Business intelligence & analytics

Not by default. Sigma runs queries on your warehouse and returns results to the workbook. Your warehouse stays the system of record.

You can use caching or materialized tables for speed, but that’s optional and configurable.
If you choose to use caching or pre-built tables for speed, that’s an optimization, not a required data copy workflow.

Either works. And most teams use a mix.

If you already standardize logic in the warehouse (SQL/dbt/semantic layer), Sigma can sit cleanly on top. If you prefer to define metrics, calculations, and governed datasets in Sigma, you can do that too. The right answer is the one your teams can maintain and trust.

Sigma can respect your warehouse’s access policies by querying as the user (OAuth) or through a service account model, depending on how you deploy it.

On top of that, Sigma gives you its own controls for governing who can view, build, edit, and share content. This ensures access and authorship don’t get tangled.

Treat metrics like products: define them once, publish them in a governed place, and reuse them everywhere.

In Sigma that usually means shared datasets/data models (and certified content), so teams aren’t rebuilding “revenue” or “active customer” five different ways.

Yes. Sigma is designed to work on live warehouse data, so scale comes from your warehouse compute and Sigma’s execution approach.

In practice, teams use a combination of smart query patterns, caching where it helps, and model/dataset design to keep things fast as usage grows.