Some modelers are attracted to abstract generic dimensions. For example, their schemas include a single generic location dimension rather than embedded geographic attributes in the store, warehouse, and customer dimensions. Similarly, their person dimension includes rows for employees, customers, and vendor contacts because they are all human beings, regardless that signiﬁcantly different attributes are collected for each type. Abstract generic dimensions should be avoided in dimensional models. The attribute sets associated with each type often differ. If the attributes are common, such as a geographic state, then they should be uniquely labeled to distinguish a store’s state from a customer’s. Finally, dumping all varieties of locations, people, or products into a single dimension invariably results in a larger dimension table. Data abstraction may be appropriate in the operational source system or ETL processing, but it negatively impacts query performance and legibility in the dimensional model.