Normalizing data is about creating a single format and point of reference for all the data in the system. This data can then be visualized, analyzed, or extracted by different systems without affecting the source data. This allows the maintenance of a single version of the truth. Normalized and standardized data are not equivalent terms. Normalized data relies on the standardization of data, but standardization does not require the data to be normalized. The aim of normalization is to have the minimum redundancy (duplicates) possible. This aids integrity, although it can be at the cost of processing and memory during reading and updating.

There are a number of normal forms; there is not enough detail to go through all of them, although a good reference is included in the Further reading section of this chapter. In this section, we will look at the first three normal forms (after which data is generally considered normalized.