Insights · AI visibility

Schema markup for AI search: a plain guide for local businesses

What structured data is, which schema types a Costa del Sol business needs, a worked example for an estate agency, and how to test it.

Schema markup is a short block of code on your website that states facts about your business in a format machines can read without guessing: what you are, where you are, who you serve and how to reach you. It helps Google, and the AI tools built on top of search, understand you correctly. It is not a ranking switch, but it removes ambiguity, and ambiguity is what keeps a business out of AI answers.

What is schema markup?

Schema markup, also called structured data, is a shared vocabulary published at schema.org. Search engines agreed on it so that websites could describe themselves in the same way. Instead of a crawler reading “Open Monday to Friday, 9 to 6” and working out what it means, your page says openingHours: Mo-Fr 09:00-18:00.

The usual format is JSON-LD. It sits in a script tag in the page’s code, invisible to visitors. Google recommends JSON-LD, and it is the easiest format to add and maintain.

Which schema types matter for a local business?

Start with the type that describes your business, then add the ones that describe your pages.

  • LocalBusiness, or a more specific subtype. Use the most precise type that fits. RealEstateAgent for an estate agency. Dentist or MedicalClinic for a clinic. LegalService for a law firm. School for an international school. Restaurant or Hotel for hospitality. A specific type tells the machine more than a generic one.
  • Organization. Useful for the company behind the business, such as a parent company or a group with several branches.
  • Person. For the founder or key professionals. A dentist’s name, qualifications and profile links help a model connect the person to the practice.
  • Service. For each thing you sell, with a name, a short description and the area you serve.
  • FAQPage. For pages with real questions and answers. Google retired FAQ rich results for most sites in 2023 and now shows them mainly for well-known government and health sites. The markup is still valid, and it still describes your questions and answers clearly. Just don’t add it expecting extra space in Google’s results.
  • BreadcrumbList. Shows where a page sits in your site. Small, but it helps machines understand your structure.
  • Review and AggregateRating. Handle with care. Google’s rules say reviews a business publishes about itself on its own site are not eligible for review stars when the item is a LocalBusiness or Organization. You can still show testimonials on your pages, but don’t expect star ratings from them. Independent reviews on Google, Trustpilot or Doctoralia do more for you anyway.

Which properties do the work?

A handful of properties carry most of the value. Fill these in accurately before anything else.

  • name: your business name, written exactly as it appears on your Google Business Profile and directory listings.
  • url: your main website address.
  • address and geo: your street address and map coordinates. If you visit clients rather than receive them, say so with areaServed and leave the address general.
  • areaServed: the towns you cover. For many Costa del Sol businesses, this is the most important property. “Marbella, Estepona, Benahavís” is far clearer than “the coast”.
  • knowsLanguage: the languages you serve customers in. If someone asks an AI for an English-speaking dentist in Fuengirola, this is a direct, machine-readable answer.
  • sameAs: links to your profiles elsewhere, such as LinkedIn, Idealista, Doctoralia or Instagram. This tells machines that these profiles are all the same business.
  • openingHours: when you are open.
  • priceRange: a rough guide, such as “€€”.

What does it look like?

Here is a short example for a made-up Marbella estate agency. The name, address and links are illustrative, not a real business.

{
  "@context": "https://schema.org",
  "@type": "RealEstateAgent",
  "name": "Casa Pinar Properties",
  "url": "https://www.example.com/",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "Calle Ejemplo 12",
    "addressLocality": "Marbella",
    "postalCode": "29660",
    "addressCountry": "ES"
  },
  "geo": { "@type": "GeoCoordinates", "latitude": 36.49, "longitude": -4.95 },
  "areaServed": ["Marbella", "Nueva Andalucía", "Benahavís", "Estepona"],
  "knowsLanguage": ["en", "es", "nl"],
  "openingHours": "Mo-Fr 09:30-18:00",
  "priceRange": "€€€",
  "sameAs": [
    "https://www.linkedin.com/company/example",
    "https://www.instagram.com/example"
  ],
  "founder": { "@type": "Person", "name": "Ana Ejemplo", "jobTitle": "Founder" }
}

Every value in that block should match what a visitor can see on the page and what your other listings say. Schema that disagrees with your visible content is a red flag, not a shortcut.

Does schema help with ChatGPT and Gemini?

Honestly: partly, and indirectly. Google has said structured data helps it understand pages, and Google AI Overviews are built on Google’s search index. That is a clear route. OpenAI and Perplexity have not published how, or whether, their engines use schema markup when they read a page.

What schema does reliably is make your core facts unambiguous. A model that reads your page, your Google Business Profile and your Idealista listing is more confident when all three agree. Schema is one of the cleanest ways to state those facts. It is a foundation, not a lever. Reviews, specific content and independent mentions still decide most of who gets named. Our guide to how AI cites sources covers that side.

How to test your schema

Two free tools cover it.

  1. Google Rich Results Test. Paste a URL and it shows which structured data Google can read, and whether any of it qualifies for rich results.
  2. Schema Markup Validator at validator.schema.org. It checks your markup against the full schema.org vocabulary, including types Google does not show as rich results.

Fix errors first. Then read the warnings, because missing properties like areaServed are often where the value is.

A worked example: this site

This website uses the same approach. Every page carries a ProfessionalService block that lists the towns we serve, the two languages we work in, and the founder as a linked Person. Each article adds Article and BreadcrumbList markup, and pages with question and answer sections add FAQPage. None of it is clever. It just says the same facts the pages say, in a format machines read easily.

Schema is also one of several small technical files worth getting right. The other one people ask about is llms.txt. For the wider picture, see how to get into Google AI Overviews and what GEO is.

Where to start

Check your homepage in the Rich Results Test today. If nothing shows, or it shows a generic type with no areaServed or knowsLanguage, that is your first fix.

If you want to know whether AI engines already describe you accurately, our free AI visibility check tests ten real questions across ChatGPT, Gemini and Perplexity and compares you with three competitors.

See what ChatGPT says about you today

Ten real buyer questions. Three AI engines. You and three competitors. A short report in your inbox within five working days, free.

Get your free AI visibility check