What is a Data Mart?

Where a data warehouse holds the organization's full curated data set, a mart is a narrower slice - often just the tables, aggregations, and metrics a specific team actually queries - built either by pulling from the warehouse (a dependent mart) or, less ideally, built independently from source systems (an independent mart, which risks drifting out of sync with the warehouse's definitions). The main reason marts exist is performance and simplicity: a finance analyst querying revenue metrics shouldn't need to understand or join across the entire company's schema, and a smaller, purpose-built table set is both faster to query and easier to reason about than the full warehouse. The tradeoff is consistency risk - if each department's mart defines a metric like "active customer" slightly differently, the organization ends up with conflicting numbers for the same underlying concept, which is why dependent marts sourced from one warehouse with shared metric definitions are the safer pattern.