Redesigned data source setup for connecting your warehouse
Connecting GrowthBook to your data now starts with a clear choice of setup path. Experiment assignment queries can be set up by mapping table columns instead of writing SQL that renames each one.
Three ways to bring in data
When your organization has no data sources yet, the Data Sources page opens with a side-by-side comparison of three options.
- Managed Warehouse gives you a warehouse that GrowthBook hosts, with built-in event tracking.
- Event Forwarder sends events to your own BigQuery, Snowflake, or Databricks warehouse without a pipeline to maintain.
- Custom connects a warehouse you already run. GrowthBook uses read access and queries it with SQL you define.
Managed Warehouse and Event Forwarder are available on GrowthBook Cloud. Self-hosted installations use the Custom option.
Fewer steps for a custom connection
To set up a custom connection, choose your warehouse and enter its connection details.
Assignment queries tell GrowthBook who saw which variation. Before, you wrote SQL that returned the exact column names GrowthBook expects. Now you can map columns from your table to their roles:
- the experiment ID, variation ID, and timestamp columns
- one column for each identifier type the query covers
GrowthBook checks that the three required columns and at least one identifier column are mapped before it saves the query. Columns whose names need quoting may still require writing the query in SQL.
Getting started
Go to Data Sources in GrowthBook, add a data source, and pick the option that fits your setup. For Custom, choose your warehouse and enter its connection details. Add an identifier type to the data source before you add an assignment query, because adding a query stays disabled until one exists. Use the warehouse connection guides to prepare credentials and access for your warehouse.
