transformation

dbt

dbt is a transformation product that version-controls SQL models. This profile covers what it does, how it is priced, and where it stops being the right choice.

What the product actually does

dbt is a transformation framework. It does not move data — it assumes the data is already in your warehouse and provides a way to express the SQL that reshapes it. Models are SQL files with references between them, and dbt resolves the dependency graph, runs the models in order, and manages the resulting tables and views.

The feature that made it ubiquitous is the testing layer. You can assert that a column is unique, not null, or a member of a set, and those assertions run as part of the build. Before dbt, warehouse data quality was verified by whoever remembered to check; after dbt it is verified by the pipeline or the pipeline fails.

How pricing works once you are inside

dbt Core is open source and free. dbt Cloud prices per seat with a developer tier that is genuinely affordable and enterprise tiers that are not. The cloud product's value is the scheduler, the IDE and the documentation site — all of which you can approximate with open-source components at the cost of engineering time.

The cost decision is therefore straightforward: if you have platform engineers, Core plus an existing orchestrator is cheaper. If you do not, Cloud's per-seat price is usually less than the engineering time to replace it.

Where it breaks down

dbt is not an ingestion tool and does not want to be. Teams that try to make it pull data end up with brittle custom macros. It also assumes a warehouse with reasonable SQL support; running it against a transactional database produces slow, awkward models.

The other limit is that dbt's model has no built-in notion of incremental correctness for complex cases. Incremental models work well for append-only data and require careful hand-written logic for anything with updates or deletes, and getting that logic wrong produces silently incorrect tables.

What switching costs

Low technically, high culturally. The SQL is portable; the tests, the documentation and the team's habits are the actual investment.

Who should not use it

Teams without a warehouse. dbt is a layer on top of one, and there is nothing for it to do without that foundation.

Sources

dbt public documentation and pricing page, read October 2026.