Skip to main content
For analytics, compliance, or backup, export conversation transcripts in bulk. Exports run asynchronously: you request one over a date window, poll until it finishes, then download the resulting files with short-lived, per-file tokens.
Exports require the conversations:export permission, and an organization can have only one active export at a time (409 otherwise).

​
Request an export

Provide a required start/end window and, optionally, the same filters available on the conversation list (channel, status, end user, and more).
The endpoint returns 202 with a job object. export_format is csv (default) or ndjson.

​
Poll for completion

Poll the job until its status is terminal:
Status moves through pending → running → succeeded (or partial, failed, expired). Once it’s succeeded or partial, the files array carries a download_token per file.

​
Download the files

Download each file with its token before the export expires:
For a single conversation’s transcript, use the paginated GET /v1/conversations/{id}/turns endpoint instead — see Run a Conversation.

​
Next steps

Privacy & data protection

Set retention policies that govern how long transcripts are kept.

Analytics

Query aggregate metrics without exporting raw transcripts.