dbt vs Airflow

Different stages.

Overview

dbt and Airflow are not competing tools; they cover different data-pipeline stages. dbt is in-warehouse SQL transformation (transforms data already in the warehouse); Airflow is workflow orchestration (moves data between systems and triggers transformations). Most modern data stacks use both: Airflow orchestrates extract-and-load, dbt handles transformation, Airflow triggers dbt as a step.

• Different pipeline stages. dbt transforms inside the warehouse; Airflow orchestrates across systems. Not interchangeable.
• dbt: SQL transformation. Templated SQL with tests, documentation, dependency graph. Default for in-warehouse transformation.
• Airflow: orchestration. DAG-based scheduling, broad operator ecosystem, cross-system data movement. Default for extract-and-load and pipeline orchestration.
• Both tools where appropriate. Airflow orchestrates extract-and-load, dbt handles transformation, Airflow triggers dbt. Standard modern pattern.

The approach

Workload-driven choice per stage, both tools where the pipeline spans both layers, documented rationale per pipeline. The discipline is recognising that the question is about pipeline stage rather than tool preference.

• Workload-driven per stage. Right tool for the stage. Reality drives the answer.
• dbt for SQL transformation. In-warehouse transformation. Default for analytics-engineering layer.
• Airflow for orchestration. Cross-system pipeline orchestration. Default for the orchestration layer.
• Operational fit plus documented rationale. Team workflow considered; per-pipeline rationale captured. Future investigations have the breadcrumb.

Why this compounds

The right tool per stage compounds across years. Cross-pipeline tooling (CI integration, lineage, alerting) gets built once per layer and reused. By year two the team's data-engineering muscle is mature and stage-appropriate tools are automatic.

• Better operational fit. Tool matches stage. Velocity stays high.
• Workload-driven decisions. Replaces tribal preference with documented rationale. Quality of choice improves.
• Better engineering velocity. Right tool means pipelines compose cleanly. Iteration speed increases.
• Year-one investment, year-two habit. First decision sets the patterns; subsequent pipelines inherit the model.