }
    }
  ]
}

Key point: each question/answer pair must be standalone and factual. LLMs extract these answers as-is to feed their syntheses. It is one of the most effective levers to be cited by ChatGPT.

Article: your contents

The Article schema reinforces the authority of your publications and allows LLMs to date and attribute them correctly.

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "JSON-LD Structured Data: The Secret Weapon of GEO",
  "author": {
    "@type": "Person",
    "name": "Jérôme Deshaie",
    "url": "https://mcva.ch/about",
    "jobTitle": "CEO & Founder"
  },
  "publisher": {
    "@type": "Organization",
    "name": "MCVA Consulting SA",
    "logo": {
      "@type": "ImageObject",
      "url": "https://mcva.ch/images/logo-mcva.png"
    }
  },
  "datePublished": "2026-03-12",
  "dateModified": "2026-03-12",
  "image": "https://mcva.ch/images/donnees-structurees-hero.jpg",
  "keywords": ["JSON-LD", "structured data", "GEO", "Schema.org"],
  "inLanguage": "en"
}

Key point: keep dateModified up to date at every revision. Web-enabled LLMs favour recent content. The inLanguage field helps multilingual systems contextualise your content.

Person: individual expertise

The Person schema serves to establish the author's authority. In an E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) context, it allows engines and LLMs to associate content with a qualified person.

{
  "@context": "https://schema.org",
  "@type": "Person",
  "name": "Jérôme Deshaie",
  "jobTitle": "CEO & Founder",
  "worksFor": {
    "@type": "Organization",
    "name": "MCVA Consulting SA"
  },
  "url": "https://mcva.ch/about",
  "sameAs": [
    "https://www.linkedin.com/in/jeromedeshaie"
  ],
  "knowsAbout": ["GEO", "SEO", "generative AI", "digital strategy", "structured data"],
  "alumniOf": {
    "@type": "EducationalOrganization",
    "name": "Reference University"
  }
}

Key point: knowsAbout on a Person schema reinforces the signal of individual expertise. When an LLM looks for an expert on a subject, this field is decisive.

Breadcrumbs help LLMs understand your site structure and the relationship between your pages.

{
  "@context": "https://schema.org",
  "@type": "BreadcrumbList",
  "itemListElement": [
    { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://mcva.ch" },
    { "@type": "ListItem", "position": 2, "name": "Insights", "item": "https://mcva.ch/insights" },
    { "@type": "ListItem", "position": 3, "name": "JSON-LD Structured Data" }
  ]
}

Impact of structured data on LLM citability

Structured data does not just serve classic SEO. Its impact on citability by LLMs is a strategic lever that is still under-exploited.

The RAG (Retrieval-Augmented Generation) systems that power ChatGPT, Perplexity or Gemini work in two stages: first they retrieve relevant documents, then they generate a synthetic answer. During the retrieval phase, structured data provides directly usable context:

  • Entity disambiguation: JSON-LD makes it unambiguous that "MCVA Consulting SA" is a Swiss company specialising in GEO, not a generic acronym. Without this precision, the LLM may confuse your brand with another entity.
  • Reliable attribution: when an LLM cites your content, the Article and Person schemas allow it to correctly attribute the source, author and date. This increases the LLM's confidence in your content.
  • Service-query response: a user who asks an LLM "which firm can audit my AI visibility in Switzerland?" will receive a response citing your business if your Service and LocalBusiness schemas are correctly set up.
  • Content freshness: dateModified signals to LLMs that your content is kept up to date, a major trust criterion in a fast-moving field.

Structured data turns your site into a knowledge base usable by LLMs, well beyond simply displaying rich snippets in Google. It is a pillar of your AI visibility.

Common mistakes in structured data

Mistake 1: inconsistent data across pages

If your Organization schema says "MCVA Consulting SA" on one page and "MCVA Consulting" on another, LLMs may treat this as two distinct entities. Use exactly the same name everywhere.

Mistake 2: outdated information

Structured data with an old address or phone number is worse than no data at all. LLMs will reproduce this incorrect information in their responses.

Mistake 3: overly minimal schemas

An Organization schema that only contains the name and URL adds almost no GEO value. The richer and more precise your schemas, the more information LLMs have to cite you correctly.

Your Service schemas should reference your Organization schema via provider. Your Article schemas should reference the author and the publisher. This semantic web reinforces the consistency of your digital identity.

Mistake 5: using the wrong @type

Confusing Organization and Corporation, or using WebPage instead of Article, sends an erroneous signal to search engines and LLMs. Always check the Schema.org documentation to choose the most specific applicable type.

Mistake 6: missing required properties

Google requires certain properties to enable rich results. For example, an Article schema without headline, image or datePublished will not generate any rich result. Consult the Google Search Central documentation for the properties required by schema type.

