Kahan Data Solutions

The 4 Common Types of Dimensions | Data Modeling 101

The video walks through four frequently encountered dimension types in star‑schema data modeling—role‑playing, conformed, junk, and degenerate—explaining when and why each should be used. Role‑playing dimensions reuse a single table for different roles within the same fact, such as a central dim_date linked to order, ship, and delivery dates, saving storage and ensuring consistent definitions. Conformed dimensions are shared across multiple fact tables, like a single dim_customer used in sales, tickets, and email facts, preventing divergent definitions and enabling cross‑process analysis. Junk dimensions bundle several low‑cardinality attributes (e.g., is_gift, payment_method, channel) into one table to keep the fact narrow, while degenerate dimensions place high‑cardinality identifiers (order number, invoice number) directly in the fact when no richer descriptive table is warranted. The presenter highlights practical examples: re‑aliasing dim_date three times, centralizing dim_customer for consistency, creating a composite key for the junk dimension, and inserting order numbers as degenerate attributes to avoid unnecessary tables. These patterns illustrate how thoughtful dimension design reduces redundancy, simplifies joins, and improves query performance. Understanding and applying these dimension types helps data teams build cleaner, more scalable warehouses, facilitates accurate reporting across business units, and minimizes storage overhead—critical factors for organizations seeking reliable analytics at scale.