dbt
Transforms raw warehouse data into numbers the business can trust.
dbt (data build tool) lets analysts write modular, version-controlled SQL that compiles into materialised views and tables inside the warehouse. It is the piece that turns a pile of raw event data into metrics the business can actually rely on.
Deliberate, not decorative.
Every warehouse project has a dbt repo. Raw tables land in a staging schema, dbt models clean, join, and aggregate them into mart tables with tests for uniqueness, freshness, and referential integrity. If a number on a dashboard is wrong, we can trace it back through the lineage.
- Single-source-of-truth metrics definitions for revenue, churn, and retention
- Automated testing that catches bad data before a dashboard breaks
- Documentation that lives next to the SQL, auto-generated for stakeholders
- Incremental models that process only new data each run
- Three different revenue numbers in three different dashboards
- Business logic copy-pasted across a dozen SQL queries
- Data bugs discovered by executives in a board meeting
Official resources for dbt.
The team behind dbt runs the docs, pricing, and sign-up. Start there when you are ready to try it.
Want help putting dbt to work?
We design, build, and maintain systems around dbt every week. If you have a problem it could solve, let’s talk about what to build.