Normalization is the process of organizing database tables to reduce data redundancy and prevent inconsistencies, by splitting data into related tables according to defined rules.
Without normalization, the same piece of information can end up duplicated across many rows — if a customer's address changes, you'd need to update it in every row that repeats it, and any row you miss creates inconsistent data.
Highly normalized databases reduce redundancy and keep data consistent, but often require more joins to reassemble related data for a query. Some systems intentionally "denormalize" parts of a database — reintroducing some redundancy — to improve read performance for specific, frequent queries.
Last reviewed: September 2026