Schema Markup for AI: The Structured Data That Gets You Cited

Schema markup is the single fastest GEO improvement most UK businesses can make, and almost none of them have done it.

That’s not an opinion. It’s what came back when we ran our GEO audit tool against fifty UK service business websites last month. Around 80% had no JSON-LD schema markup at all, or only had basic Organization and WebSite schema — the equivalent of writing your name on a blank envelope and hoping the post office knows what to do with it.

This article is the technical guide. Which schema types matter for AI citation, how they actually work under the hood, what the data says about their impact, and how to implement them on a real UK service business website. Code examples included. No fluff.

Why schema matters more for GEO than it ever did for SEO

For traditional SEO, schema markup was always a “nice to have.” You could add it and maybe get a rich snippet in Google — a star rating, a recipe card, an FAQ dropdown. Nice, but not essential. Plenty of sites ranked #1 without a single line of JSON-LD.

For GEO, the calculation is different. AI engines don’t read your website the way a human does. They don’t browse, they don’t appreciate your design, and they don’t infer meaning from context. They parse. They extract. They look for machine-readable signals that tell them, with zero ambiguity, what your business is, what you do, who you serve, and what questions you can answer.

Schema markup is that signal. It’s the difference between an AI engine having to guess what your page is about from unstructured prose, and an AI engine reading a structured fact sheet that says “this is a plumbing business in Manchester, these are the services it offers, these are the questions it answers, this is its price range, these are its reviews.”

The data backs this up. SE Ranking’s November 2025 study found that sites with profiles on structured review platforms (Trustpilot, G2, Capterra, Yelp) have 3x higher chances of being cited by ChatGPT than sites without such presence. Our own audit data at GeoRankLocal shows an even starker pattern: sites with GEO-relevant schema (FAQPage, Service, LocalBusiness) consistently score 15-25 points higher on our rubric than sites with only basic Organization/WebSite schema, even when the underlying content quality is similar.

The reason is mechanical. When ChatGPT or Perplexity retrieves your page as a citation candidate, the RAG pipeline has to decide whether your content is trustworthy, relevant, and extractable. Schema markup directly addresses all three: it proves you’re a legitimate entity (trustworthy), it maps your content to specific topics and services (relevant), and it packages your information in a format the model can quote from without risk of misinterpretation (extractable).

The six schema types that actually matter for AI citation

There are hundreds of schema types defined at schema.org. Most of them are irrelevant for GEO. Here are the six that move the needle, ranked by impact.

1. FAQPage — the highest-impact GEO schema

FAQPage schema tells AI engines: “This page contains specific questions and specific answers. Here they are. Quote from them.”

This is the single most impactful schema type for GEO because it maps directly to how people use AI search. When someone asks ChatGPT “how much does an accountant cost in London?” the model is looking for a page that answers exactly that question. If your page has FAQPage schema with that exact question and a clear, factual answer, you’ve just handed the model a gift-wrapped citation.

From our audits, FAQPage schema is present on roughly 15% of UK SMB sites. The 85% without it are missing the easiest GEO win available.

How to implement it:

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "How much does an accountant cost in London?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Accountancy fees for small businesses in London typically range from £75 to £200 per month for basic bookkeeping and tax returns, rising to £300-500 per month for full management accounts, VAT, and payroll services. One-off self-assessment tax returns usually cost between £150 and £350."
      }
    }
  ]
}

The rules that matter:

• Each answer should be 60-120 words. Long enough to be substantive, short enough to be extractable as a single citation.
• Include specific numbers, prices, timeframes, and locations wherever possible. AI engines prefer concrete claims over vague hedging.
• Write in third person or direct declarative statements, not marketing copy. “Accountancy fees typically range from...” not “We offer amazing value for...”
• Aim for 5-10 FAQ items per page. Our scoring rubric gives additional points at 5+ items because it signals genuine topical depth rather than a token FAQ bolted on as an afterthought.
• The questions should match real conversational queries — the way someone would ask ChatGPT, not the way a marketing team would write a heading.

2. LocalBusiness — essential for any business with a physical presence or service area

LocalBusiness schema (or its subtypes like AccountingService, Plumber, LegalService, Dentist) tells AI engines where your business operates, what it does, and how to contact it. Without this, AI engines have to guess your location from page content, which they often get wrong.

{
  "@context": "https://schema.org",
  "@type": "AccountingService",
  "name": "A&Y Financial Services",
  "url": "https://ayfinancials.com",
  "telephone": "+44-20-XXXX-XXXX",
  "address": {
    "@type": "PostalAddress",
    "addressLocality": "London",
    "addressRegion": "Greater London",
    "postalCode": "EC1A 1BB",
    "addressCountry": "GB"
  },
  "areaServed": {
    "@type": "Country",
    "name": "United Kingdom"
  },
  "priceRange": "££",
  "openingHours": "Mo-Fr 09:00-17:30"
}

Critical details most implementations miss:

• Use the most specific @type available. Don’t use generic “LocalBusiness” when schema.org offers AccountingService, Plumber, Electrician, LegalService, Dentist, RealEstateAgent, and dozens of others.
• Include areaServed. This is what AI engines use when someone asks “best accountant in Manchester.”
• Include sameAs links to your Trustpilot, LinkedIn, Google Business Profile, and any industry directory listings.
• priceRange is optional but powerful. AI engines frequently cite price ranges when recommending businesses.

3. Service — one per service you offer

Service schema itemises exactly what your business does. If you’re a web design agency that also does SEO and GEO, you should have three separate Service schema blocks.

