SegmentStream layers cross-channel attribution and budget optimization on top of Amplitude — reading raw event data exported from Amplitude's Destination Catalog into your data warehouse, not the Amplitude reporting interface.
Amplitude exports raw event data into your own data warehouse via its warehouse destinations. Once that export is configured, point SegmentStream at the same warehouse and SegmentStream reads the events from there.
Amplitude supports BigQuery, Snowflake (and Data Share), Redshift, Amazon S3, GCS, and Azure Blob Storage. SegmentStream connects to whichever one you use.
SegmentStream does not use Amplitude's reporting / dashboards API. Those expose only processed data with Amplitude's own session and identity model. SegmentStream consumes the raw exported events and builds attribution using its own identity graph and algorithms.
On initial setup, Amplitude can export historical events alongside the forward feed. SegmentStream becomes able to use that history as soon as the warehouse is reachable. For exact retention and historical-coverage limits, see Amplitude's Destination Catalog documentation.