Structured data validation can check whether markup contains the expected properties and whether those properties meet the standards used by the tool. However, a single page passing validation does not mean that entity relationships across pages have been fully established. Errors and warnings also need to be interpreted in light of the scope of the checks: is the tool checking Schema.org specifications or the requirements for specific search features? To decide whether the markup needs fixing, first confirm the tool, version, and validation checks used, rather than relying solely on whether validation passed.
Validation helps identify URLs and properties that need attention
When introducing SEO Spider 11.0 in 2019, Screaming Frog explained that it could extract JSON-LD, Microdata, and RDFa in bulk and validate them against Schema.org specifications and Google’s search feature requirements at the time. Rather than simply providing an overall score, these checks help identify URLs and properties that need attention by letting you:
- Filter URLs by whether they have structured data, the data format, and whether errors or warnings are present.
- View property values, validation types, severity levels, and specific issue messages.
- Distinguish issues with specifications from issues with search feature requirements, so different validation results are not confused.
In the article’s Product example, a missing image was classified as an error because the image was required at the time, while missing recommended properties were classified as warnings. This example comes from a historical version and cannot be treated as a rule for all Product markup today.
Page-by-page checks have limits when assessing cross-page relationships
In a 2026 Search Engine Journal article, Bill Hunt noted that most validation tools check properties one page at a time. They are suitable for checking standalone entities or the data required for rich results, but are less suited to evaluating interconnected knowledge graphs.
The article gives an example in which a branch page may use @id to reference an organization definition on the homepage. The address, legal information, and social profile data are stored in that definition rather than repeated on the branch page. The author noted that this arrangement may trigger warnings in page-by-page testing. When a check reports missing information, it is therefore important to distinguish between data that is genuinely absent and data located outside the page being tested. This does not, however, mean that all warnings can be ignored.
Likewise, having separate markup for branches, accounts, and loans does not necessarily establish which services are available at which branches. Checks of individual page properties and checks of entity relationships answer different questions.
Check the tool version and validation criteria before interpreting results
In its 2019 article, Screaming Frog reported that results from Google’s Structured Data Testing Tool sometimes differed from the search feature documentation in its tests. The Yandex validator also had inconsistencies, and its requirements differed from Google’s. These were the author’s findings at the time; they do not establish that current tools still have the same issues.
The article also acknowledged that Screaming Frog’s own new feature would not be perfect initially and recommended using it alongside Google’s tool at the time. When interpreting validation results, first identify what a tool checked, then consider the issues it did or did not find. A tool reporting no errors does not mean that the markup meets all standards or that all relationships across pages have been verified.
Structured data validation: perspectives from across the web
If a tool reports no errors, can formatting issues be ruled out?
No. In a 2019 example, Screaming Frog described Local Business markup that used “UK” as the country code. According to the author, this did not meet the requirements for two-letter country codes at the time, but Google's Structured Data Testing Tool did not detect the issue. This only shows that the issue was missed in that particular test; it does not establish that current tools would also miss it. To interpret actual results, you still need to confirm the tool, version, and checks used.
How should different results from validators checking the same markup be interpreted?
First identify which set of requirements each tool checks, then compare the specific issues rather than treating one result as definitive. Screaming Frog's 2019 tests reported that Google's tool at the time sometimes confused required and recommended properties. The Yandex validator also had inconsistencies, and its requirements differed fundamentally from Google's feature requirements. These were the author's findings from testing the tools at the time, not evidence that current versions still behave the same way.
Does support for bulk validation mean a tool covers all search features?
Not necessarily. The ability to process URLs in bulk and the range of features covered are separate matters. When introducing SEO Spider 11.0 in 2019, Screaming Frog stated that the version could extract three structured data formats at scale, but at the time it validated only 26 of Google's 28 search features. The absence of reported issues for a feature is therefore not enough to prove that it was checked. Confirm the coverage of the version being used.
References
- The Integrity Graph: The Missing Infrastructure Layer For AI Visibilitywww.searchenginejournal.com · 2026-06-10T13:30:06+00:00 — Source used in the article above.
- Screaming Frog SEO Spider Update – Version 11.0www.screamingfrog.co.uk · 2019-03-05T09:52:50+00:00 — Discusses: If a tool reports no errors, can formatting issues be ruled out?; How should different results from validators checking the same markup be interpreted?

