Method
How this lab is set up, what it records, and what it refuses to publish.
This site exists to answer one question under controlled conditions: what does structured data do on its own, with nothing else helping it?
On a normal site the question cannot be answered. Age, links, brand recognition and existing rankings all push results around, and no amount of careful markup separates its effect from theirs. So this site removes them. It has no reputation to trade on and nothing to sell, and it says so on every page.
How a specimen works
A specimen is one schema.org type implemented in full on a working page. The page is written as that kind of page would really be written. The markup is complete, and it is truthful about a subject that is openly fictional.
The JSON-LD shown on the specimen page is the same object that is printed into the page head. The theme generates both from one stored value, so the page cannot display one thing and ship another. That is the only claim this site makes about itself that you have to take on trust, and it is the one you can check most easily: read the source.
What gets recorded
Four things are tracked per specimen, each with the date it was checked:
- Validator — whether the markup parses and passes schema validation.
- Rich result — whether the type is eligible for a rich result, per the testing tool.
- Indexed — whether the page is in the index at all.
- AI citation — whether an AI assistant cites the page when asked a question it plainly answers.
Nothing on this site is backdated. An observation carries the date it was taken. Where a check has not been run, it says not run, rather than showing a zero that looks like a finding.
What is deliberately withheld
Rating and review properties state other people’s judgement as fact to every parser that reads them, and a disclaimer on the page never reaches the parser. Where a specimen’s authored markup contains them and no real source is named, the theme strips them before printing and says so on the page. The collection would rather have a gap than a number nobody can check.
The control
At least one specimen carries no structured data at all. Without it, any result here would be a story rather than a comparison.