What is a Data Normalization?

In the database-design sense, normalization means organizing tables so each fact is stored in exactly one place - splitting a table that repeats a customer's address on every order row into separate customer and order tables, for instance - which prevents update anomalies where the same fact gets out of sync across duplicated copies. In the data-cleaning sense, it means standardizing formats: making sure "NY," "New York," and "ny" all resolve to one canonical value before they're compared or aggregated. Both senses share a purpose: making data comparable and reliable to query. Without normalization, a query counting customers in "New York" would silently miss every row recorded as "NY," undercounting the true total without raising any error. Normalization typically happens early in a pipeline, before deduplication or enrichment, because those later steps depend on values already being in a consistent, comparable form to work correctly.