Choose a relevant type
Start with the main thing represented by the page: an organisation, person, article, product, event or breadcrumb trail. Do not add every schema type that sounds related, and do not mark hidden testimonials, invented ratings or unavailable offers.
A normal page with no useful structured-data case is not broken. Relevance and accuracy matter more than schema count.
Syntax is the first test
JSON-LD must be valid JSON: property names and string values use double quotes, trailing commas are not allowed, and multiple entities can be connected with @graph. Syntax validation does not establish eligibility for a search feature.
Keep data maintainable
Generate values from the same trusted source that renders the visible page. Product price, availability, dates and canonical URLs become misleading when hardcoded markup drifts from the interface.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "A practical delivery guide",
"datePublished": "2026-08-25",
"author": { "@type": "Person", "name": "Site owner" }
}
</script>Validate the current requirements
Use a schema validator for vocabulary and syntax, then consult the current documentation of any search feature you intend to support. Detection never guarantees that a rich presentation will be shown.