Free Schema Checker

What a schema checker reads, and what schema markup can and cannot do.

Structured data is a set of claims a page makes about itself in a form machines can parse. Here is how it is written, what Google does with it now that several rich results are gone, and the traps that matter most for firms that answer to a regulator.

What schema markup is

Schema markup is structured data written in the schema.org vocabulary, a shared dictionary of types (Organization, Person, Article, LegalService) and the properties each type can carry (name, author, datePublished, sameAs). Google, Microsoft and Yahoo launched schema.org in 2011, and Yandex joined the same year, so that a page could describe its contents once and every search engine could read the description. The vocabulary now runs to hundreds of types, most of which no search engine does anything special with.

The same vocabulary can be written three ways. Microdata and RDFa are attributes woven into the visible HTML. JSON-LD is a separate block of JSON inside a <script type="application/ld+json"> tag, usually in the page head, which keeps the markup apart from the design. Google says it supports all three and recommends JSON-LD, and it is what WordPress SEO plugins and most site builders write. It is also the only format this checker reads.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@graph": [
    { "@type": "LegalService", "@id": "https://yourfirm.com/#firm",
      "name": "Your Firm LLP", "url": "https://yourfirm.com/" },
    { "@type": "Person", "@id": "https://yourfirm.com/#jane",
      "name": "Jane Doe", "jobTitle": "Partner",
      "sameAs": ["https://www.linkedin.com/in/jane-doe"] },
    { "@type": "Article", "headline": "What probate costs in 2026",
      "author": { "@id": "https://yourfirm.com/#jane" },
      "publisher": { "@id": "https://yourfirm.com/#firm" } }
  ]
}
</script>

That example is a graph: three entities in one block, joined by @id references, so the article points at its author and its publisher instead of repeating them. Yoast and Rank Math write graphs like this by default. It is also why a page can declare a Person without the word "Person" appearing at the top level of any block, and why this checker walks every nested object when it looks for an author or a firm.

A checker, a rich results test and a markup tool are three different jobs

Google once offered one tool for all of this, the Structured Data Testing Tool. It took that tool off its own site and split the job in two: the Rich Results Test, which answers a narrow Google question (is this page eligible for one of the result features Google supports, and are that feature's required properties present?), and a general markup tool it handed to schema.org in 2021, which reads all three formats and reports every schema.org type it finds without judging Google eligibility.

This checker is a third thing. It reads the JSON-LD in the HTML your server sends, lists the types, confirms each block parses, and runs the four markup checks Verand's site audit runs on every customer article: Article markup on an article, FAQPage where the page is built from questions, sameAs on the author or firm, and citation markup where there is a Sources section. It does not check required properties, and a pass here does not mean a page is eligible for any Google feature. Run the Rich Results Test for that. What this checker adds is speed on a single URL, the same result every time, and the audit's own view of the markup an article should carry.

Rich results, and the ones Google has retired

A rich result is a search listing with something extra drawn from structured data: breadcrumbs in place of a raw URL, a product's price, a recipe's rating. Google is explicit that correct markup earns eligibility and nothing more: it does not guarantee that structured data will show up in results even when the page is marked up correctly, and it chooses per query, per device and per searcher whether a plain result serves better.

Several features that site owners were told to mark up have since been withdrawn, which is the part most schema guides leave out. In August 2023 Google limited FAQ rich results to well-known, authoritative government and health sites; on 7 May 2026 it stopped showing them in Search at all, and in June it removed the FAQPage documentation (Google's changelog). The markup can still parse cleanly, but it produces nothing visible in Google. HowTo rich results were removed from mobile in August 2023 and from desktop the following month. Google's documentation changelog records other retired types too, among them Course Info, Estimated Salary, Learning Video and Vehicle Listing.

The practical reading for a firm: FAQPage markup still describes a question-and-answer page to any machine that reads it, and Verand's audit still flags an article of questions that has none, but nobody should add it expecting expandable answers under their Google listing. Put the effort into the markup that describes who you are.

Why markup has to match the visible page

Google's structured data guidelines say not to mark up content that is not visible to readers, and give the example that if the JSON-LD describes a performer, the page must describe that same performer. Markup that breaks the guidelines can bring a manual action, which removes the page's eligibility for rich results; Google says it does not affect how the page ranks in web search.

