Schema.org

Long definition

Schema.org provides the vocabulary; the delivery format (JSON-LD, Microdata, RDFa) provides the syntax. They're separate concerns often conflated — you use Schema.org to say "this is an Article with these properties" and JSON-LD as the mechanism to put that claim in the HTML.

The vocabulary is hierarchical. Everything descends from Thing:

Thing
├── CreativeWork
│   ├── Article
│   │   ├── NewsArticle
│   │   ├── BlogPosting
│   │   └── TechArticle
│   ├── Book
│   ├── Recipe
│   └── ...
├── Person
├── Organization
│   ├── Corporation
│   ├── LocalBusiness
│   │   ├── Restaurant
│   │   ├── Hotel
│   │   └── ...
│   └── ...
├── Product
│   └── ProductGroup
├── Event
└── Place

Each type accepts properties defined on itself or inherited from parents. NewsArticle inherits everything Article and CreativeWork accept, plus its own (dateline, printEdition, etc.).

Not every type produces rich results in Google. Google's structured data documentation lists the types that enable visual SERP features. Marking up types outside that list is still valuable — it helps entity resolution and knowledge graph coverage — but you won't see stars, prices, or FAQ expansions in results.

Schema.org vs other vocabularies: there's also Dublin Core, FOAF, GoodRelations (absorbed into Schema.org), and domain-specific ones. For SEO targeting Google/Bing/Yandex, Schema.org is the answer. Other vocabularies are for research, archives, linked data projects — not search visibility.

Common misconceptions

• "Schema.org is a Google standard." It's a joint initiative. Google is the heaviest user in SEO context, but the spec is governed by the Schema.org community (W3C community group).
• "You must use the newest version." Schema.org publishes incrementally with backward compatibility. Older types keep working. Only watch for deprecations (e.g. itemReviewed.aggregateRating patterns that shifted in 2020).
• "Schema.org replaces meta tags." It adds a parallel semantic layer. You still need <title>, <meta description>, Open Graph, etc. Schema.org and meta tags don't compete — they coexist.
• "More schema types = better rankings." No. Overlapping or inappropriate types confuse the parser and can trigger misuse penalties for rich-result eligibility. One primary type per page (Article, Product, Recipe) plus universals (BreadcrumbList, Organization) is the right scope.