Personal Data and Pseudonymization
According to the GDPR, pseudonymization is the process of replacing identifiers with tokens; however, such data remains personal if a person can be identified using additional information. It is important to understand that simply removing a name or hashing a column does not make a dataset anonymous. If you use a "simple" hash for a large number of identifiers, a dictionary attack can easily return the real values.
Risks of Re-identification
According to Recital 26 of the GDPR, data is considered anonymous only when it is impossible to identify a person using means that are reasonably likely to be used. The risk increases in small groups: if your report contains only one client with specific characteristics (e.g., age, occupation, and a specific region), their identification becomes possible even without a name. Never store a mapping table (where a token connects to a real ID) in the same environment as the analytical dataset. This table must have the highest level of protection and remain isolated.
Recommendations for DWH
To minimize risks, use aggregation: reports should operate on statistical groups, not individual records. Avoid "thin" groups where the sample size is too small to ensure confidentiality. If you need behavioral analytics, use random tokens generated through a secure procedure rather than simple hashing. Always assess residual risk: could the combination of your dataset with open data lead to de-anonymization? If so, the data is not anonymous, and all GDPR requirements apply, including the right to erasure and rectification of records. Adhering to these rules ensures that you remain within the legal framework, using the reference https://eur-lex.europa.eu/eli/reg/2016/679 .
Author