Schema markup tells search engines what your content is — explicitly. Done well it earns rich results, AI Overview citations and knowledge graph alignment. Done poorly it warns and gets ignored. Here's the practical guide.
JSON-LD vs Microdata
Use JSON-LD. Microdata is legacy and harder to maintain. JSON-LD is a single self-contained block, easy to template, easy to validate, easy to debug.
The Core Types
Organization — sitewide, in your global head. Establishes your entity. WebSite — sitewide, with optional SearchAction for sitelinks search box. BreadcrumbList — on every page with a breadcrumb. Article or BlogPosting — on every content page, with author and dateModified. Product — on commerce pages, with offers, aggregateRating, brand. FAQPage — for genuine FAQ content (don't fake it). HowTo — for genuine how-to content.
The Schema Graph
The advanced move: combine multiple types into a single graph using @graph syntax. Link entities together with @id references. This builds a richer entity model and significantly increases the chance of knowledge graph pickup.
Validation Workflow
Every schema deploy goes through three checks: Google's Rich Results Test (for eligible rich result types), the Schema Markup Validator (for syntax correctness), and a manual inspection in Search Console once indexed. Schema that doesn't pass these three gets removed — broken schema is worse than no schema.
Common Mistakes
Marking up content that doesn't exist on the page. Wrong type for the content. Missing required properties. Stale data. Faking ratings. Each of these gets you a Search Console warning or a manual action.
Schema is invisible to users but it's a direct conversation with Google about what your content means. Have that conversation properly and you'll earn placements competitors never see.