Adobe Experience Cloud
AEP external audience API cannot start with an unhosted laptop CSV
Choose the right upload path before designing automation for a locally generated audience.
November 17, 2026
A local CSV and an API-driven external audience sound like two ends of the same workflow. In AEP they are different ingestion paths. A recent Experience League question asks how to call the external-audience API directly against a file on a local device without cloud storage. The API reference describes a cloud-storage source specification with a file path and cloud type; a laptop path is not that source.
Choose the path that matches the operating model
For an occasional human upload, Audience Portal provides an Import audience CSV flow. Adobe specifies permission requirements and CSV constraints, including a 1 GB maximum and no spaces or dashes in column names. Test the identity column and consent treatment before importing real records.
For repeatable API ingestion, Adobe's tutorial first uploads the CSV to a cloud storage provider, then uses the external-audience API to ingest it. Design the storage location, credentials, retention, and failure monitoring as parts of the pipeline. A local-file source connector is another UI ingestion path, but it is not evidence that the external-audience endpoint accepts an arbitrary local filesystem path.
Verify membership, not just request success
Confirm the audience creation operation completes, then sample expected and excluded identities. Document how a replacement file updates membership and when downstream activation sees the change. If the requirement prohibits cloud storage entirely, state that constraint before promising an API-only solution.
Write the handoff contract
Name who generates the file, who validates the identity column, who owns the storage endpoint and who confirms a completed ingestion operation. This turns “upload a CSV” into a repeatable business process with a clear failure owner.
Adovo’s view
The ingestion route is an operating contract. Choose it from the frequency, security boundary and owner of the data, then automate it. A clever API call cannot erase a missing source system.
