What is normalization from a modeling perspective?

< 1 min read

From a modeling perspective, normalization means splitting data into multiple related tables to remove redundancy.
For example, Customer address, city, and country were stored in separate lookup tables.
This reduces duplicate storage and keeps updates consistent.
If a city name changes, we update it in one place instead of many rows.
It improves data quality and maintains clear relationships between entities.
However, too much normalization makes reporting slower due to multiple joins.
In analytics projects, we usually normalize in source systems but not in reporting models.
So OLTP databases stay normalized while Power BI models are simplified.
This balances data integrity in source and performance in reporting.

Power BI Essentials