What dimensions do you typically create in a dimensional data model, and how do you decide which dimensions to include?

In a dimensional data warehouse, the number and type of dimensions are driven by business requirements and query patterns. Common dimensions include Time, Customer, Product, Geography, Channel, and Supplier. You typically create a dimension for each entity that is queried or filtered on in reports. The decision criteria are: 1) Cardinality – low‑cardinality attributes (e.g., gender, status) become dimensions; high‑cardinality attributes (e.g., transaction_id) stay in the fact. 2) Stability – attributes that change rarely (e.g., product category) are good candidates. 3) Query frequency – dimensions that appear in most queries should be modeled. 4) Granularity – you may create a hierarchy (e.g., day → month → year) within the Time dimension. A typical star schema might have 5–7 dimensions, but the exact number can vary from 3 for a simple sales fact to 10+ for a complex enterprise model. The key is to balance query performance, storage, and maintainability. You also consider whether to use a snowflake schema for highly normalized dimensions, but that adds join overhead.