dbt turns a SQL SELECT into a named dataset in your warehouse.
In an ELT workflow, extraction and loading happen before that transformation. You start with raw records, write the business rule in a model file, then ask dbt to build the relation. A model is usually a SELECT, without CREATE TABLE wrapped around it.
Your first job is to make the output explicit. Choose the columns another analyst needs and preserve missing values unless the request tells you how to replace them. A NULL amount means unknown; replacing it with zero changes the meaning.
The exercises use Harbor Cart orders and Pulse Path subscriptions, two fictional businesses. Run the model, inspect the result, and check both the column names and the rows.