Schema.org for AI Search: 4 Schemas That Matter

Q: How do I validate my schema markup?

Use two free tools: (1) Google's Rich Results Test at search.google.com/test/rich-results — tells you which rich snippets your schema qualifies for. (2) Schema.org's official validator at validator.schema.org — surfaces structural errors and missing required fields. Run both. If either returns errors, the schema is invisible to AI engines.

Most small business websites have no structured data at all. To a human, the page looks fine. To ChatGPT, Gemini, and Google's AI Overviews, the page is a mystery — they have to guess what your business does, who you serve, and whether to cite you. Usually they cite a competitor instead. Here are the four schema types that fix that, with copy-paste JSON-LD you can drop into your site today.

TL;DR

Schema markup is the "machine-readable summary" AI engines use to confidently mention your business in answers.
You need four schemas: Organization (or LocalBusiness), FAQPage, Service, and Article on blog posts.
Use JSON-LD, not Microdata. Drop it in <head>, no visible-page changes needed.
Validate every schema in Google's Rich Results Test and schema.org's validator. Invalid schema = invisible to AI.
Most small businesses get one specific schema wrong — see "The one most get wrong" below.

What schema actually does for AI search

Schema.org is a shared vocabulary that lets you tell search engines, in their own machine-readable language, exactly what your page is about. It's not visible to your visitors — it sits in a <script type="application/ld+json"> tag in the page's <head>. But to AI engines, it's the cleanest, fastest signal they have about who you are.

Traditional Google SEO can survive without schema. AI search increasingly can't. When ChatGPT or Gemini decides whether to cite your business in an answer, it weighs the confidence it has that you are who you say you are. A page with valid Organization + LocalBusiness schema is a confident pick. A page with no schema is a coin flip — and AI engines hedge by citing the bigger brand or directory listing instead.

The gap: A 2026 audit of 500 small business websites found that fewer than 1 in 3 had any structured data at all. Of those that did, over 40% contained invalid or incomplete schema — which AI engines treat as if it weren't there. That means roughly 80% of small business sites are effectively unintelligible to AI search.

The fix isn't complicated. Three to four schemas, copy-pasted into your <head>, validated once, and forgotten. Below are the four that matter for almost every small business.

1. Organization (or LocalBusiness)

This is the foundation. Every page on every website should have one Organization or LocalBusiness schema — usually placed on the homepage, but it's fine to include it on every page. It tells AI engines your business name, where you are, how to contact you, and what entity you are.

If you have a physical address or geographic service area — a salon, restaurant, dentist, plumber, accountant — use LocalBusiness. If you're purely online — a SaaS, a remote agency, an ecommerce-only brand — use Organization.

Here's a copy-paste-ready LocalBusiness schema. Replace the placeholder values with your actual business details, then drop it in your homepage <head>:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "LocalBusiness",
  "name": "Your Business Name",
  "url": "https://yourdomain.co.uk/",
  "telephone": "+44 20 1234 5678",
  "email": "hello@yourdomain.co.uk",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "12 Example Street",
    "addressLocality": "London",
    "postalCode": "EC1A 1AA",
    "addressCountry": "GB"
  },
  "openingHoursSpecification": [
    {
      "@type": "OpeningHoursSpecification",
      "dayOfWeek": ["Monday","Tuesday","Wednesday","Thursday","Friday"],
      "opens": "09:00", "closes": "17:30"
    }
  ],
  "sameAs": [
    "https://www.linkedin.com/company/your-handle",
    "https://www.facebook.com/your-handle"
  ]
}
</script>

The sameAs array is doing more work than it looks. It cross-links your website to your social profiles — which gives AI engines independent corroboration that your business exists at the address you claim. Add every social profile you actively run. Skip any that are dormant; a dead Twitter from 2018 hurts you.