The reason is structural. Structured data is a claim the page makes about itself, and it is cheap to write. If search engines believed markup the page does not back up, every site would mark up five-star ratings, awards and credentials it does not have. So the markup is trusted as far as the visible page corroborates it. A jobTitle of "CFP" that appears nowhere in the byline, an AggregateRating with no reviews on the page, an FAQ block whose answers differ from the answers readers see: each one is at best ignored, and at worst the kind of thing a reviewer acts on. Write the page first, then describe it.

Schema markup for AI answers: what it does and does not do

Several pages ranking for this topic suggest schema markup gets a brand cited in AI answers. Google's own documentation for AI Overviews and AI Mode says the opposite of a special requirement: there are no additional requirements to appear in them, no special optimizations, and no special schema.org structured data you need to add. It lists structured data that matches the visible text as one of the ordinary SEO practices that still apply.

So say it plainly: structured data helps machines read what the page already says. It can make an author, a firm and a credential unambiguous, and a sameAs link to a LinkedIn profile or a regulator's record ties a name to one real person rather than every person who shares it. It does not make a thin page citable, and no checker, including this one, can tell you a page will be named by ChatGPT or Perplexity. What earns a place in an answer is the substance of the page. The markup describes it.

Schema types for law firms, advisory firms and medical practices

Generic Organization markup is where most professional sites stop. Schema.org has more specific types, and using the closest one tells a machine what kind of business you are before it reads a word of the page. The last column is what this checker does with the type: its sameAs check recognises a fixed set of entity types by exact name and does not follow the vocabulary's subtype tree.

"Not recognised by name" means that if the only entity on a page is, say, an InsuranceAgency with no Person or Organization beside it, the sameAs check finds no entity to test and passes without looking. Add a Person for the author, which most articles need anyway, and the check has something to read. On the person, the properties that carry credentials are jobTitle, hasCredential, memberOf, knowsAbout and sameAs. The strongest sameAs targets for a regulated professional are records a stranger can check: a FINRA BrokerCheck or SEC adviser page, a state bar directory entry, a licensing board lookup.

The review and rating trap

Review and AggregateRating markup on a firm's own site is the schema most likely to cause trouble in a regulated business, for two separate reasons. The first is Google's: since 2019 it has not shown review stars for LocalBusiness and Organization types when the business being reviewed controls the reviews, which it calls self-serving, whether the reviews are in the markup or an embedded widget. Marking up your own testimonials mostly produces nothing.

The second is the regulator's. For an SEC-registered investment adviser, client testimonials, endorsements and third-party ratings in marketing are governed by the Marketing Rule, Rule 206(4)-1 under the Advisers Act, which attaches disclosure and oversight conditions to each. A star rating in JSON-LD is a testimonial rendered in machine-readable form, stripped of whatever disclosure sits next to it on the page. Many state bars and medical boards have their own advertising rules on testimonials as well. Treat review markup as marketing that needs the same review as the testimonial itself, and flag it for whoever signs off your advertising before it goes live.

How schema gets onto a page

On WordPress, an SEO plugin such as Yoast or Rank Math writes an Organization or Person graph and an Article node for every post, filled from the settings you give it; the author profile fields are where sameAs links usually go. Site builders such as Webflow and Wix let you paste a JSON-LD block into a page's head. On a static or hand-built site, the block is written into the template. The one method this checker cannot see is markup injected by JavaScript, for example through Google Tag Manager: Google can render scripts and may read it, but any tool that reads the served HTML, this one included, will report the page as having none.

Common mistakes

• A block that does not parse. A trailing comma, a smart quote pasted from a document, an unescaped quotation mark inside a headline. JSON has no tolerance: one bad character and the whole block is unreadable. This is the check that runs on every page.
• An article with no Article type. Themes that emit only WebPage leave a post undescribed as a post. BlogPosting and NewsArticle count as Article here.
• An author with no way to verify them. A Person with a name and nothing else. Two or three sameAs links to profiles a reader can check turn a name into an identifiable person.
• Markup describing content that is not on the page. Ratings, awards or FAQ answers that exist only in the JSON-LD. Search engines ignore it at best.
• Two plugins writing two graphs. An SEO plugin and a theme both emitting Organization, with different names or logos. The type list on this card shows both; pick one source and turn the other off.
• Marking up for a rich result that no longer exists. FAQ and HowTo markup added in 2026 for the listing it produced in 2022.
• Assuming a clean result means eligibility. Every block parsing is the floor, not the finish. Required properties per feature are what the Rich Results Test is for.