1. Decide what the object describes
Write down the page’s purpose and intended object type before adding fields. An article, a service and the organisation that publishes them are different things.
Record each field’s source: visible heading, editorial date, official organisation name or verified URL. Leave out information without a reliable source.
2. Compare markup with visible content
Read the page as a visitor and compare every JSON-LD statement with visible text and components. Do not mark up hidden FAQs, invented ratings or prices absent from the page.
Check translations separately. The heading, description, canonical and breadcrumb should all belong to the same language version.
3. Validate the implementation and access
Parse JSON during the build and use appropriate validators for supported properties. A tool can catch technical errors but cannot establish that information is true or useful.
Check HTTP status, canonical, robots and crawlability. When a URL changes, review identifiers and internal relationships too.
4. Test the update path
Change a visible field in a page type and confirm its matching schema property updates. Also test removal, noindex and language switching so stale markup does not remain.
Document an owner and review steps. Passing a test promises neither rankings nor a rich result; markup should describe published content accurately.
| Field group | Review |
|---|---|
| Identity | Publisher and name are correct |
| Content | Information is visible and current |
| Language version | URL and text share a language |
| Release | JSON, canonical and crawlability are checked |