Skip to main content
Structured data is machine-readable markup describing what a page contains, using Schema.org vocabulary. It is how a system knows a number on the page is a price rather than a weight. For AI search and shopping agents, it converts a page from prose to be interpreted into data to be used.

​
Format

Use JSON-LD in a <script type="application/ld+json"> block. It is Google’s recommended format and is easier to maintain than microdata because it is separate from the visible markup.

​
The types that matter for ecommerce

Product with a complete offers block carries the most weight for both shopping surfaces and agents.

​
Rules that cause markup to be ignored

These are the failures that appear most often in audits:
  • Markup must match the visible page. Marking up a price, rating, or availability that differs from what a user sees is a violation and can cost rich results entirely.
  • Do not mark up content that is not on the page. Reviews, ratings, and FAQs must be genuinely present and visible.
  • FAQPage requires real question-and-answer content, not marketing copy reshaped into questions.
  • Availability must track real inventory. Stale InStock on a sold-out product is worse than no markup.
  • Every required property must be present. Incomplete Product markup is frequently ignored wholesale rather than partially honoured.

​
Validate

Validate the rendered page, not the template. Markup assembled by JavaScript may differ from what you expect.

​
Does it drive AI citations?

Structured data reliably drives rich results and shopping surfaces. Its direct effect on citation in generated answers is less firmly established. The defensible position: structured data makes a page unambiguous, and unambiguous pages are easier to use as a source. It also happens to be required for surfaces where the effect is proven. It is worth doing on that basis alone — not because a specific citation lift can be promised.