71% of pages cited by ChatGPT use schema markup. FAQPage schema delivers a 350% increase in AI citations. Organisation schema with @id and sameAs builds entity authority. Speakable schema has under 10% adoption but delivers a 127% citation increase. The @graph pattern connects all entities into a compound signal that AI platforms can verify.

Schema types ranked by AI search impact

Schema type	AI citation impact	Adoption rate	Priority
FAQPage	+350% citation increase	~25%	Critical
Organisation	Entity authority compound signal	~40%	Critical
Article + Author/Person	3x citation lift	~35%	High
Speakable	+127% citation increase	<10%	High (competitive edge)
HowTo	+20% higher citation rate	~15%	Medium
BreadcrumbList	Site structure signal	~50%	Medium
Product	Ecommerce citation driver	~30%	High (ecommerce)
AggregateRating	Trust signal for recommendations	~20%	High (review-dependent)

1. FAQPage schema - 350% citation increase

FAQPage schema delivers the highest measurable impact on AI citations. It produces a 350% increase in citation likelihood. AI platforms can extract question-answer pairs directly from structured data. They match these to user queries without parsing page content.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is AI search optimisation?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "AI search optimisation (also called GEO) is the process of making your business visible in AI-generated answers from platforms like ChatGPT, Gemini, Perplexity, and AI Overviews."
      }
    },
    {
      "@type": "Question",
      "name": "How much does AI search optimisation cost?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "AI search optimisation typically costs between £500-£5,000 per month from specialist agencies, depending on scope and competition level."
      }
    }
  ]
}
</script>

FAQPage best practices for AI

Match questions to actual search queries (use "People Also Ask" data)
Keep answers concise - 40-80 words is ideal for AI extraction
Include specific data in answers (percentages, numbers, dates)
Limit to 5-10 questions per page to maintain topical focus

2. Organisation schema - entity authority

Organisation schema with @id, @graph, and sameAs links builds entity authority. AI platforms use this to verify your brand exists and is trustworthy. Gemini cross-references Organisation schema with Google Business Profile, Wikidata, and LinkedIn to verify entity claims.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "@id": "https://example.com/#organization",
  "name": "Your Company Name",
  "url": "https://example.com",
  "logo": "https://example.com/logo.png",
  "description": "Your consistent brand description used everywhere.",
  "foundingDate": "2020",
  "areaServed": {
    "@type": "Country",
    "name": "United Kingdom"
  },
  "sameAs": [
    "https://www.linkedin.com/company/your-company",
    "https://www.wikidata.org/wiki/Q123456",
    "https://twitter.com/yourcompany",
    "https://www.youtube.com/@yourcompany",
    "https://g.co/kgs/yourGBPid"
  ],
  "contactPoint": {
    "@type": "ContactPoint",
    "contactType": "customer service",
    "email": "[email protected]"
  }
}
</script>

Organisation schema best practices

Use @id to create a persistent identifier other schema can reference
Include sameAs links to every authoritative profile (LinkedIn, Wikidata, YouTube, GBP)
Use areaServed - especially important for local and regional businesses
Match the description exactly to what appears on your GBP, LinkedIn, and other profiles
Include foundingDate - adds entity permanence signal

3. Article + Author/Person - 3x citation lift

Pages with Article schema that include a linked Person schema for the author see a 3x increase in AI citation likelihood. AI platforms use author authority as a quality signal. Connecting content to a known, credentialed person increases trust in the factual claims.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Your Article Title",
  "description": "A clear description of the article content.",
  "datePublished": "2026-04-01",
  "dateModified": "2026-04-06",
  "author": {
    "@type": "Person",
    "@id": "https://example.com/#author-jane",
    "name": "Jane Smith",
    "jobTitle": "AI Search Analyst",
    "url": "https://example.com/about/jane-smith/",
    "sameAs": [
      "https://www.linkedin.com/in/janesmith",
      "https://twitter.com/janesmith"
    ]
  },
  "publisher": {
    "@id": "https://example.com/#organization"
  },
  "mainEntityOfPage": "https://example.com/article-url/"
}
</script>

Author/Person best practices

Create an author bio page on your site and link to it via url
Connect the Person to LinkedIn and other professional profiles via sameAs
Include jobTitle to establish expertise relevance
Use @id so other pages can reference the same author entity
Always include dateModified - AI platforms use this for freshness

4. Speakable schema - 127% increase, under 10% adoption

Speakable schema marks content suitable for text-to-speech and AI voice assistants. It delivers a 127% citation increase. Fewer than 10% of websites implement it. This is the biggest competitive opportunity in AI schema. High-impact markup that almost nobody uses.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Your Article Title",
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": [".answer-capsule", "h1", ".key-stat"]
  },
  "publisher": {
    "@id": "https://example.com/#organization"
  }
}
</script>

Speakable implementation tips

Use cssSelector to target your answer capsule paragraphs
Include the page title (h1) as speakable
Mark any key statistics or data summary sections
Keep speakable content concise - 2-3 sentences per section
Test with Google's Rich Results Test tool

5. HowTo schema - 20% higher citation rate

HowTo schema marks step-by-step instructions in a format AI platforms can extract directly. Pages with HowTo schema see a 20% higher citation rate. It works best for guides, tutorials, and process-oriented content where users need clear sequential steps.

