Skip to content



Our metadata formats are still in flux, and are likely to change substantially over the coming year.

A key principal of our data pipeline is that the data that flows through needs to be properly described and attributed to the correct source. This is done using a metadata system that is built into the pipeline.

Snapshot metadata

Snapshots are the raw data provided by upstream data providers. At minimum, they must capture:

  • The URL of the upstream data source
  • The license the data was provided under
  • A human readable description of the data

See the SnapshotMeta class for supported fields.

Dataset metadata

In our data model, datasets contain tables of data, and those tables contain variables. Each of these levels supports metadata.

See the classes DatasetMeta, TableMeta and VariableMeta in the owid.catalog.meta module for more details on the available fields.