dbt

Transform data with confidence

SQL-based data transformation tool with testing, documentation, and version-controlled semantic layers.

Our take

dbt is the transformation layer that defined the modern data stack. The model is simple and powerful: write SQL SELECT statements that define data models, dbt handles materialization (table, view, incremental), testing, documentation, and lineage automatically. The open-source dbt Core is free and handles most data engineering needs — teams self-host scheduling via Airflow or Prefect. dbt Cloud adds IDE, scheduling, CI/CD, and the Semantic Layer (business metric definitions shared across BI tools) for $100/seat/month. The Semantic Layer is the most strategically important feature: define 'Monthly Active Users' once in dbt, and every BI tool (Looker, Tableau, Metabase) uses the same definition — no more inconsistent metric calculations across dashboards. The learning curve is SQL fluency + Git workflow; teams without either face a steeper adoption path. Fivetran comparison: Fivetran moves raw data in, dbt transforms it — they're complementary, not competing.

Product roadmap

dbt Labs is expanding the Semantic Layer to become the universal metrics layer across the data ecosystem, building dbt Explorer for interactive lineage and documentation, and developing AI-powered SQL generation to lower the barrier for non-engineers. The MetricFlow acquisition enables metric federation across any BI tool.