Advanced Structured Data: The Schemas Most Sites Forget

You've implemented Organization and FAQPage on your site. You think your structured data is in order. Yet there are a dozen schemas and fields you may not even know about.

The Princeton study (Aggarwal et al., KDD 2024) demonstrated that making content more verifiable (citing sources, adding statistics and quotations) can boost its visibility in generative answers by up to 40%. Structured data extends that logic: it makes the author, date, reviews or steps explicit. But the key isn't having "some schema" — it's having the right schemas with the right fields, accurately describing what the page shows. Still, many sites stop at Organization and FAQPage, ignoring a dozen useful types.

What advanced schemas can (and cannot) do for AI

RAG systems (Retrieval-Augmented Generation) used by ChatGPT, Perplexity and Gemini don't read your pages like a human. They look for information they can extract, compare and cross-reference. A well-populated JSON-LD schema makes information already on the page (last update, steps, reviews) explicit in a standardized format.

But markup is not what gets a page cited. For its AI Overviews and AI Mode, Google states that no special optimization is required: no special markup or schema.org. And in the Detekia calibration study (317 pages from 40 sites, cited pages compared with non-cited pages from the same site), neither Organization schema nor a named author distinguished cited pages. What did: a short answer right under the headings, question headings, tables, a recent update date, more complete content. The schemas below therefore describe good content properly; they do not replace it. AI citability is decided first in the visible text.

The schemas almost nobody implements

1. HowTo — step-by-step guides

If your site contains tutorials, practical guides or multi-step processes, the HowTo schema lets AI engines extract them as structured lists. It's the ideal format for direct answers.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "HowTo",
  "name": "How to [your process]",
  "step": [
    { "@type": "HowToStep", "name": "Step 1", "text": "[description]" },
    { "@type": "HowToStep", "name": "Step 2", "text": "[description]" },
    { "@type": "HowToStep", "name": "Step 3", "text": "[description]" }
  ]
}
</script>

Why use it: numbered steps, visible on the page, make good answers to "how to..." queries. A HowTo describes them explicitly; the visible list is what matters first.

2. Review and AggregateRating — structured social proof

Customer reviews are part of the external signals in the Detekia score (reviews or testimonials on the page). Review markup makes each review's rating, author and date explicit.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "[your product]",
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "[average rating]",
    "reviewCount": "[number of reviews]",
    "bestRating": "5"
  },
  "review": [{
    "@type": "Review",
    "author": { "@type": "Person", "name": "[customer name]" },
    "datePublished": "[date]",
    "reviewBody": "[review content]",
    "reviewRating": { "@type": "Rating", "ratingValue": "[rating]" }
  }]
}
</script>

AI engines cross-reference ratings across sources (your site, Trustpilot, Google Reviews). An AggregateRating on your site strengthens the consistency of the signal.

3. Person — author expertise

A Person schema describes the author: qualifications, social links, affiliation. It is useful to readers and expected in regulated sectors (health, law, finance). But don't expect a direct effect on citations: according to Google, E-E-A-T is not in itself a ranking factor, and in the Detekia calibration study, a named author did not distinguish cited pages from other pages on the same site.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Person",
  "name": "[author name]",
  "jobTitle": "[job title]",
  "worksFor": { "@type": "Organization", "name": "[company]" },
  "sameAs": [
    "[LinkedIn URL]",
    "[Twitter URL]"
  ],
  "knowsAbout": ["[area of expertise 1]", "[area of expertise 2]"]
}
</script>

The knowsAbout field is rarely filled in. It specifies the topics the author covers; no effect on citations has been shown.

4. SpeakableSpecification — content readable by voice assistants

This schema indicates which passages of your page are best suited for reading aloud: you yourself designate the sentences that make the best answers.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "WebPage",
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": [".answer-capsule", ".key-takeaway"]
  }
}
</script>

Few sites use it. Google still presents it as a beta feature, and its effect on AI answers has not been shown.

5. DefinedTerm and DefinedTermSet — your domain vocabulary

If your site uses technical or specialized vocabulary, this schema signals that you define terms in your field. It suits glossaries on B2B, legal, medical or financial sites.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "DefinedTermSet",
  "name": "[your field] glossary",
  "hasDefinedTerm": [{
    "@type": "DefinedTerm",
    "name": "[technical term]",
    "description": "[clear definition]"
  }]
}
</script>

When a user asks ChatGPT "what is [technical term]", what can get you cited is first a short, clear definition, visible right under the heading. The DefinedTerm describes it; it does not replace it.

