What is a Data Anonymization?

True anonymization is a higher bar than most people assume: removing a name isn't enough if a combination of remaining fields (zip code, birth date, and gender, famously) can still re-identify someone. Techniques include generalization (replacing exact values with ranges), suppression (dropping fields entirely), noise injection, and k-anonymity approaches that guarantee each record is indistinguishable from at least k others in the set. This is distinct from pseudonymization, where identifiers are replaced with tokens but a mapping back to the original still exists somewhere - pseudonymized data is still personal data under regulations like GDPR, while properly anonymized data falls outside its scope entirely because re-identification is no longer possible. Organizations reach for anonymization when they want to analyze trends, train models, or share data sets for research without carrying the compliance burden of handling identifiable personal data, but claiming something is anonymized when it's actually just pseudonymized is a common and risky mistake.