Why bother building a user hierarchy in a dimension if a user can manually arrange attributes in a hierarchy in the browsing tool?

A user hierarchy is more than simply an aid to the user. In combination with the attribute relationships for the attributes defining each level of the hierarchy, the user hierarchy is used for designing aggregations. You can think of aggregations as a collection of summary tables that are managed for you by Analysis Services. The addition of aggregations to a cube is one way to improve performance. You probably won't notice much if your fact table contains thousands of records, but once you cross into the "millions of records" zone, you'll notice a difference in query performance between a cube with aggregations and one without aggregations. 

Tip by Stacia Misner, a consultant and author specializing in business intelligence solutions since 1999, providing consulting and training services through Data Inspirations, and writing about BI at blog.datainspirations.com.