AEO AEO Content AI Get free audit

Knowledge · criteria

Schema.org JSON-LD: The Scoreboard AI Actually Reads

Tidio runs 4 JSON-LD schema types. Crisp runs zero. That's not a coincidence -it's the difference between a 63 and a 34. Structured data is the machine-readable layer AI trusts most.

Add JSON-LD script blocks to your pages with Schema.org types -Organization, FAQPage, Article, BreadcrumbList. In our audits, sites with 3+ schema types score 15-20 points higher than sites with none. Tidio uses 4 types and scores 63. Crisp uses zero and scores 34.

What this article answers

  • What is Schema.org JSON-LD and how does it help AI find my site?
  • Which JSON-LD schema types should I add to my website?
  • How do I add structured data to my homepage and blog posts?

Key takeaways

  • Add Organization JSON-LD to your homepage first - it is the single highest-impact schema type and takes 10 minutes.
  • Use JSON-LD format (not Microdata or RDFa) because it is what Google and AI systems prefer.
  • Validate all schema with Google Rich Results Test before deploying - invisible JSON-LD errors sit undetected for months.
  • Keep entity names consistent across every schema block and every page on your site.
  • Add FAQPage schema to any page with Q&A content for both AI extraction and Google Rich Results.

What Is JSON-LD Structured Data?

Schema.org structured data is a standardized vocabulary you embed in your HTML to help machines understand your content -not just read it, but parse it into facts.

JSON-LD (JavaScript Object Notation for Linked Data) is the format. You drop a script block into your page and suddenly AI systems don’t just see text -they see structured knowledge:

  • This page describes a Product with a specific price and availability
  • This article was written by a specific Person with these credentials
  • This business is an Organization at this address with this phone number
  • These are Frequently Asked Questions with verified answers

Here’s what an Organization block looks like:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Business",
  "url": "https://example.com",
  "description": "What your business does",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "123 Main St",
    "addressLocality": "City",
    "addressRegion": "ST",
    "postalCode": "12345"
  }
}
</script>

That’s raw, machine-readable fact. No ambiguity. No interpretation needed.

Why Does Structured Data Give Sites a Scoring Advantage?

Put on Claude’s glasses for a moment. You’re scanning a website about live chat software. One site has four JSON-LD schema types -Organization, WebSite, Product, FAQPage -all machine-parseable. Another site has… divs. Just divs all the way down.

Which one do you trust with a citation?

We’ve tracked this across the entire live chat vertical. Tidio (score: 63) runs 4 schema types. LiveChat (59) runs 3. Crisp (34) runs zero. Zero. HelpCrunch (33) -also zero. The pattern isn’t subtle.

Here’s what structured data buys you:

  • Google Rich Results: FAQ dropdowns, star ratings, product carousels -all powered by schema
  • AI citation accuracy: when your data is structured, AI cites specific facts (prices, ratings, hours) instead of hallucinating them
  • Entity recognition: schema connects your business to the knowledge graph as a real entity, not just another website
  • Voice search answers: smart assistants rely on structured data for spoken responses

The culprit behind most low scores? Not bad content -missing metadata. The content exists but AI can’t parse it into facts.

Which Schema Types Should You Add to Each Page?

Every page type needs different schemas. Here’s what goes where:

Homepage -Organization + WebSite + SearchAction:

{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "name": "Business Name",
      "url": "https://example.com",
      "logo": "https://example.com/logo.png",
      "sameAs": ["https://instagram.com/handle"]
    },
    {
      "@type": "WebSite",
      "url": "https://example.com",
      "potentialAction": {
        "@type": "SearchAction",
        "target": "https://example.com/search?q={search_term_string}",
        "query-input": "required name=search_term_string"
      }
    }
  ]
}

Product pages -Product schema with offers, reviews, and domain-specific properties.

Blog articles -Article or BlogPosting with author, datePublished, and image.

FAQ pages -FAQPage with Question and AcceptedAnswer pairs. This one’s a force multiplier -it makes your FAQ content parseable AND eligible for Google’s rich results.

Validate everything with Google’s Rich Results Test (search.google.com/test/rich-results) and Schema.org’s validator (validator.schema.org). Don’t guess. Test.

Start here: Add Organization JSON-LD to your homepage today. It takes 10 minutes and it’s the single highest-impact schema type.

What Are the Most Common Structured Data Mistakes?

Schema that doesn’t match visible page content. Google will penalize you. If your schema says “4.8 stars” but your page shows “4.2 stars” -that’s a trust destroyer.

Microdata or RDFa instead of JSON-LD. JSON-LD is the format Google and AI systems prefer. It’s cleaner, easier to maintain, and doesn’t tangle with your HTML structure.

Inconsistent entity names. “Acme Inc” on one page, “Acme Incorporated” on another, “ACME” in the schema. Pick one. Stick with it everywhere.

Missing required properties. Every schema type has required and recommended fields. Missing them silently degrades your structured data quality.

Never validating. We’ve seen sites with broken JSON-LD sitting undetected for months. Invisible errors are the worst kind -you think you’re covered and you’re not.

Duplicate schemas. Two Organization blocks on one page confuses parsers. One per type, one truth.

Score Impact in Practice

Structured Data Coverage carries 5% weight in the Content Organization tier - one of the heavier individual criteria. Sites with 3+ JSON-LD schema types consistently score 7-9/10 on this criterion. Sites with zero schema score 0-1/10 with no partial credit for Microdata or RDFa.

In the live chat vertical, the correlation is direct. Tidio deploys Organization, WebSite, Product, and FAQPage schemas and scores 63/100 overall. LiveChat runs three schema types and scores 59. HelpCrunch and Crisp deploy zero schema types and score 33 and 34 respectively. The structured data gap alone doesn’t explain the full point difference, but it’s a reliable indicator - sites that invest in schema tend to invest in other AI-readability criteria too.

Our own site (aeocontent.ai, 88/100) runs Organization, WebSite, Service, FAQPage, Article, BreadcrumbList, CollectionPage, and Dataset schemas. That coverage contributes to a 9/10 on this criterion. The specific schema types matter less than the coverage - having any four types is significantly better than having one, because each type gives AI a different lens into your business.

How AI Engines Evaluate This

AI engines don’t just check whether you have structured data - they use it as a primary knowledge extraction layer, and each engine leans on different schema types.

ChatGPT relies heavily on Organization and Product schemas when building its understanding of businesses. When a user asks “What does company X do?” ChatGPT looks for Organization schema first because it provides the most concise, machine-parseable answer. FAQPage schema is ChatGPT’s preferred format for Q&A-style queries - it can extract individual question-answer pairs directly from the structured data without parsing the surrounding HTML at all.

Claude applies a stricter validation on schema consistency. If your Organization schema says “Acme Corp” but your WebSite schema says “ACME Corporation,” Claude notices the inconsistency and downgrades its confidence in both signals. Claude also cross-references sameAs links in Organization schema against actual social media profiles, treating verified external links as corroboration of entity identity.

Perplexity uses structured data as a speed optimization. When Perplexity processes a page with JSON-LD, it can extract facts from the schema in milliseconds rather than parsing the entire HTML document. This means pages with schema get processed more completely within Perplexity’s time budget. Pages without schema may get partially processed or skipped when Perplexity is under load.

Google AI Overviews consume structured data through the same pipeline as traditional search rich results. FAQPage and HowTo schemas are particularly valuable here - they map directly to the structured answer format AI Overviews uses in search results.

External Resources