Three things changed in 2024-2026: (1) AI engines became major traffic sources and they read schema as authoritative, (2) Google's AI Overviews started preferring schema-marked sources for citation, and (3) the entity graph (Google's understanding of who you are) is now a ranking factor in its own right and schema is how you populate it.
Schema is the cheapest, fastest, most-leveraged on-page work you can do in 2026. The cost is one engineering sprint; the upside compounds for years.
Use JSON-LD. Always. Google has explicitly preferred it for years, every other engine handles it cleanly, and it does not entangle your structured data with your HTML markup. Microdata and RDFa are legacy formats; the only reason to use them is if you cannot inject script tags (which is essentially never).
Three tools in order: (1) Schema.org Validator (validator.schema.org) for spec compliance, (2) Google Rich Results Test for Google-eligibility, (3) your CMS preview for production rendering.
'Valid' has two layers: spec-valid (parses cleanly) and rich-result-eligible (Google is willing to use it for a rich snippet). You want both. A page can be spec-valid and still ineligible because Google requires specific properties for specific snippet types.
AI engines specifically reward: well-formed Article schema with `author` (Person with `sameAs` to LinkedIn / Wikipedia / authoritative profiles), `dateModified` (recency signal), `mainEntityOfPage`, and `publisher` with logo.
If you want to be cited inside ChatGPT and Perplexity answers, the single biggest schema upgrade you can make is adding a properly-marked Person entity for every author with sameAs links to their authoritative profiles. This is the Person-entity grounding step LLMs use to decide whose claims to trust.
Indirectly, strongly. Schema is not a direct ranking factor in classic Google blue-link rankings, but it is the dominant input to rich-snippet eligibility, AI-citation selection, and entity-graph strength — all of which materially drive visible search and AI traffic.
Only if you mark up content that does not exist on the page, or use FAQPage / HowTo for promotional content. Used honestly on real content, schema is purely upside.
Update it whenever the underlying content changes. The `dateModified` field is read by both Google and AI engines as a freshness signal — keep it accurate.
Yes — and you should. A typical article page should carry Article + BreadcrumbList + Person (author) + Organization (publisher). Combine them in a single JSON-LD block as a graph (`@graph`) for cleanliness.