HowTo schema is particularly effective for service businesses that want to show expertise. A solicitor explaining "How to contest a will" or an accountant explaining "How to register for VAT" creates exactly the kind of structured, extractable content that AI platforms prefer.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "HowTo",
  "name": "How to Submit Your Site to Bing Webmaster Tools",
  "description": "Step-by-step guide to getting indexed in Bing for ChatGPT visibility.",
  "step": [
    {
      "@type": "HowToStep",
      "name": "Create a Bing Webmaster Tools account",
      "text": "Go to bing.com/webmasters and sign up with your Microsoft account."
    },
    {
      "@type": "HowToStep",
      "name": "Add your website",
      "text": "Enter your domain and verify ownership via DNS, meta tag, or CNAME."
    },
    {
      "@type": "HowToStep",
      "name": "Submit your sitemap",
      "text": "Navigate to Sitemaps and submit your XML sitemap URL."
    }
  ]
}
</script>

HowTo best practices for AI

Keep each step concise - one clear action per step
Use the name field as a short summary and text as the full instruction
Include 3-10 steps per HowTo block (too many steps reduce extraction quality)
Match the step names to how users would describe the process
Combine HowTo with Article schema to connect the guide to an author entity

6. BreadcrumbList - site structure signal

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "BreadcrumbList",
  "itemListElement": [
    { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://example.com/" },
    { "@type": "ListItem", "position": 2, "name": "Guides", "item": "https://example.com/guides/" },
    { "@type": "ListItem", "position": 3, "name": "Schema Markup", "item": "https://example.com/schema-markup-ai-search/" }
  ]
}
</script>

BreadcrumbList helps AI crawlers understand your site hierarchy and topical relationships between pages. It tells AI platforms where a page sits within your site structure, which helps with topical authority assessment.

BreadcrumbList best practices

Include BreadcrumbList on every page, not just deep pages
Use descriptive names that include relevant keywords (not just "Page 1")
Match the breadcrumb trail to your actual URL structure
Keep trails logical - Home > Category > Subcategory > Page

7. Product + AggregateRating - ecommerce AI visibility

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Product Name",
  "description": "Clear product description with key features.",
  "brand": {
    "@type": "Brand",
    "name": "Your Brand"
  },
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.7",
    "reviewCount": "342"
  },
  "offers": {
    "@type": "Offer",
    "price": "99.00",
    "priceCurrency": "GBP",
    "availability": "https://schema.org/InStock"
  }
}
</script>

ChatGPT mentions brands in 99.3% of ecommerce responses, averaging 5.84 brands per response. Product schema with genuine AggregateRating data feeds these recommendations.

The @graph pattern - compound entity signals

The @graph pattern connects multiple schema entities into a single, interlinked data structure. Instead of separate disconnected blocks, @graph tells AI platforms your Organisation, Person, Article, and FAQPage entities are all part of the same verified network. This compound signal is stronger than individual schema blocks.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "@id": "https://example.com/#organization",
      "name": "Your Company",
      "url": "https://example.com",
      "sameAs": ["https://linkedin.com/company/yourco"]
    },
    {
      "@type": "Person",
      "@id": "https://example.com/#author",
      "name": "Author Name",
      "jobTitle": "Expert Title",
      "worksFor": { "@id": "https://example.com/#organization" }
    },
    {
      "@type": "Article",
      "headline": "Article Title",
      "author": { "@id": "https://example.com/#author" },
      "publisher": { "@id": "https://example.com/#organization" },
      "speakable": {
        "@type": "SpeakableSpecification",
        "cssSelector": [".answer-capsule"]
      }
    }
  ]
}
</script>

How @graph strengthens AI signals

Connects author to organisation - establishes institutional authority
Links article to verified publisher - content trust signal
Uses @id references - avoids redundant data, creates clean entity web
Gemini specifically cross-references @graph entities with Google Knowledge Graph

Common schema mistakes that hurt AI visibility

The most common schema mistakes are missing Organisation schema (44% of UK SMEs), using schema without @id references, having sameAs links that point to dead profiles, and implementing schema that contradicts the visible page content. AI platforms detect inconsistencies and reduce trust when structured data does not match what users see.

