In short. Keep Dataset.description between 50 and 5000 characters and use it as a short summary of the dataset.
Definition: Google Search Central requires the
Dataset.descriptionsummary to be between 50 and 5000 characters. The value is a short summary of the dataset, while the surrounding dataset metadata describes its subject, measured variables, and creator.
Key facts
- Google's Dataset structured data documentation places
descriptionamong the required properties and says its summary must be between 50 and 5000 characters. - The same documentation recommends limiting all textual properties to 5000 characters or less and says Google Dataset Search uses the first 5000 characters of any textual property.
- Google describes
descriptionas a short summary and permits Markdown syntax, with absolute path URLs for embedded images. - Google describes dataset metadata as information about what a dataset is about, which variables it measures, and who created it. It separates that metadata from specific values for the variables.
- Schema.org's Dataset definition describes a dataset as structured information about topics of interest and lists
descriptionasTextorTextObject.
What length does Dataset.description need?
Google's documentation, updated 8 September 2026, gives the direct answer: description is a required Dataset property, and its summary must contain 50 to 5000 characters. The same page recommends a 5000-character ceiling for every textual property and says Dataset Search uses only the first 5000 characters of any textual property. For this field, keep the complete summary inside the documented 50 to 5000 character range.
The lower and upper figures describe Google's Dataset structured data guidance. Schema.org supplies the vocabulary and value type, while the Google page supplies the feature-specific length rule. The Google page provides no sentence template or separate word-count rule.
What should the description say?
Google calls the value “A short summary describing a dataset.” Its guidance on dataset metadata says that metadata states what the dataset is about, which variables it measures, and who created it. That guidance describes the dataset record as a whole, so the description should identify the dataset's subject and scope in a compact summary, with related facts represented in their corresponding properties.
Schema.org defines Dataset as “A body of structured information describing some topic(s) of interest.” Its inherited description property accepts Text or TextObject and means “A description of the item.” These definitions support a plain-language summary of the dataset itself. They do not set a second character minimum for the field.
How should the value be formatted?
Google permits Markdown syntax in the summary. An embedded image uses an absolute path URL rather than a relative path URL. In JSON-LD, a line break is represented by the two characters \\n.
Google's page shows the description inside a Dataset JSON-LD object:
{
"@context": "https://schema.org/",
"@type": "Dataset",
"name": "NCDC Storm Events Database",
"description": "Storm Data is provided by the National Weather Service (NWS) and contain statistics on..."
}
The description is one required property in the example's Dataset object. Google also says that required properties must be included for content to be eligible for display as a rich result, and recommends validating the markup with the Rich Results Test.
Related terms
- Dataset structured data
- Structured data
- How can dataset markup describe measured variables and methods?
- How can a dataset describe its geographic and time coverage?
FAQ
What is the minimum length for Dataset.description?
Google's Dataset documentation says the summary must be at least 50 characters long.
What is the maximum length?
The documented maximum is 5000 characters. Google Dataset Search uses the first 5000 characters of any textual property.
Does the description have to list measured values?
Google describes dataset metadata as covering what the dataset is about, the variables it measures, and who created it. It describes the metadata as representing the dataset's contents and gives specific variable values a separate place.
Sources
- Google Search Central, "Dataset (Dataset, DataCatalog, DataDownload) structured data", 8 September 2026
- Schema.org, "Dataset", undated
Check how a page reads to answer engines. Open the GEO score analyzer