In dbt, what is the difference between ref and source?

In dbt, ref() and source() are two functions used to reference tables, but they serve different purposes. ref() is used to reference another dbt model within the same project. It creates a dependency graph so that dbt knows the order of execution and can rebuild only the affected models when a change occurs. source() is used to reference external tables that are not managed by dbt, such as raw tables in a data warehouse. It tells dbt where to find the data but does not create a dependency; instead, it allows you to document and test those tables. In practice, you use ref() for internal transformations and source() for raw, external data. This distinction helps maintain clear lineage and ensures efficient incremental builds.