dbt

Definition

dbt (data build tool) is an open-source, SQL-based transformation framework that lets analytics engineers build, test, and document data transformation pipelines directly inside the data warehouse.

Overview

dbt handles the "T" in ELT: once raw data has already been loaded into a warehouse, dbt transforms it into clean, analytics-ready tables using modular, version-controlled SQL files called models, rather than requiring a separate ETL tool to do the transformation work. Models use Jinja templating for reusable macros and are compiled down to plain SQL that runs against the warehouse. dbt includes a built-in testing framework for checks like uniqueness, not-null, and referential integrity, and it automatically generates documentation and a dependency graph (DAG) showing how models relate to one another. Because models are just text files, they live naturally in Git alongside the rest of a team's code. dbt runs on top of warehouses like Snowflake, BigQuery, and Databricks, and is commonly orchestrated by tools such as Dagster or Apache Airflow. Teams typically pair dbt with a separate ingestion tool to load raw data before dbt transforms it. dbt Core is the free, open-source CLI, while dbt Cloud adds scheduling, CI/CD, and a hosted IDE — the full workflow is the focus of SkillVeris's dbt & Analytics Engineering course.

Key Features

• SQL-first transformation models with Jinja templating for reusable logic
• Built-in testing framework for data quality checks
• Automatic dependency graph (DAG) generation between models
• Auto-generated documentation and column-level data lineage
• Version-control friendly — models live as plain text files in Git
• Incremental models for efficiently processing only new or changed data
• Shareable macros and packages across projects
• dbt Cloud adds scheduling, CI/CD, and a hosted IDE on top of dbt Core

Use Cases

Frequently Asked Questions

Related in DevOps