Mistake 1: No Organisation schema at all

44% of UK SME websites have no Organisation schema. This is the single biggest gap we find in audits. Without Organisation schema, AI platforms have to guess your business name, type, and location from unstructured text. Adding Organisation schema is a 15-minute task that immediately strengthens your entity signal.

Mistake 2: Schema without @id references

If your Organisation schema does not include an @id, other schema blocks cannot reference it. This means your Article schema and Person schema float as disconnected entities. Use @id on every schema block and reference them using { "@id": "..." } instead of duplicating data.

Mistake 3: Dead sameAs links

If your sameAs array includes links to deleted social profiles or old domains, AI platforms find broken references. This reduces confidence in your entity. Audit your sameAs links quarterly. Remove any that point to pages that no longer exist.

Mistake 4: Schema that contradicts page content

If your Organisation schema says "London" but your page content says "Manchester", AI platforms detect the conflict and trust neither. Schema must match visible content exactly. Your business description in schema should be the same description that appears on your homepage, GBP, and LinkedIn.

Mistake 5: Missing dateModified on articles

AI platforms use dateModified to assess content freshness. 76% of cited pages were updated within 30 days. If your Article schema has no dateModified, AI platforms cannot verify freshness. Always include both datePublished and dateModified, and update dateModified when you genuinely update the content.

Mistake 6: Not using Speakable

Fewer than 10% of websites use Speakable schema, despite it delivering a 127% citation increase. This is the biggest competitive gap in AI schema. If your competitors are not using it (and statistically they are not), adding Speakable gives you an immediate edge.

Testing and validating your schema

Use Google's Rich Results Test and Schema.org's validator to check your schema. Test every page type, not just your homepage. Check that @id references resolve correctly, sameAs links work, and schema matches visible content. Run tests after every site update.

Testing tools

Google Rich Results Test (search.google.com/test/rich-results) - validates JSON-LD and shows which rich results your schema supports
Schema.org Validator (validator.schema.org) - checks syntax and structure against the Schema.org specification
Google Search Console - shows schema errors and warnings across your entire site

What to check

Every @id reference resolves to a real schema block
Every sameAs URL returns a live page
Organisation name matches your GBP, LinkedIn, and other profiles
dateModified is accurate and recent
Author Person schema links to a real bio page on your site
No duplicate schema blocks on the same page

Implementation checklist

Step	Schema type	Page type
1. Add Organisation schema	Organization	Every page (via site-wide template)
2. Add Person schema for authors	Person	Author bio pages + articles
3. Add Article schema	Article	Blog posts, guides, insights
4. Add FAQPage schema	FAQPage	FAQ sections, Q&A pages
5. Add Speakable	SpeakableSpecification	All content pages
6. Add BreadcrumbList	BreadcrumbList	Every page
7. Add HowTo schema	HowTo	Step-by-step guides
8. Add Product + Rating	Product, AggregateRating	Product pages
9. Connect with @graph	@graph	All pages
10. Validate	-	Google Rich Results Test

How each AI platform uses schema

Different platforms weight schema signals differently. Understanding these differences helps you prioritise.

Platform	Schema priority	Key behaviour
ChatGPT	Organisation, FAQPage	Uses schema to build entity profiles from Bing index data
Gemini	Organisation, Person, sameAs	Cross-references schema with Google Knowledge Graph and GBP
Perplexity	Article, dateModified	Prioritises freshness signals and source attribution
AI Overviews	FAQPage, Speakable, HowTo	Extracts structured Q&A and step-by-step content for answers
Claude	Organisation, Article	Uses schema in training data for entity understanding

Frequently asked questions

Does schema directly improve AI citations?

Yes. 71% of pages cited by ChatGPT use schema markup. FAQPage schema delivers a 350% increase in citation rates. Schema does not guarantee citation, but it significantly increases the probability.

Which schema type should I add first?

Organisation schema. It is the foundation that all other schema types reference. Add it site-wide through your template. Then add FAQPage to your most important content pages. Then add Speakable for competitive advantage.

Can I use a WordPress plugin for schema?

Yes, but check the output carefully. Plugins like Yoast, RankMath, and Schema Pro generate valid schema but often miss key fields like sameAs, areaServed, and @id references. Always supplement plugin output with manual additions for the fields that matter most for AI visibility.

How many FAQ questions should I add per page?

3-10 questions per page. Keep them focused on the page topic. Use questions from Google's "People Also Ask" and from your own customer enquiries. Each answer should be 40-80 words - concise enough for AI extraction but detailed enough to be useful.

Does schema work the same for ecommerce and service businesses?

The core schema types (Organisation, Article, FAQPage) work for both. Ecommerce sites should prioritise Product and AggregateRating schema. Service businesses should prioritise LocalBusiness (or ProfessionalService), Person, and HowTo schema. Read our guide on AI search for ecommerce for platform-specific advice.

Schema Markup for AI Search