Structured data: describe real facts and verify the markup
Structured data gives machines an explicit description of information on a page. Its value depends on whether that description is true, relevant and maintained—not on how many schema types you can add.
Choose a real information problem
Consider markup when an organisation, product, article or other supported entity is described ambiguously, or when an existing implementation contains an incorrect price, identity or relationship. A missing JSON-LD block is not automatically a defect: useful pages can be indexed without it, and a theme may use another supported format.
Leave accurate, maintained markup alone unless you have identified a specific gap. Adding a second generator can introduce conflicting facts. Begin by listing what the visible page says, where those facts are maintained and which component currently publishes structured data.
Distinguish meaning, syntax and eligibility
Google supports JSON-LD, Microdata and RDFa. JSON-LD often makes maintenance simpler because its data is separate from the surrounding HTML. A successful JSON parse establishes syntax; it does not prove that the entity, claims or feature-specific properties are correct.
Google’s quality rules require relevant, accurate information that represents the page. Do not invent reviews, awards, stock, credentials or business locations. Even valid, eligible markup does not guarantee a rich result. Schema.org vocabulary and Google’s supported search features are related, but not interchangeable lists.
Start with a small, verifiable example
Suppose the fictional Harbour Bicycle Studio homepage visibly identifies that business and links to its own website. This illustrative organisation description contains only those facts. It is not a complete template for product offers, local-business features or review stars.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Organization",
"@id": "https://harbour-bicycle.example/#organization",
"name": "Harbour Bicycle Studio",
"url": "https://harbour-bicycle.example/"
}
</script>Replace every example value before use. Keep a stable identifier for the same entity across templates. Add properties only when applicable and verified; a fabricated logo URL or address is worse than an omitted optional property. Google’s organisation documentation explains its supported properties and does not require every possible field.
For a product or other specialised feature, start from that feature’s current documentation instead. A business-identity example cannot establish product availability, shipping promises or review eligibility.
Implement once, close to the source of truth
- Inventory existing output. Look for JSON-LD scripts and other structured-data formats in the delivered page and relevant templates.
- Choose one owner for each entity. Decide whether the theme, plugin, application or custom template should generate it.
- Map fields to maintained data. Prices should come from the same catalogue that displays prices; identity should come from a reviewed business record.
- Preview one representative page. Compare every claim with visible content before applying the pattern elsewhere.
WordPress: if an installed SEO or commerce plugin already generates the needed entity, correct its data or documented settings. Do not assume the block editor preserves scripts or that every theme has the same schema controls. A developer should handle template-level output when appropriate.
Shopify: themes commonly supply product structured data. Inspect theme and app output before adding another block. Change catalogue facts in their owning fields and test a theme duplicate when code changes are necessary.
Custom HTML: generate JSON with a serializer, using the framework’s safe embedding method. Do not interpolate untrusted strings into a script element. Protect against values containing closing-script sequences and confirm that escaping preserves the intended data.
Test the document, then the search feature
First check that the document parses and describes the intended entity. Then use Google’s Rich Results Test for supported search features. A generic schema validator answers a different question: whether the vocabulary is used consistently, not whether Google will show an enhancement.
Inspect the public URL after release, not only an editor preview. Where you have access, review Search Console enhancement or manual-action information relevant to your implementation. Confirm that the page remains accessible under its intended policy and that no visible layout or checkout behaviour changed.
Keep the previous template, plugin configuration and representative output. If new errors or conflicting facts appear, remove the new block or revert the narrow change. Do not delete unrelated working entities as a blanket repair.
Keep expectations proportionate
Common errors include copying a competitor’s reviews, publishing expired offers, assigning the wrong entity type and treating every optional-property warning as mandatory work. Avoid promises that a schema block will make an AI assistant recommend your business.
SingCompare checks JSON syntax and selected published fields in supported JSON-LD profiles. Open “Structured-data review” in your diagnostic for script numbers, field paths, source evidence and suggested corrections. Unknown contexts or representations remain unassessed; optional fields are not required. Proposed values need your review, and a fresh diagnostic can check the published result. This is not a complete Schema.org validator, Google eligibility test or independent fact checker. It does not prove rich-result appearance or AI citations. Keep the business facts understandable in the visible page, and use clear content practices alongside accurate markup.
Sources and review date
Official documentation reviewed on 17 November 2026. Platform controls and search policies can change. Examples are fictional; implementation and prioritisation advice is SingCompare’s own.