Why I Use dbt with BigQuery (And You Should Too)

Why I Use dbt with BigQuery (And You Should Too)

If you’re writing SQL transformations in BigQuery, whether through scheduled queries, stored procedures, or Airflow operators, dbt can make your life significantly better. Here’s why.

SQL Gets Version Control

dbt projects are just SQL files in a Git repository. Every model, test, and macro is tracked, reviewed, and deployed through standard CI/CD. Compare this to scheduled queries managed through the BigQuery console — no history, no review process, no rollback.

Built-In Testing and Documentation

dbt lets you declare tests inline: unique, not_null, accepted_values, and relationships tests catch data quality issues before they propagate downstream. Documentation is generated automatically from schema YAML files and rendered as a searchable website.

models:
  - name: dim_customers
    description: "One row per customer with latest attributes"
    columns:
      - name: customer_id
        tests:
          - unique
          - not_null
      - name: segment
        tests:
          - accepted_values:
              values: ['enterprise', 'mid_market', 'smb']

Incremental Models Save Money

Instead of rebuilding a table from scratch every run, incremental models process only new or changed records:

{{ config(
    materialized='incremental',
    partition_by={'field': 'event_date', 'data_type': 'date'},
    incremental_strategy='merge',
    unique_key='event_id'
) }}

SELECT *
FROM {{ source('raw', 'events') }}
{% if is_incremental() %}
WHERE event_date >= (SELECT MAX(event_date) FROM {{ this }})
{% endif %}

On BigQuery specifically, this means you scan less data and pay less.

The Dependency Graph

dbt builds a DAG of your models automatically via {{ ref('model_name') }}. This means transformations run in the correct order, and you can visualize your entire data lineage. When something breaks, you know exactly what’s upstream and downstream.

Takeaway: dbt brings software engineering practices — version control, testing, documentation, modularity — to SQL transformations. If BigQuery is your warehouse, dbt is the natural companion.