Data Masking, Anonymization, Pseudonymization and Tokenization: The KVKK Difference
Anonymization is irreversible, pseudonymization is not. The difference between masking, tokenization and anonymization under KVKK, with a comparison table.
Quick answer: Anonymization transforms data irreversibly so it can no longer be linked to a person, and the result falls outside the scope of KVKK (Turkey's data protection law). Pseudonymization hides identity with a separate key, but because it is reversible the data is still personal data. Masking hides part of the data in the display. Tokenization replaces sensitive data with a token that carries no meaning of its own. These four serve different purposes and are not interchangeable.
Why are these concepts confused?
In practice these four terms are often used interchangeably. Yet the difference between them has both technical and legal consequences. The most critical distinction is reversibility. If data is truly anonymous it is no longer personal data and most KVKK obligations do not apply to it. But if it is only pseudonymized, it remains personal data because identity can be restored with the right key, and all protection obligations continue.
Saying "we anonymized our data" without knowing this difference is a common mistake. Often what was actually done is pseudonymization, and the organization believes it left the scope of KVKK when it did not.
Comparison of the four methods
| Method | What it does | Reversible | KVKK view | Typical use |
|---|---|---|---|---|
| Anonymization | Permanently destroys the identity link | No | Result is not personal data | Statistics, open data, analysis |
| Pseudonymization | Replaces identity with a separate key | Yes, with the key | Still personal data | Research, restricted access |
| Masking | Hides part of the data (e.g. card number) | Depends | Context dependent | Screen, report, support view |
| Tokenization | Replaces sensitive data with a meaningless token | Yes, with the token vault | Narrows scope | Payment, card storage (PCI DSS) |
When is each method used?
Anonymization is suitable when there is no further need to link the data to a person. It is aimed at publishing statistics, training a model or sharing open data. But true anonymity is hard. Inadequately anonymized data can make a person re-identifiable when combined with other sources. So any claim of anonymity should be tested against re-identification risk.
Pseudonymization is suitable when you want to keep data usable for analysis while separating identity. For example, in a study, the identity of patients is kept in a separate key. This is a strong protective measure, but because the data is still personal data, KVKK obligations continue.
Masking is used when data needs to be seen by an employee or support team but not shown in full. Showing only the last four digits of a credit card is the typical example.
Tokenization is used, especially in the payment world, to keep sensitive data away from systems. A card number is replaced with a token, and the real number is stored in a separate, protected token vault. This lightens the compliance burden by reducing the number of systems in PCI DSS scope.
A common mistake: real data in test environments
One of the riskiest habits of organizations is copying real personal data from the production environment into test and development environments. Test environments are usually less protected, open to more people and less monitored. Real customer data moved there becomes the silent source of a breach.
The correct approach is to use masked or synthetic data instead of real data in test and development environments. That way developers work with realistic data while personal data is not put at risk.
The KAOS and DSET approach
In KVKK compliance consulting, DSET determines which data an organization should protect with which method and tests anonymization claims against re-identification risk. Our KAOS security engine detects and reports real personal data leaked into test environments and unmasked sensitive fields during source code and data flow scans. Our goal is not compliance on paper but protection verified in the field.
Related reading: KVKK compliance consulting and VERBIS, data classification and labeling, how much is a KVKK breach fine, data breach 72 hour notification matrix, source code security review.
Frequently asked questions
Is anonymized data within the scope of KVKK? Data that is truly anonymized, meaning it can no longer be linked to a person in any way, is not considered personal data and most KVKK obligations do not apply to it. But if data is only pseudonymized, it remains personal data because identity can be restored with the right key, and protection obligations continue. Getting this distinction right is critical.
Are masking and tokenization the same? No. Masking hides part of the data in the display, for example showing only the last four digits of a card number. Tokenization replaces the whole of the sensitive data with a token that carries no meaning and keeps the real data in a separate protected vault. Tokenization is used especially in payment systems to narrow compliance scope.
Is using real customer data in a test environment a problem? Yes, it is a significant risk. Test and development environments are usually not as protected as production and are open to more people. Real personal data copied there becomes a silent source of breach. The right path is to use masked or synthetic data in test environments.
Sources
- KVKK, guidance and decisions on anonymization of personal data: kvkk.gov.tr
- PCI Security Standards Council, tokenization and payment data protection: pcisecuritystandards.org
- DSET Blog, KVKK compliance consulting and VERBIS
To protect your data with the right method and test your anonymization claims, you can talk to our team at Ankara Hacettepe Technopark. Contact DSET.
Kimliğinizi doğrulayın
Yetkilendirilmiş erişim alanı. Tüm giriş denemeleri kayıt altına alınır.