How Datasets Work

Datasets hold all data in Narrative's ecosystem.

Datasets in Narrative serve as the primary system of record for any piece of data that is processed, stored, or transferred within the platform. As the heart of the ecosystem, datasets organize and store data, structured in schemas, enabling the efficient and orderly processing of vast amounts of data.

Storage and Organization

A dataset works as a virtual container where data is organized and stored. Like a table in a relational database, each dataset holds numerous fields, akin to columns, representing structured data.

Datasets ingest data, which can then be overwritten if needed. Retention policies define how long the data is kept before it is automatically deleted. These policies ensure the datasets only retain the most relevant and up-to-date information.


A schema is the blueprint that defines how data in a dataset is structured. It outlines the types of the fields or columns that will exist in a dataset, along with their names, descriptions and any validation that should be applied to the field. Schemas in Narrative are immutable.

Valid types for data within Narrative include


Every dataset is owned by a single company. The data held within a dataset cannot be accessed by any other entity unless allowed by the owner company.

Permissions are given via access rules, which define who can view and manipulate the data. For further information on access rules, refer to the Access Rules Documentation.

In summary, datasets are pivotal in the data processing pipeline within Narrative's ecosystem. By encapsulating and structuring data, they ensure the efficient, secure, and clear flow of information across the platform.