Analytics Engineering
Start free
GlossaryData modeling

Data mart

Also: mart

A curated, business-facing set of tables built for a specific team or subject area — finance, marketing, product — that analysts and BI tools query directly.

A data mart is the final, polished layer of the warehouse: the facts and dimensions a particular team actually uses, named in their language and shaped for their questions. In dbt these are your `marts/` models — the end of the pipeline that dashboards and analysts point at.

Marts are organized by domain rather than by source. A finance mart, a marketing mart, a product mart — each containing the handful of well-tested tables that team needs. This keeps consumers away from raw and intermediate models and gives every team a stable, governed contract.

The discipline of marts is restraint: expose what teams need, hide the plumbing, and make the grain and definitions of each table obvious.

Why it matters

Marts are the interface between your pipeline and the business. A clean mart layer is what lets analysts self-serve confidently without spelunking through raw tables or re-deriving metrics.

They also localize change: you can refactor everything upstream as long as the mart's contract stays stable.

Common mistakes
  • Letting analysts query staging/intermediate models directly, bypassing the governed mart layer.
  • Organizing marts by source system instead of by business domain.
FAQ
What's the difference between a data mart and a data warehouse?
The warehouse holds all the modeled data; a data mart is a curated, domain-specific subset of it (finance, marketing, etc.) shaped for a particular team's questions.
Go deeper
Related terms

Go deeper in the Data Modeling hub.

Learn this by building, not memorizing.

Definitions get you the vocabulary. The platform gets you the skill — graded exercises, real projects, and a portfolio capstone.

Try free practiceSee the course →