What Is Schema.org? The Code That Introduces Your Business to Machines
Schema.org is a shared vocabulary that labels the information on your page as "this is a business, this is its name, this is its phone number". Search engines have read it for years; AI models draw on the same labels.
Your site says "Monday–Friday 09:00–18:00". When you read it, you understand those are opening hours. When a machine reads it, it's one of thousands of character strings on the page. Schema.org is the standard way to stick a label on that string that says "this is an opening time".
The short definition
Schema.org is a vocabulary that Google, Microsoft, Yahoo and Yandex founded together in 2011. It contains hundreds of types such as "LocalBusiness", "Product", "FAQPage" and "Service", and each type has fields: name, address, phone, price, rating, author…
Information written in this vocabulary is called structured data. The most common way to write it is JSON-LD: a small block of code placed in the page's <head>, invisible to visitors.
Why it matters for AI
Language models understand text by inference. Is "Protan" a business name, a person, a product? The model looks at the sentence, usually gets it right, sometimes gets it wrong. When your JSON-LD says "@type": "Organization", "name": "Protan", there is nothing left to guess.
When models search the web while generating an answer (Perplexity always, ChatGPT and Gemini often), the structured data on the retrieved pages directly affects the answer's accuracy. When a model gets your phone number wrong, the source is usually an unlabelled page.
Which types for businesses
Not every site needs every type. For a small or medium-sized business, the order of priority is:
- Organization or LocalBusiness — on the homepage, once. Name, logo, address, phone, opening hours, service area. If you have physical premises, pick one of LocalBusiness's subtypes: Dentist, Plumber, Restaurant, AutoRepair…
- Service — on every service page. The name of the service, who it's for, which area.
- FAQPage — on the frequently asked questions page. Question-and-answer pairs are the format models like best.
- Product — if you sell products; price, stock, rating.
- BreadcrumbList — page hierarchy; small but useful.
How to add it
If you use WordPress, plugins such as Yoast and Rank Math generate Organization and Breadcrumb automatically; for services and FAQs you usually need to add blocks by hand. On a custom-built site, you put the JSON-LD in the head section of your template.
A LocalBusiness example:
{
"@context": "https://schema.org",
"@type": "LocalBusiness",
"name": "Example Glass Balcony",
"url": "https://example.com",
"telephone": "+90 232 000 00 00",
"address": {
"@type": "PostalAddress",
"streetAddress": "Example Street No:1",
"addressLocality": "İzmir",
"addressCountry": "TR"
},
"openingHours": "Mo-Fr 09:00-18:00",
"areaServed": "İzmir"
}
Testing
Once the code is in place, validate it with Google's Rich Results Test or the Schema Markup Validator. Two errors are very common: invalid JSON (a missing comma, an extra bracket) and a value that contradicts what's visible on the page. The second is worse than the first: if your phone number reads one way on the page and another way in the JSON-LD, there's no telling which one to believe.
If you'd rather not write it from scratch, the Schema.org generator at protan.ai produces LocalBusiness, Service and FAQ blocks from a form. The free scan then shows which types are present on your site and whether they're valid.
Measure where your site stands in AI answers with 19 free checks; score and roadmap in minutes.
Keep reading
What Is llms.txt and How Do You Create One? A Guide with Examples
llms.txt is a small Markdown file at the root of your site that tells AI models "this is who we are, these are...
Why Does ChatGPT Recommend My Competitor and Not Me?
Same city, same service, similar website; yet your competitor is in the answer and you're not. The cause is us...
ChatGPT, Gemini, Claude and Perplexity – Which Does What?
Four engines, four different behaviours. One always looks at the web, one leans on Google's data, one is cauti...