Schema.org
Schema.org is a shared vocabulary that website operators use to label the content of their pages for machines: for instance, "this is a price," "this is a cooking recipe," "this is a date." Search engines and AI systems read out these labels instead of merely guessing at the text.
A normal web page consists of text, images, and layout. For a human, it’s immediately clear which part is the price and which is the opening hours. A computer program, on the other hand, sees only letters without meaning. Schema.org solves this problem with a fixed vocabulary of labels. The site operator attaches invisible tags to individual elements: here is a price, there is an author’s name, next to it a rating. The vocabulary was jointly launched in 2011 by Google, Microsoft, Yahoo, and Yandex and is free for everyone to use.
Why machines need labels
Without such labels, a program has to guess. If a page shows the number 24.90 with a euro sign, that’s probably a price. But it could also be a free-shipping threshold or a sample value in body text. Such misjudgments happen millions of times over and distort the result.
With a clear label, the guessing becomes unnecessary. That’s precisely why search engines display additional details directly in the results list for some hits: star ratings, cooking times, prices, dates. These enhanced hits are called rich results. They almost always arise from Schema.org markup in the page’s source code.
For site operators, this is a tangible economic incentive. A result with stars and a price stands out and gets clicked more often. But an important distinction applies: Schema.org does not automatically improve ranking position in the results list. It improves presentation and comprehensibility for machines. That’s a difference that is often confused.
Types, properties, and the JSON-LD format
At its core, Schema.org is a large list of types. A type is a category of thing, for example Person, Book, Restaurant, Event, or Product. Each type has permitted properties. A Product, for instance, includes name, price, brand, and availability. An Event includes start time, location, and organizer. In total, the vocabulary comprises more than 800 types.
The types are arranged hierarchically, similar to biology. A Restaurant is a special case of a Local Business, which in turn is a special case of an Organization. As a result, a Restaurant automatically inherits properties such as address and phone number. So anyone wanting to describe a new category doesn’t have to start from scratch.
Technically, this information is usually written into the page in the JSON-LD format. That’s a small block of data in the source code that remains invisible to visitors. It repeats the page’s information in machine-readable form. Older methods instead marked up the visible text directly, which was more error-prone. One basic rule always applies: the information in the data block must match what the visitor actually sees. Search engines treat incorrect labeling as deception and either ignore it or penalize the site.
From recipe search to AI assistants
Schema.org is most visible in everyday search queries. Anyone searching for a recipe sees ratings and preparation time right in the result. The same applies to job listings, movie showtimes, and product prices. Voice assistants also draw on such data when reading out an opening time.
In recent years, the term has become important again because of AI systems. Language models and search agents have to evaluate web pages without understanding them the way a human does. Cleanly tagged data noticeably lowers the error rate here. That’s why Schema.org comes up in discussions about how publishers and retailers keep their content discoverable for AI systems.
You’re likely to encounter the term mainly in two contexts: in search engine optimization and in reports about data standards on the web. If you open the source code of pretty much any news page, you’ll very likely find a Schema.org block with author, headline, and publication date.