dbt Minimal Project Structure

Simple dbt setup for small data teams. Flat model structure, basic tests, and minimal configuration overhead.

#dbt #sql #data #analytics #warehouse #elt

PNG PDF

Project Directory

dbt_project/

models/

SQL transformations

sources.yml

Source definitions

schema.yml

Model docs & tests

customers.sql

orders.sql

revenue.sql

seeds/

Static CSV data

country_codes.csv

tests/

Custom data tests

assert_positive_revenue.sql

macros/

Reusable SQL/Jinja

cents_to_dollars.sql

snapshots/

SCD Type 2

orders_snapshot.sql

dbt_project.yml

Project config

profiles.yml

Connection config

.gitignore

README.md

Why This Structure?

This flat structure works for small teams with <20 models. All models live in one folder with a single schema.yml for documentation and tests. Simple to navigate, minimal overhead. Upgrade to modular when you hit 20+ models.

Key Directories

models/-SQL files that become views or tables in your warehouse
models/sources.yml-Declare raw tables from your warehouse
models/schema.yml-Column docs, tests, and model descriptions
macros/-Reusable Jinja functions across models

Getting Started

pip install dbt-postgres (or dbt-snowflake, dbt-bigquery)
dbt init my_project && cd my_project
Configure profiles.yml with your warehouse credentials
dbt debug to test connection
dbt run to build models

Model Example

-- models/customers.sql
with source as (
    select * from {{ source('raw', 'customers') }}
),

renamed as (
    select
        id as customer_id,
        name as customer_name,
        created_at
    from source
)

select * from renamed

When To Use This

Small teams with <20 models
Single data source or simple pipelines
Getting started with dbt
Proof of concept or prototyping

When To Upgrade

More than 20 models becoming hard to navigate
Multiple developers stepping on each other
Need clear staging vs. marts separation
Complex dependencies across domains