dbt (data build tool) is a tool for transforming and modelling data directly in a data warehouse or data platform, using mainly SQL. Instead of moving data to a separate processing environment, dbt lets you build a collection of reusable data models on top of your existing data. It also adds features such as dependency management, data testing, documentation and lineage. dbt has become increasingly popular among organizations doing modern data engineering, especially as cloud data platforms and the ELT approach have become more common.
I have been playing around with dbt recently and decided to write about how it works in practice with BigQuery. I wanted to do something simple but realistic, so I took some open data from Traficom about used cars imported into Finland.
The idea was not to build anything particularly fancy. I mainly wanted to see how dbt handles the kind of data transformation work that I have previously done with other tools, especially Databricks.
Getting the data into BigQuery
First I downloaded data about used cars imported into Finland during 2026 from the Traficom website.
Looking at the data, it was immediately obvious that I wanted to do some formatting, filtering and translations before the data would be in an optimal form for reporting.
