In short. Dataset structured data uses Dataset, DataDownload and DataCatalog to describe data collections, their downloadable forms and catalog membership.

Definition: Dataset structured data is machine-readable metadata describing a collection of structured information. Schema.org's Dataset type identifies the collection, while related properties describe its subject, creator, access conditions and downloadable forms.

Key facts

As of October 2026, Google's Dataset documentation describes metadata for its Dataset Search tool and lists name and description as required properties. The description must contain between 50 and 5000 characters.
Schema.org's Dataset definition connects a dataset to downloadable forms through distribution. Different distributions can contain different subsets of the data.
Schema.org defines a DataDownload as all or part of a dataset in downloadable form. Its contentUrl identifies the actual file content.
A DataCatalog is a collection of datasets. The includedInDataCatalog property associates a dataset with its catalog.

What belongs in a dataset description?

Google's property definitions focus on metadata: what a dataset covers, which variables it measures and who creates it. Google includes tables, CSV files and collections of files among the formats that can qualify as datasets.

The name identifies the dataset, and description summarizes it. Google recommends distinct names for distinct datasets. Additional properties include creator for the author, identifier for an identifier such as a Digital Object Identifier (DOI), and variableMeasured for a measured variable such as temperature or pressure.

Coverage describes the scope of the observations. spatialCoverage identifies the geographic area, and temporalCoverage identifies the period covered by the data. These properties apply to datasets with the corresponding geographic or temporal dimension.

How are downloads and access conditions described?

Google's DataDownload guidance assigns separate roles to the page and download addresses: url locates the dataset description, while distribution.contentUrl provides the download link. For a download, Google lists contentUrl as required and recommends encodingFormat to state the file format. Multiple distributions can describe separate CSV and Excel downloads.

The W3C's JSON-LD specification defines a JSON-based format for linked data. The following illustrative JSON-LD example follows Google's structure and uses placeholder values and addresses:

<script type="application/ld+json">
{
  "@context": "https://schema.org/",
  "@type": "Dataset",
  "name": "Monthly store visits",
  "description": "Monthly counts of visits to stores, grouped by store location and month.",
  "url": "https://example.com/datasets/store-visits",
  "isAccessibleForFree": true,
  "distribution": [
    {
      "@type": "DataDownload",
      "encodingFormat": "CSV",
      "contentUrl": "https://example.com/files/store-visits.csv"
    }
  ]
}
</script>

Access and reuse have separate fields in the same documentation. isAccessibleForFree records whether access requires payment. license identifies the licence under which the data is distributed; Google recommends a URL identifying the specific licence version.

Where does the markup go, and how is it checked?

Google's implementation guidance recommends adding dataset markup to the canonical landing page: the main page describing that dataset. When descriptions also appear on listing pages, sameAs can identify the canonical dataset page.

The same guidance directs publishers to validate markup with the Rich Results Test and inspect deployed pages with URL Inspection. Google must be able to access the description page: robots.txt restrictions, a noindex directive or a login requirement can prevent access or indexing. A sitemap helps Google find the dataset URLs and follow changes.

FAQ

Can you describe a paid dataset?

Yes. Google's Dataset properties include isAccessibleForFree to record whether the dataset is available without payment. The description page still needs to be accessible to Google.

Does each download need its own Dataset object?

Schema.org's distribution property can repeat within a dataset description. You can therefore associate several downloadable forms with the same dataset, including forms containing different subsets.

How do you identify the catalog containing a dataset?

Use includedInDataCatalog with a DataCatalog object. Schema.org's catalog definition also provides the inverse dataset property for listing datasets contained in a catalog.

Sources

Check how a page reads to answer engines. Open the GEO score analyzer