Dataset schema markup generator
Dataset markup is documented for Google Dataset Search. That is a different surface from ordinary Google Search, and marking up a dataset does not affect ordinary results.
Reference copy unfinished. The generator below works and produces valid markup. The explanatory sections are still being written, so this page is not listed for search engines yet.
Google Search features this type can produce
From the registry| Feature | Status | Surface | Last verified |
|---|---|---|---|
| Dataset (opens in a new tab) Surface is Dataset Search, not ordinary Search. Reports must not imply general Search eligibility from this feature. | ACTIVE | dataset_search | 2026-08-29 |
Meeting a feature’s documented requirements does not guarantee that the feature appears. No tool can promise that, and any that does is describing something it cannot know.
Generate the markup
Runs in your browser<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Dataset",
"name": "UK rainfall by county, 2019–2025",
"description": "Monthly total rainfall in millimetres for every county in the United Kingdom, from January 2019 to December 2025, derived from Met Office station readings.",
"url": "https://example.com/data/uk-rainfall",
"creator": {
"@type": "Organization",
"name": "Example Research Unit"
},
"distribution": {
"@type": "DataDownload",
"contentUrl": "https://example.com/data/uk-rainfall.csv",
"encodingFormat": "text/csv"
},
"license": "https://creativecommons.org/licenses/by/4.0/",
"temporalCoverage": "2019-01-01/2025-12-31"
}
</script>Paste this into the <head> of the page it describes. Markup that describes a page must live on that page.
The surface is the whole point
Most structured data guides talk about “being eligible” without saying eligible for what. Dataset is the type where that vagueness becomes obviously wrong: the documented destination is Google Dataset Search, a separate product with its own index and its own interface. Adding Dataset markup to a page does not make it eligible for anything in ordinary Search results.
That is not a criticism of the markup. If you publish research data, government statistics, or any structured collection that people search for as data rather than as a web page, Dataset Search is the surface you want, and this is how you reach it. It is simply a different question from the one most people think they are asking.
What a good Dataset entity carries
A name and a description are required, and the description does real work here — it is often all a researcher sees before deciding whether to open your page. Say what the data covers, over what period, and in what units.
creator, license and distribution are recommended and worth the effort. A distribution with a contentUrl and an encodingFormat is what turns “there is data here” into “the data is a CSV at this address”, which is the difference between being findable and being usable.
Mistakes worth avoiding
Expecting it to affect ordinary Search
It is documented for Dataset Search. Our reports name the surface on every finding for exactly this reason.
A description that describes the page, not the data
“Download our data here” tells a researcher nothing. State coverage, period and units.
No licence
Data with no stated licence is data most people cannot use. license is recommended, and it is the field that decides whether your dataset gets cited.
Check what you already have
If the page you are marking up already carries structured data, audit it before adding more. Duplicate entities from a theme and a plugin are the most common finding on established sites, and adding a third block makes it worse rather than better.