dbt (data build tool)

dbt (data build tool) is an open-source framework from dbt Labs for the transformation step of an ELT workflow. It lets teams define transformations as SQL select statements organized into version-controlled models, which dbt compiles and runs against a warehouse, handling dependency ordering, testing and documentation along the way.

Unlike hand-written scripts or logic buried inside a BI tool, dbt applies software engineering practices, version control, code review, automated tests and documentation, to data modeling. It builds a dependency graph across models automatically, generates data lineage documentation from that graph, and supports incremental builds so only changed data is reprocessed. This is a specific product rather than a generic concept: "dbt" refers to this tool and its cloud offering, unlike broader terms such as a semantic layer.

dbt matters because it professionalized the transformation layer, giving rise to the analytics engineer role that sits between data engineering and analytics. It has become a common backbone of a data pipeline's transformation stage across many warehouses. Pitfalls include letting the model graph grow unchecked without clear naming or ownership conventions, skipping tests on business-critical models, and treating dbt as a substitute for upstream data quality rather than a place to enforce it.