2. FAQPage

This is the single highest-ROI schema for AI search visibility. AI engines — especially Google AI Overviews and Perplexity — pull questions and answers directly out of FAQPage schema and quote them verbatim in their generated responses. If your page has good FAQ schema, you're statistically more likely to be the source the model cites.

Where to add it: any page that genuinely answers questions a customer might ask. Service pages. Pricing pages. Long blog posts. Don't fake it — Google can detect FAQ blocks that don't correspond to visible page content and will penalise the page.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "How much does an SEO audit cost?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Our comprehensive PivotScan™ audit is free. Paid Fix Delivery starts at £299."
      }
    },
    {
      "@type": "Question",
      "name": "How long does it take to see SEO results?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Technical fixes show within 2-4 weeks. Content and citation work compounds over 3-6 months."
      }
    }
  ]
}
</script>

Three rules: keep each answer under 60 words (AI engines truncate beyond that), match the visible page content exactly (no schema-only answers), and aim for 4-8 questions per page. More than 8 dilutes the signal; fewer than 4 isn't worth the schema's lift.

3. Service

If you sell services — not products — this is the schema that tells AI engines exactly what you do, who you do it for, and where. It's how a query like "AI receptionist for small business" gets matched to your AI Receptionist service page rather than a generic directory listing.

One Service schema per dedicated service page. If you have a page covering five services, you can include five Service blocks — or, better, split into separate pages with one Service each. AI engines reward focus.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Service",
  "serviceType": "AI Receptionist",
  "name": "AI Phone Receptionist for Small Businesses",
  "description": "24/7 AI-powered phone receptionist that answers calls, books appointments, and qualifies leads — without hiring a human.",
  "provider": {
    "@type": "Organization",
    "name": "Your Business Name",
    "url": "https://yourdomain.co.uk/"
  },
  "areaServed": { "@type": "Country", "name": "United Kingdom" },
  "offers": {
    "@type": "Offer",
    "price": "99",
    "priceCurrency": "GBP",
    "priceSpecification": {
      "@type": "UnitPriceSpecification",
      "price": "99",
      "priceCurrency": "GBP",
      "unitText": "MONTH"
    }
  }
}
</script>

The areaServed field is what turns this from a generic service into something AI engines feel safe citing for location-specific queries. Use Country for nationwide, AdministrativeArea for a region, or City for hyper-local.

4. Article (or BlogPosting)

Every blog post needs Article schema. It tells AI engines this page is editorial content, when it was published, who wrote it, and what it's about — all of which feed into whether your post gets pulled into an AI answer about that topic.

Use BlogPosting (a subtype of Article) for editorial posts. Use Article for any longer-form written content.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "Schema.org for AI Search: the 4 Schemas That Move Rankings",
  "description": "Short summary, 150-160 characters.",
  "image": "https://yourdomain.co.uk/og-image.png",
  "author": {
    "@type": "Organization",
    "name": "Your Business Name",
    "url": "https://yourdomain.co.uk/"
  },
  "publisher": {
    "@type": "Organization",
    "name": "Your Business Name",
    "logo": {
      "@type": "ImageObject",
      "url": "https://yourdomain.co.uk/logo.png"
    }
  },
  "datePublished": "2026-05-28",
  "dateModified": "2026-05-28",
  "mainEntityOfPage": "https://yourdomain.co.uk/blog/post-slug.html"
}
</script>

dateModified matters more than people realise. AI engines weight freshness heavily — a 2024 post with no dateModified looks stale. If you update a post, bump dateModified to the day you updated it.

The one schema most small businesses get wrong

If we audit ten small business sites that have made an effort with schema, eight of them have the same mistake: incomplete Organization schema. They've added the @type and name, maybe a url, and stopped there.

To AI engines, an Organization schema without address, telephone, sameAs, and logo is suspicious — it's just a name floating in space. The schema technically validates, but it doesn't add the trust signal that drives citations. AI engines treat partial Organization schemas as roughly equivalent to no schema at all.

