Warehouse connection

Overview

Warehouse destinations land curated event tables alongside modeled traits for analytics engineering teams. Connect read-only service credentials, pick sync frequency, and map property packs to columns.

Steps

1. Open Sources → Warehouse and choose Snowflake, BigQuery, or Redshift.
2. Provide a least-privilege role that can write to a dedicated dataset and create incremental tables.
3. Select grain: user-day, event-level, or summary rollups for cost control.

-- Example staging table contract
CREATE TABLE tl_events_raw (
  event_id STRING,
  event_name STRING,
  received_at TIMESTAMP,
  payload VARIANT
);

4. Run a backfill window (7–30 days) before enabling incremental hourly loads.

Troubleshooting

• Auth expiry: rotate keys on schedule; TrackLayer surfaces JWT/OAuth expirations in the connection health card.
• Schema drift: version columns when teams add nested JSON; use variant columns during pilot phases.
• Cost spikes: narrow high-volume marketing pixels or sample debug streams before warehouse export.