{
  "@context": "https://schema.org",
  "@type": "Service",
  "name": "GEO Website Build",
  "description": "Full AI-citable website built from scratch with schema markup, structured data, answer-formatted content, and technical SEO foundation.",
  "provider": {
    "@type": "Organization",
    "name": "GeoRankLocal"
  },
  "offers": {
    "@type": "Offer",
    "price": "2997",
    "priceCurrency": "GBP"
  }
}

Why this matters for GEO: When ChatGPT answers “who builds GEO-optimised websites in the UK?” it’s looking for services, not businesses. Service schema tells the model this specific service exists, at this specific price, from this specific provider.

4. Article — for every blog post and knowledge hub piece

Article schema tells AI engines that a page is a published piece of content with an author, a date, and a topic.

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "What Is Generative Engine Optimisation?",
  "author": {
    "@type": "Person",
    "name": "Awais M.",
    "jobTitle": "Founder of GeoRankLocal"
  },
  "datePublished": "2026-04-07",
  "dateModified": "2026-04-07"
}

The critical field is dateModified. Research from The SEO Works found that 76.4% of ChatGPT’s most-cited pages were updated within the last 30 days (The SEO Works, 2026). If you update an article, update the dateModified field. Every time.

5. HowTo — for process-oriented and tutorial content

HowTo schema breaks a process into numbered steps that AI engines can extract individually.

{
  "@context": "https://schema.org",
  "@type": "HowTo",
  "name": "How to Check Your AI Visibility",
  "step": [
    {
      "@type": "HowToStep",
      "position": 1,
      "name": "Run a GEO audit",
      "text": "Enter your website URL into a GEO audit tool to check schema markup, content depth, fact density, and citation signals."
    }
  ]
}

6. BreadcrumbList — site navigation for AI crawlers

BreadcrumbList schema tells AI engines how your pages relate to each other hierarchically. Small signal, but additive.

What we learned from scoring schema across 50 UK sites

From the GeoRankLocal audit data this month:

Around 80% had no GEO-relevant schema at all. They either had zero JSON-LD blocks, or only had basic Organization/WebSite schema that modern CMS platforms auto-generate.

The sites that did have FAQPage schema consistently scored 15-25 points higher on our rubric than comparable sites without it, even when the underlying content quality was similar.

Almost no sites combined multiple schema types. The ideal implementation — LocalBusiness + Service + FAQPage + Article + BreadcrumbList — was present on fewer than 5% of the sites we audited.

The most common mistake was duplicate or conflicting schema. Several sites had Organization schema with one business name and a Google Business Profile with a slightly different name. AI engines use cross-referencing to verify entity identity. Inconsistency drops citation confidence.

Consistency across all platforms matters more than having schema at all. Your business name, address, phone number, and service descriptions need to be identical in your JSON-LD schema, your Google Business Profile, your Trustpilot page, your Yell listing, and any industry directory.

How to check if your site already has schema

Before implementing anything, check what you’ve got:

1. Visit your website in Chrome
2. Right-click anywhere and click “View Page Source”
3. Press Ctrl+F and search for application/ld+json
4. Each match is a JSON-LD schema block — read the @type field to see what kind

Alternatively, paste your URL into Google’s Rich Results Test (search.google.com/test/rich-results) or Schema.org’s validator (validator.schema.org). Or use our free GEO audit tool — it checks schema as part of the overall assessment.

Implementation paths by platform

WordPress: Install Rank Math or Yoast SEO. Both generate LocalBusiness and Article schema automatically. For FAQPage, use the built-in FAQ block in the editor — both plugins detect it and generate the schema.

Shopify: Install the JSON-LD for SEO app by Ilana Davis. For FAQPage, add a custom JSON-LD snippet to your theme’s head.

Custom builds (Next.js, React): Add JSON-LD blocks directly in your page head as script tags. This is the approach we use at GeoRankLocal — each page template has its schema blocks defined as TypeScript objects in a dedicated lib/schema.ts file.

Squarespace / Wix: Both have built-in schema for basic types but limited FAQPage and Service support. Use the code injection feature to add custom JSON-LD.

The minimum viable schema set for a UK service business

If you do nothing else from this article, implement this combination:

1. LocalBusiness (or the most specific subtype) on your homepage
2. Service — one block per distinct service you offer
3. FAQPage — on every page with questions and answers, 5-10 Q&A pairs each
4. Article — on every blog post, with author name, datePublished, and dateModified
5. BreadcrumbList — on every page except the homepage

That combination takes a competent developer half a day on a custom site, or about two hours on WordPress with the right plugins. The impact on your GEO score is typically 15-30 points.

The bottom line

Schema markup is not optional for GEO in 2026. From the fifty UK sites we audited this month, 80% are missing it entirely. The 20% that have it are scoring measurably higher on every GEO readiness metric we track.

If you’re going to do one thing after reading this article, check whether your site has FAQPage schema. If it doesn’t, add it this week. It’s the fastest path from invisible to citable.

Sources

1. GeoRankLocal internal audit data, March-April 2026 (50 UK service business sites)
2. SE Ranking, “AI Citation Factors Study,” November 2025
3. The SEO Works, “75 AI SEO Statistics for 2026.” https://www.seoworks.co.uk/downloads/ai-seo-statistics/
4. Position Digital, “100+ AI SEO Statistics for 2026.” https://www.position.digital/blog/ai-seo-statistics/
5. Schema.org specification, https://schema.org
6. Google Rich Results Test, https://search.google.com/test/rich-results