Mistake 7: untested structured data

Publishing structured data without validating it is risky. A JSON syntax error, a malformed URL or a missing field can render the whole block invisible to Google and LLMs. Test systematically before publishing.

How to verify your structured data

Four essential tools:

  1. Google Rich Results Test: Google's official tool to check whether your structured data is eligible for rich results. It detects syntax errors, missing properties and compliance issues. Test each page individually.
  2. Schema.org Validator: validates conformity to the Schema.org standard, independent of Google's specific requirements.
  3. Google Search Console: the "Enhancements" section displays structured-data errors detected during the crawl of your site. It is the best long-term monitoring tool.
  4. Manual test on LLMs: ask questions to ChatGPT and Perplexity to verify that your data is taken into account. Ask, for example, "Who is MCVA Consulting?" or "Which firm offers an AI visibility audit in Switzerland?".

Structured data and GEO: the synergy

LLMs do not "see" your site like a human: they do not browse your pages visually. They read the source code, and JSON-LD is the most readable and structured part of that code.

Concretely, here is why structured data is a multiplier for GEO performance:

LLMs natively read JSON-LD. When a RAG system indexes your page, it extracts visible text but also <script type="application/ld+json"> blocks. These blocks provide structured metadata that the LLM can use directly, without having to infer them from free text.

The Knowledge Graph feeds on Schema.org. Google's Knowledge Graph, which influences AI Overviews, relies heavily on structured data to build its knowledge graph. An entity well-described in JSON-LD has more chance of joining this graph, and therefore of being cited by Gemini and Google AI Overviews.

The content + structure synergy. A substantive article without structured data is good content. A substantive article with Article, Person, Organization and FAQPage schemas is citable and attributable content. This approach also applies to broader AI projects, as shown in our guide on the architecture of a high-performing AI project. LLMs favour sources they can identify and attribute with confidence.

The Swiss trust signal. For companies based in Switzerland, the LocalBusiness schema with addressCountry: "CH" and a complete address reinforces a specific trust signal. LLMs, in response to geolocated queries, favour entities whose location is clearly established.

JSON-LD and GEO: the measurable impact

At MCVA Consulting, we have observed that adding complete structured data (Organization + Service + Article + BreadcrumbList) improves the average GEO Score by 15 to 25 points in 3 months.

Structured data is not optional for GEO. It is the technical bedrock on which any AI-visibility strategy rests.

Summary

  • JSON-LD structured data lets LLMs understand your business unambiguously.
  • The essential schemas: Organization, LocalBusiness, Service, FAQPage, Article, Person and BreadcrumbList.
  • knowsAbout and serviceType are the most structuring fields for GEO.
  • Adding complete structured data improves the GEO Score by 15 to 25 points on average.
  • Maintain perfect consistency across all your schemas (name, address, description).
  • Systematically test with the Google Rich Results Test before publishing.

FAQ

Is structured data enough for GEO?

No. Structured data is a necessary technical pillar, but it is not sufficient on its own. A complete GEO strategy combines structured data, quality content optimised for citability, domain authority, presence on reliable third-party sources and semantic linking. Structured data provides the framework; content provides the substance. Without expert and up-to-date content, even the best JSON-LD schemas will not generate citations from LLMs.

What impact on Google ranking?

Structured data is not a direct ranking factor in Google's algorithm. However, it allows you to obtain rich results (FAQ, reviews, breadcrumbs), which significantly increase click-through rate (CTR). A better CTR sends a positive engagement signal to Google. Furthermore, in the context of AI Overviews, structured data helps Google understand and cite your content more precisely. The impact is therefore indirect but measurable.

Do you need a developer to implement structured data?

Not necessarily. For a WordPress site, plugins such as Yoast SEO or Rank Math automatically generate basic schemas. For a Next.js site or a JavaScript framework, JSON-LD integration happens directly in the component code. If you need advanced and custom schemas (recommended for a GEO strategy), technical support is preferable to ensure data consistency and completeness. In Switzerland, most web agencies can integrate JSON-LD in a few hours of development.

Want to know whether your structured data is correctly set up for GEO? MCVA Consulting can audit your JSON-LD schemas and optimise your visibility in AI responses. Discover our GEO audit for a complete diagnosis.

Related articles

Guide

AI visibility for businesses: getting recommended by ChatGPT, Claude and Perplexity

The reference article on AI visibility. How to be recommended by LLMs, the GEO Score™ explained, and the 7 levers of citability.

10 min

How-to

How to be cited by ChatGPT: 7 actionable strategies

ChatGPT recommends some companies and ignores others. Here are 7 concrete strategies to get your brand to appear in its responses.

9 min

Technique

AI architecture: the key to a scalable and high-performing project

How to structure an AI project so it scales. APIs, fine-tuning, RAG, agents — an architecture guide for Swiss SMEs.

8 min