schema markup

advancedwidespread

Standard tags hidden in the website code that tell search engines and AI systems exactly what is what on the page.

In plain terms

You easily understand a website: you see that this is a recipe, that is the price, and the other is the opening hours. The machine must infer this from the text and can make mistakes during the process. Schema markup is like a label stuck on a jar: it does not change the content, but at a glance, it reveals what is inside, when it was made, and who put it there.

The tags come from a shared vocabulary called Schema.org. It contains predefined types (such as product, event, recipe, company, article, person), each with its own properties (price, date and time, preparation time, title, author). Since everyone uses the same vocabulary, the markup only needs to be written once, although individual search engines do not use every type.

The benefit is twofold. In search engines, your page may appear as a more visually striking result, for example with star ratings, prices, or event dates. AI search engines and agents can also more easily extract facts from it, as they do not have to guess which number is the price and which is the phone number.

It differs from plain text in that the visitor does not see it: the markup is for machines and does not alter the page's appearance.

How it works

First, you select what the page describes: a product, event, recipe, local business, or article. Each has its own type in the Schema.org vocabulary.
You fill in the fields that belong to the type: name, price, currency, date, title, author, rating. These must match what is visibly shown on the page.
You place the data in a small code block in the page source. The most common format is JSON-LD, which most e-commerce and content management systems create automatically or via an extension.
Search engine crawlers read the block when visiting a page and store the facts contained within. AI systems partly rely on the database built by these crawlers; it is unclear to what extent their own crawlers use the markup directly.
The search engine decides whether to use it: it can display an enriched result, embed the data in a knowledge panel, or the AI answer can rely on it. You can check with free verification tools whether your markup is flawless.

An everyday example

You search Google for the phrase 'túrós csusza recipe', and in one of the results you already see the photo, the 30-minute preparation time, and the 4.8-star rating in the list. The search engine does not guess these: it read them from the page's schema markup.

Who is affected

marketerA well-tagged page can achieve more prominent search results, and AI search engines can also interpret it more easily. → Marketing

entrepreneur, executiveThe accurate display of company data, prices, and opening hours in search engines and AI assistants' responses depends on this, alongside business profile data (such as Google Business Profile).

developerThey must embed the markup into the website templates and keep it up to date along with the data. → Code

everyoneDuring searches, you see stars, prices, and event dates in the results daily, which largely stem from this labeling.

A common misconception

MythIf I add schema markup to my site, I will rank higher in search results and definitely be included in AI answers.

In realityMarkup on its own does not improve ranking: it helps machines understand the page precisely and can make the result more prominent. Nothing guarantees an enhanced display or AI citation, and markup does not rescue weak content.

Limits and risks

The search engine decides whether to display the enriched result; even flawless markup may remain invisible.
The labeling must reflect what is actually visible on the page: search engines may penalize fabricated ratings or misleading data.
The range of supported types changes: Google has narrowed or discontinued several extended match types in recent years, so what worked last year may not work this year.
It requires maintenance: if prices or opening hours change but the markup remains old, machines will propagate incorrect data.

Similar, but not the same

meta tags (title, meta description) – Meta tags provide the page title and short description, while schema markup conveys detailed, categorized facts about the content.
expanded hit – The enriched result is what you see in the search engine, while schema markup is the raw material in the code from which it can be generated.
llms.txt – llms.txt is a separate file, proposed in 2024 and not an official standard, that would give AI systems guidance on an entire website; not every AI provider takes it into account. Schema markup, by contrast, is an established standard that labels the content of individual pages.

Related concepts

Broader

structured data, search engine optimization, szemantikus web, metaadat

Narrower

JSON-LD, mikroadat, RDFa

On the edge

expanded hit, generative search engine optimization, knowledge graph, llms.txt, AI visibility, AI search

A brief history

The Schema.org common vocabulary was launched in 2011 by Google, Microsoft (Bing), and Yahoo, and Yandex joined later that same year. The goal was that websites would not have to use different markup for each search engine.

Where things stand

By the end of 2026, schema markup will be considered standard equipment for websites, with most content management systems and e-commerce platforms generating it automatically. Attention is shifting from star ratings to how it helps AI search engines and agents understand content, though the extent of this benefit is debated.

Status: 2026 November

Also known as

schema markup · schema markup · schema markup · Schema.org · schema.org markup · structured data markup · structured data markup · JSON-LD markup

Sources

This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.