6. ItemList — rankings and comparisons

"Best...", "top 10...", "comparison..." queries are among the most frequent on AI engines. In the Detekia calibration study, "best..." comparisons were among the most cited pages: what counts is explicit criteria, a table and visible facts. The ItemList schema then describes the order of your ranking.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "ItemList",
  "name": "[title of your ranking]",
  "itemListElement": [
    { "@type": "ListItem", "position": 1, "name": "[item 1]", "url": "[link]" },
    { "@type": "ListItem", "position": 2, "name": "[item 2]", "url": "[link]" }
  ]
}
</script>

7. Service and Offer — your offerings with pricing and terms

If you sell services, the Service schema with an embedded Offer (price, availability, geographic area) lets AI engines compare your offering with your competitors' in a structured way.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Service",
  "name": "[service name]",
  "provider": { "@type": "Organization", "name": "[your company]" },
  "areaServed": "[geographic area]",
  "offers": {
    "@type": "Offer",
    "price": "[price]",
    "priceCurrency": "EUR",
    "availability": "https://schema.org/InStock"
  }
}
</script>

Queries such as "how much does [service] cost" or "which [service] in [city]" rely on this information: show it plainly on the page as well, since the schema only describes it. For deeper implementation guidance, see our complete Schema.org guide.

Forgotten fields in common schemas

Even if you already use Organization or Article, certain fields are almost always empty:

Organization: the missing fields

• foundingDate — the company's founding date (a longevity signal)
• numberOfEmployees — team size (a credibility signal)
• award — awards and distinctions received
• sameAs — links to ALL your profiles (LinkedIn, Twitter, Crunchbase, Wikipedia)
• knowsAbout — the company's areas of expertise

Article: the missing fields

• dateModified — the date of the last update (a freshness signal)
• wordCount — the number of words (a depth signal)
• citation — the sources cited in the article
• about — the article's main topic (helps AI engines categorize it)
• author with a complete Person schema (not just a name as a string)

How to prioritize implementation

On WordPress, some of these schemas can be handled by a plugin and the rest with manual JSON-LD: see WordPress and GEO: plugins and optimizations.

Don't try to implement all these schemas at once. Here is a sensible order, from most useful to most specialized:

1. Enrich your existing schemas — add the missing fields to Organization and Article. It's the fastest win, in 1-2 hours, starting with dateModified.
2. Add Person on your author and team pages. Useful to readers, expected in regulated sectors.
3. Add HowTo on your guides and tutorials, on top of visible numbered steps.
4. Add Review/AggregateRating if you have customer reviews. They make your evidence machine-readable.
5. Add niche schemas (DefinedTerm, Service, ItemList) depending on your industry.

Common mistakes to avoid

• Don't duplicate schemas — one Organization per site, not one per page.
• Always validate with Google's Rich Results Test and the Schema Markup Validator.
• Don't fill them with invented data — AI engines cross-reference information. A numberOfEmployees: 500 when you are a team of 5 will be detected.
• Use JSON-LD, not microdata — AI engines parse JSON-LD more easily.
• Update dateModified on each real content change, not automatically on every visit.

Measuring the impact of advanced schemas

The free Detekia score counts only one structured data element (a named Organization, in the Entity clarity criterion): adding schemas will barely move it. Its citation test asks ChatGPT 5 questions for one page: note the result before adding your schemas, then run it again a few weeks later. This test only covers a sample of queries: to measure what AI engines say about you compared with your competitors across all your strategic queries, a full AI visibility audit takes over.

Stay cautious about the size of the gain. An experiment run by Otterly.AI (December 2025 – March 2026) saw a sharp rise in Google AI Overviews citations after adding markup (+611%), but a drop on ChatGPT (−71%), and 6 out of 7 AI platforms could not read the markup directly. Schemas work mostly through Google SEO and signal consistency, not as an automatic multiplier; Google itself notes that no special markup is required for its AI answers. And most of the gain comes from the first 3-4 well-populated schemas, not from quantity.

→Discover the 6 criteria of the Detekia GEO score

Conclusion

Advanced structured data isn't a technical luxury reserved for large companies. It's a standardized description of what your pages already show: an update date, steps, reviews, a definition. It does not replace what gets a page cited (short answers under the headings, tables, evidence, up-to-date content), but it removes ambiguity. Start with the visible content, then describe it properly.

→The complete GEO guide for 2026