Fact table

A fact table is the central table in a dimensional model, holding the numeric measurements of a business process — an order amount, a page view, a claim payout — one row per event or per aggregation grain. Each row carries foreign keys pointing out to the dimension tables that describe who, what, where and when the event happened.

Fact tables are typically narrow and very long: few columns, one of which is usually a measure like revenue or quantity, plus a handful of foreign keys, but potentially billions of rows. They sit at the center of a star schema or snowflake schema, radiating out to dimension tables, and grain — what one row represents — is the single most important design decision: getting it wrong forces every downstream query to guess at aggregation level.

Analysts and BI tools join fact tables to dimensions to slice measures by any descriptive attribute, which is why fact tables are the workhorse of a data mart or warehouse built for reporting. A common pitfall is mixing grains in one fact table (for example, both order-level and line-item-level rows), which silently double-counts totals; another is storing descriptive text directly in the fact table instead of pushing it out to a dimension, which bloats storage and breaks the dimensional model's reuse.