What is pseudonymisation?
The GDPR defines pseudonymisation as processing in which data can no longer be attributed to a specific person without additional information – provided that this information is kept separately and protected (Art. 4 no. 5 GDPR). A typical approach is to replace names and identifiers with substitute values and keep the mapping separately.
The advantage: you can keep working with the pseudonymised data, and the result can be attributed to the right person again when needed.
What is anonymisation?
Anonymisation removes the link to the person in such a way that the person can no longer be identified with reasonable effort (cf. Recital 26 GDPR). Common methods are removal, generalisation – such as an age group instead of a date of birth – and aggregation into groups.
The catch lies in the combination: date of birth, postcode and gender together can make a person identifiable again, even if the name is missing.
The difference at a glance
| Criterion | Pseudonymisation | Anonymisation |
|---|---|---|
| Link to the person | can be restored with the mapping | permanently removed |
| Legal classification | remains personal data for anyone who has the mapping | no longer personal data |
| Data protection law applies | yes – for the controller; for recipients without the mapping, depending on the circumstances | no |
| Typical use | Sharing with AI and service providers, analyses that link back | Statistics, research, retention without reference to individuals |
| Route back to the person | possible | not intended |
An example
The same record, three times (all details fictitious):
| Field | Original | Pseudonymised | Anonymised |
|---|---|---|---|
| Name | Anna Muster | Customer K-4711 | – |
| Date of birth | 12.03.1984 | 12.03.1984 | Age group 40–49 |
| Place of residence | Lucerne | Lucerne | Central Switzerland |
| Request | Query about the policy | Query about the policy | Query about the policy |
When to use which method?
Pseudonymisation is the right choice when the result must find its way back to the person: for an AI response on a case, for processing by a service provider or for analyses with follow-up questions. Anonymisation is the right choice when the link to the person is no longer needed, for example for statistics or after a retention period has expired.
In Switzerland – and since a 2025 ruling by the CJEU also in the EU – a relative approach applies: what matters is whether the respective recipient can identify the person with reasonable effort. For the organisation that keeps the mapping, pseudonymised data always remains personal data.