The fix is filling in the fields you can corroborate elsewhere — your address (on your contact page), your phone (on your contact page), your logo (the image at the top of every page), and your social profiles (the sameAs array). If a field is genuinely not applicable — you're a remote-only consultancy with no phone — omit the field entirely. Empty strings or placeholders like "TBD" are worse than missing.

🛠️ Your 60-minute schema deployment checklist

Pick the right Organization variant. Physical address or service area? Use LocalBusiness. Purely online? Use Organization.
Draft your Organization/LocalBusiness JSON-LD from the template above. Fill in every field you have data for.
Drop it in <head> on every page (yes, every page — not just homepage).
For each service page: add a Service schema. One per page, with realistic areaServed and pricing.
For each blog post: add Article or BlogPosting schema with accurate datePublished and dateModified.
For pages with genuine FAQ content: add FAQPage schema mirroring the visible questions.
Validate every page in Google's Rich Results Test.
Cross-validate in schema.org's official validator. Fix every error.
Re-audit your site in 2 weeks — most AI engines re-crawl within that window.

How to validate schema (don't skip this)

Schema with errors is worse than no schema. Errors silently disqualify your page from rich snippets, FAQ accordions, and — increasingly — AI citations. Two free tools, takes 30 seconds per page:

Google's Rich Results Test (search.google.com/test/rich-results) — paste in your URL or raw HTML. It tells you which rich result types your schema qualifies for and surfaces required-field errors.
Schema.org Validator (validator.schema.org) — same input, stricter validator. Catches issues Rich Results Test waves through.

Run both. If either returns an error, fix it before moving on. If both return zero errors, you're done.

Frequently asked questions

Do small business websites really need schema markup in 2026?

Yes. AI search engines like ChatGPT, Gemini and Google AI Overviews rely on structured data to confidently cite your business in answers. Without schema, the AI has to guess what your page is about — and it usually guesses your competitor instead. Adding 3-4 well-formed schemas is one of the highest-ROI SEO tasks for a small business, and takes about an hour.

What is the difference between Organization schema and LocalBusiness schema?

Organization schema describes any business or entity. LocalBusiness schema is a subtype for businesses with a physical address that serves a geographic area — restaurants, dentists, plumbers, salons. If you have a shop front or service area, use LocalBusiness (it inherits everything from Organization and adds address, opening hours, and geo coordinates). If you're purely online, use Organization.

Will schema markup directly improve my Google rankings?

Schema is not a direct ranking factor for traditional Google, but it strongly influences how your pages appear in search — rich snippets, knowledge panels, FAQ accordions, and most importantly, AI Overview citations. In practice, sites with valid schema get more clicks and more AI citations, which compounds into stronger organic performance over 3-6 months.

What is JSON-LD and why is it the recommended schema format?

JSON-LD (JavaScript Object Notation for Linked Data) is the format Google, Bing and AI engines explicitly recommend for structured data. It lives inside a <script type="application/ld+json"> tag in your page's <head>, completely separate from your visible HTML. Compared to the older Microdata and RDFa formats, JSON-LD is easier to add, less likely to break, and doesn't clutter your page markup.

How do I validate my schema markup?

Use two free tools: (1) Google's Rich Results Test at search.google.com/test/rich-results — tells you which rich snippets your schema qualifies for. (2) Schema.org's official validator at validator.schema.org — surfaces structural errors and missing required fields. Run both. If either returns errors, the schema is invisible to AI engines.

What comes next

Schema is the technical foundation. It's necessary but not sufficient. Once your structured data is clean, the next levers to pull are AI search visibility tactics — llms.txt, GEO, AEO, and Q&A-formatted content. Together they take a small business website from "invisible to AI" to "actively cited by ChatGPT and Gemini" within 3-6 months.

If you'd rather not work through schema deployment yourself, our free PivotScan™ audit tells you exactly which schemas your site is missing and which existing ones contain errors. Or if you want it done for you, our Fix Delivery service includes per-page schema deployment as part of every tier.