Schema markup is structured code you add to a page to describe explicitly what it is: an article, a recipe, a product, an event, a person, an FAQ, a step-by-step guide. Without schema, Google and LLMs infer context from the HTML; with schema, you are telling them.
The format Google recommends is JSON-LD (JavaScript Object Notation for Linked Data), placed in the <head> or at the end of the <body>. Microdata and RDFa still exist but are on the way out.
The types that pay off
You do not need to implement every schema.org type. On SEO projects these are the ones that pay back:
- Article / NewsArticle: blog posts and editorial content.
- FAQPage: frequently-asked-question blocks. Enables the expandable accordion in the SERP.
- HowTo: numbered steps for processes. Rich result with visible steps.
- Product: product pages, with price, stock, ratings.
- Organization and LocalBusiness: the business entity, name, address, sister-brand relationships.
- BreadcrumbList: for navigable breadcrumbs.
- DefinedTerm / DefinedTermSet: for glossaries and definitions (this dictionary uses it).
Why it matters more in 2026
Two reasons. First, rich results beat the plain blue result in CTR: in many verticals a result with FAQ schema gets 20-30% more CTR than one without. Second, and newer, generative LLMs re-use schema as a structured source. When ChatGPT or Perplexity generate an answer, they prefer to cite fragments that come already structured, because they are easier to parse and attribute.
Validating the schema is not optional. Google Rich Results Test catches syntax errors and missing required fields. Broken schema is worse than no schema: Google ignores it and you miss the opportunity without knowing.