How we check structured data

What we check

Three things are checked on every guide: the JSON-LD examples it shows, the markup pattern it recommends, and, for WordPress how-tos, the steps a reader is asked to follow. Examples are display only. They use clearly invented values such as example.com, Example Author or Example Recipe, never real people or businesses, and they are not emitted as live markup on our own pages.

Each example is written to the vocabulary on the relevant schema.org type page and to the requirements in the matching Google Search Central feature guide, so that a reader who copies it starts from something that validates.

Tools and versions

Two public validators do most of the work. The Rich Results Test shows which Google rich results the markup on a page or code snippet can generate, and flags errors and warnings against Google's requirements. The Schema Markup Validator checks markup against the whole schema.org vocabulary, including types Google does not use. Google's own testing page recommends starting with the Rich Results Test and using the Schema Markup Validator for generic schema validation (Google Search Central, test your structured data).

The label at the top of each guide names the tools used and the date of the check. Where a check involves software with a version number, such as WordPress, WooCommerce or a schema plugin, that version is recorded with it.

Live result checks

A clean result in a validator only shows that markup is eligible. Where a guide describes markup on a live page, the editors also check the live URL rather than a pasted snippet, confirm that the structured data matches the text a visitor can see, and use the URL Inspection tool in Search Console where a property is available to check the rendered page.

Google is explicit that it does not guarantee structured data will appear in search results, even when a page is marked up correctly (Google Search Central, general structured data guidelines). For that reason no guide here reports a rich result or an AI citation as something markup earns on demand.

AI crawler fetch checks

Not every crawler runs JavaScript the way a browser does, and a JSON-LD block injected by a script can be missing from what a crawler receives. For each page an example applies to, the editors fetch the raw HTML the server returns, without executing scripts, and confirm that the JSON-LD block, the visible answer and the dates are present in that response.

Where a guide discusses a specific AI crawler, it names it by user agent, such as OAI-SearchBot, GPTBot or PerplexityBot, and checks that the robots.txt rules being described would allow it.

Dates and rechecking

Every guide shows a published date and an updated date. The updated date equals the dateModified value in the page's markup and moves only when the content changes, not when a page is merely republished. Each recorded check carries its own date.

A guide is rechecked when the documentation it relies on changes, when a validator changes its output, or when a reader reports a problem. The recheck follows the same steps as the first check, and the guide's label is updated with the new date.

Limits

Validators show eligibility, not outcomes. Nobody outside a search engine can see how its ranking or answer systems weigh markup, and AI answer engines change their behaviour without notice. Guides therefore state what is documented, say where the evidence ends, and date any statement that may change with phrases such as "as of November 2026". To return to the guides themselves, go back to the handbook.