Warehouse

Amazon S3

Gzip CSV per table and day (CSV now; Parquet later), with a _SUCCESS marker.

Built, not yet tested liveBuilt. Not yet tested against a live Amazon S3 account.

Live: shipped in OrcLink's production release before this one. Built, not yet tested live: built and tested with sample data, not yet run against a live account of the vendor. Coming soon: listed, not available to connect yet.

How to connect

Data out > Warehouse > Amazon S3: name, bucket, region, an optional key prefix, an access key id and secret access key (stored encrypted, never shown again). Save (paused), send a test, then start it.

What OrcLink reads

Nothing from the destination.

What OrcLink writes back

The tables outcomes, touchpoints, credits, spend, calls: one per UTC day, the whole day replaced each time. The columns are the read API's fields (published at /docs/export-schema.json).
These tables are person-level (one row per outcome, touch, credit or call): never an email, phone, IP address, page path, hash or click id.
Once a day (07:30 UTC) for the day before, plus the failed days of the last 7 days, while the scheduled jobs export and jobs run (your OrcLink operator schedules them).
A _SUCCESS marker after each day.

Never stored

Your credentials in clear: stored encrypted (AES-256-GCM) and never shown again.

Known limits

Files are CSV (gzip); Parquet is not built.
Backfill: up to 90 days back, one day at a time.
HIPAA mode: it runs only when the workspace has accepted OrcLink's BAA and an owner has recorded a BAA with the vendor; campaign names only with "Include campaign names in data out".
It runs only while the person who set it up (or last started it) can still manage the workspace.
Plans: the warehouse export needs the Medium plan or above.