Dataset: Google Dataset Search & Metadata Markup

Dataset is the most "specialist" type in our guide: it won't give you a pretty card in regular web search, but serves a specialized product—Google Dataset Search. If you publish research data, statistics, government open data, or machine learning datasets, proper Dataset metadata helps those actually looking for data discover it. Its value isn't traffic—it's "being found by the right people."

What is Dataset#

Dataset describes a dataset—a collection of data that can be cited, downloaded, or accessed. Its primary destination is Google Dataset Search, a vertical search engine for researchers and data users, not rich media cards in regular web search.

Official documentation: Google Search Central: Dataset structured data.

Audience is data seekersDataset doesn't serve regular traffic. Before deciding to use it, ask: Do you actually publish datasets that can be cited/downloaded? If yes, use it; if no, don't force it.

Required & Recommended Fields#

Field	Required	Description
`name`	Required	Dataset name
`description`	Required	Specific, informative description
`distribution`	Recommended	`DataDownload`, including contentUrl, encodingFormat
`creator`	Recommended	Creator, Person or Organization
`license`	Recommended	License URL or description
`identifier`	Recommended	Persistent identifier like DOI
`variableMeasured`	Recommended	Variables measured in the data
`temporalCoverage`	Recommended	Time coverage range
`spatialCoverage`	Recommended	Spatial/geographic coverage range

Complete JSON-LD Example#

dataset.json

{
  "@context": "https://schema.org",
  "@type": "Dataset",
  "name": "Chinese Site Structured Data Adoption Rate Annual Dataset",
  "description": "Sampled statistics of structured data adoption rates on Chinese websites from 2020–2026, covering type distribution, JSON-LD share and error rates, aggregated by year.",
  "identifier": "https://doi.org/10.0000/example.sd-cn-2026",
  "keywords": ["structured data", "JSON-LD", "SEO"],
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "creator": {
    "@type": "Organization",
    "name": "MagicSEO",
    "url": "https://magic-seo.com/"
  },
  "temporalCoverage": "2020-01-01/2026-12-31",
  "spatialCoverage": "Mainland China",
  "variableMeasured": ["adoption rate", "type distribution", "error rate"],
  "distribution": [
    {
      "@type": "DataDownload",
      "encodingFormat": "text/csv",
      "contentUrl": "https://example.com/data/sd-cn-2026.csv"
    },
    {
      "@type": "DataDownload",
      "encodingFormat": "application/json",
      "contentUrl": "https://example.com/data/sd-cn-2026.json"
    }
  ]
}

distribution & Download Formats#

distribution is key to making datasets actually "usable." Each DataDownload:

contentUrl points to directly downloadable data files, URLs must be genuinely accessible;
encodingFormat specifies format, such as text/csv, application/json, application/vnd.ms-excel;
Multiple DataDownloads can be provided for the same dataset in different formats;
When providing via API, combine with appropriate fields to describe access methods.

Metadata for Discoverability#

Field	Why Important
`description`	Primary basis for users to judge relevance, be specific
`license`	Whether and how it can be used, directly impacts usability
`identifier` (DOI, etc.)	Persistent citation, essential for academic and data contexts
`temporalCoverage` / `spatialCoverage`	Time and geographic scope, helps precise filtering
`variableMeasured`	Explains what data measures, improves matching
`creator` / `publisher`	Source and credibility

License Information Must Be Clear#

No license = no one dares use itlicense tells users whether and how data can be used. Recommend pointing to URLs of standard licenses (like CC licenses) or providing clear descriptions. Missing license makes potential users hesitate and reduces dataset appeal.

Why This Site Doesn't Implement Dataset Yet#

This page is a tutorial article, not publishing a real downloadable, citable dataset. The DOI and download URLs in the example above are placeholders. Adding Dataset to a page without an actual dataset equals marking non-existent content.

Site demonstrationIf this site later publishes real SEO datasets (like the example topic above) and provides downloadable files with licenses, we'll add Dataset to that dataset page. This article continues using Article, BreadcrumbList, FAQPage.

Validation & Monitoring#

Use Rich Results Test or Schema validators to check name, description, etc.
Confirm distribution contentUrl is genuinely downloadable and encodingFormat correct.
Verify license, identifier, coverage and other metadata accuracy.
Search in Google Dataset Search to confirm discoverability.
Large datasets can combine with sitemaps to aid crawling.

Common Errors#

Error	Why Problematic	Fix
Vague description	Hard to judge relevance	Write specific, informative description
No distribution	Data not downloadable	Provide DataDownload + contentUrl
Missing license	No one dares use it	Give standard license URL or description
Applying Dataset to non-dataset pages	Marking non-existent content	Only mark real datasets
Broken download URLs	Promise not fulfilled	Ensure contentUrl accessible

Dataset Launch Checklist#

Actually publish downloadable/citable datasets
name and description complete and specific
distribution provides DataDownload (contentUrl + encodingFormat)
license uses standard license URL or clear description
identifier provides persistent identifier like DOI (if available)
temporalCoverage / spatialCoverage / variableMeasured supplemented as appropriate
creator / publisher indicate source
Download URLs genuinely accessible
Discoverable in Google Dataset Search
Validation tools show no critical errors

Frequently Asked Questions#

Does Dataset structured data produce regular rich results?

No, it does not appear as typical rich media cards in regular web search. Dataset structured data primarily serves Google Dataset Search, a vertical search product for researchers and data users. Its target audience is people looking for data, not general web search traffic.

What fields are required for Dataset?

Google's required fields for Dataset are name and description. Description should be written with sufficient detail and information value. To make datasets truly usable and discoverable, it's strongly recommended to supplement with recommended fields like distribution (including DataDownload's contentUrl and encodingFormat), creator, license, identifier, variableMeasured, temporalCoverage, and spatialCoverage.

How do I provide data download URLs?

Use the distribution field with one or more DataDownload objects. Each object should use contentUrl to point to directly downloadable data files and encodingFormat to specify the format (such as text/csv, application/json). If data is provided via API, combine with appropriate fields to describe access methods. Download URLs should be genuinely accessible.

Is license information important?

Very important. License tells users whether and how data can be used, significantly impacting dataset usability and credibility. Recommend using URLs pointing to standard licenses (like CC licenses) or clear license descriptions. Missing license information makes potential users hesitate and reduces dataset appeal in Dataset Search.