There is no single schema markup checker that answers every question, so use the one that matches yours. Google's Rich Results Test tells you which Google rich results your markup can generate and shows a preview. The Schema Markup Validator at validator.schema.org checks syntax and schema.org compliance for every type, with no Google-specific validation. After launch, the URL Inspection tool and the rich result reports in Search Console show what Google actually picked up from the indexed page. Google's own advice is to start with the Rich Results Test, and to keep watching the reports after deployment, because markup can break when templates change. A passing result is not a promise: Google states it does not guarantee that correct markup will show up in search results.
For this article, Sebastian, my AI employee for SEO, GEO and blogging, checked the structured data of kevinwelter.com: all 873 URLs in the sitemap, as of November 6, 2026, along with the parts that are not clean. Where structured data fits into a full audit is covered in How to Run an SEO Audit With AI; every article on the topic is collected under SEO and AI search.
Which schema markup checker tests what
The old Structured Data Testing Tool moved to schema.org in 2021 as the Schema Markup Validator, without the Google-specific checks. So a page can pass the validator and still miss Google's requirements. How to set up the account is in Setting Up Google Search Console.
What schema markup looks like in the page source
Google reads JSON-LD, Microdata and RDFa, and recommends JSON-LD because it sits in its own block instead of being woven into the visible text. This is the breadcrumb of an article on kevinwelter.com as it appears in the page source (from the German version, so the labels are German):
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "BreadcrumbList",
"itemListElement": [
{ "@type": "ListItem", "position": 1, "name": "Start",
"item": "https://kevinwelter.com/" },
{ "@type": "ListItem", "position": 2, "name": "SEO und KI-Suche",
"item": "https://kevinwelter.com/seo" },
{ "@type": "ListItem", "position": 3,
"name": "Website für die KI-Suche optimieren",
"item": "https://kevinwelter.com/blog/website-fuer-ki-suche" }
]
}
</script>
Readers see the same trail as a bar above the article. That pairing is the rule: Google says not to add structured data about information "that is not visible to the user, even if the information is accurate". And for Google Search behaviour, the Search Central documentation is definitive, not schema.org, which defines far more types than Google uses.
Structured data Google still shows (November 2026)
Google's feature list (page updated October 16, 2026) includes Article, Breadcrumb, Event, Local business, Organization, Product, Review snippet, Video, Profile page and Recipe, among others. For a typical business site or blog, four matter: Article, Breadcrumb (shown on desktop only, per the docs), Organization or Local business, and Video if you host your own.
Still in older tutorials, but without effect in Google Search:
Leftover markup doesn't hurt. Google, 2023: "Structured data that's not being used does not cause problems for Search, but also has no visible effects in Google Search." And on the 2025 removals: "This update won't affect how pages are ranked." So you don't have to rip out old FAQ blocks, but nobody should sell them to you as a fix.
Five things to get right before you test
- Pick the type from the page, not from the result you want. An example: marking up a coaching offer as
Product aims at merchant listings, but "Only pages where a shopper can purchase a product are eligible for merchant listing experiences". For a service without checkout, schema.org's Service is the honest type, even though Google has no feature for it.
- Mark up only what is visible. Violations can lead to a manual action, which removes rich result eligibility; per Google it "doesn't affect how the page ranks in Google web search".
- Fill required properties completely. Breadcrumb: at least two
ListItems, each with position and name, and item on all but the last. Video: name, thumbnailUrl, uploadDate. Article has no required properties; recommended are headline, image, datePublished, dateModified and author. Google prefers "fewer but complete and accurate recommended properties".
- Keep dates visible and identical. ISO 8601, ideally with a time zone, matching the date on the page. Why bumping a date without new content isn't optimisation is covered in Optimizing Your Website for AI Search.
- Treat a schema markup generator as a typing aid. Generators and plugins save typing, not checking: their output goes through the same tests and has to match the page.
Checking 873 pages on kevinwelter.com
Both web tools test one URL at a time through a form. For 873 URLs, Sebastian built his own run: fetch every sitemap URL one by one, extract the JSON-LD, and check it locally against the schema.org vocabulary, Google's required properties and the visible text. As a control, the checker first had to fail a test page with deliberately broken markup; it caught every planted error. On top of that came URL Inspection for eight pages, read-only.
Four findings remain. None of them is an error in the test's sense:
- References to details on other pages. 53 pages point via
@id to a node that is missing from the page itself, 47 of them to the website node that only exists on the homepage. That is valid schema.org; cleaner is the node on every page that uses it.
- Breadcrumb without a visible bar. On 40 pages, the team pages and five topic hubs per language, the trail exists only in the markup. URL Inspection still reports it as valid on the Sebastian team page. It doesn't quite fit Google's visible-content rule; the clean fix is a visible bar.
- Video without an author name. The six new video pages point to the author only by an identifier whose person details are not on the page. For a person given as
creator or author, Google requires name or alternateName. The video page that was inspected was still unknown to Google on the day of the check.
- Dates without a time. All articles give the day only. Google recommends a time zone; otherwise it uses Googlebot's.
URL Inspection added one more lesson: of five indexed pages with breadcrumb markup, it reported valid breadcrumbs on three and nothing at all on two, last crawled on October 21 and 25. Clean markup and what Google reports are two different measurements, and a check run needs both.
Schema markup and AI search
Structured data is often pitched as a lever for AI search. Google's guide to its generative AI features says: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." The same paragraph gives the reason to keep it anyway: it helps "with being eligible for rich results on Google Search". What does matter for AI search is in Optimizing Your Website for AI Search. Why an llms.txt changes nothing for Google Search is covered in llms.txt: Hype vs. Evidence. How to check whether your pages show up in AI answers is in AI Visibility Tracking.
Frequently asked questions
What is the best schema markup checker?
For Google Search, the Rich Results Test, because it shows which Google rich results your markup can generate. For pure schema.org validity, including types Google ignores, the Schema Markup Validator.
Can I check schema markup for a whole site at once?
Not with the two web tools; they test one URL at a time. After indexing, the rich result reports in Search Console cover the whole site. Before that, a crawl that extracts the JSON-LD, like the run above, does the job; some audit crawlers include schema validation only in their paid edition, see SEO Audit Tools Compared.
Does old FAQ or HowTo markup cause problems?
No. Per Google, unused structured data causes no problems and has no visible effect. A visible FAQ section can still help readers; building one from real customer questions is covered in Build FAQs From Customer Conversations.
Does schema markup improve rankings?
Nobody can promise that. Google describes structured data as a way to become eligible for rich results, without a guarantee, and says a structured data manual action doesn't affect web search ranking.
What to do next
Run one URL each of your three most important page types, say home, service page and article, through the Rich Results Test. Note the detected type, missing required properties, and any marked-up value that isn't visible. Drop whatever Google no longer shows, and check URL Inspection once it shows a crawl after your change; when an indexing request helps is in Google Indexing Request: When It Helps. For a whole site: how an audit with real tools is built is in How to Run an SEO Audit With AI, which audit tools include schema checks and what they cost in SEO Audit Tools Compared, and an SEO tool with an API under test in SE Ranking Review: API, MCP and Real Numbers. After markup comes linking: Internal Linking: Finding the Links You Miss. How hub and supporting pages fit together is in Pillar Pages and Topic Clusters.
Sebastian's template with tool chain, writing guide and project folder structure is in my community AI Employee Framework. If you want Sebastian working on your own site, get in touch via contact.
Your first AI employee up and running within an hour
Join the community$5/month founding price, first 50 members
