Import dataset records

Asynchronously import a large volume of Data Set Records, supporting both create and update operations in a single request. This endpoint initiates a background job to process your request, allowing you to submit up to 1000 records per operation. Once complete, the job will provide a detailed report on the status of each record in CSV format.

Each record in the data array must include a mode field to declare the intended operation:

• Create — Creates a new record. Returns a per-record validation error if a record with the same primary key already exists, missing required columns or invalid permission key.
• Update — Updates an existing record matched by its primary key. The request must include the primary key column(s) for lookup. Non-primary-key columns are merged into the existing record. Returns a per-record validation error if no matching record is found. Primary key and permission key columns cannot be changed.

The completed job summary includes separate counts for created, updated, failed, and cancelled records, along with a report_url to download the full per-record CSV report.

This process is asynchronous; you can find a complete guide to the workflow in our Asynchronous Bulk Import API documentation.