Platform / Anonymize
Anonymize without breaking the data model
Every sensitive field is transformed with the technique it needs, and the same source value receives the same substitute in every table. Applications and tests keep working.
Field-level rules
The right technique for every piece of data
There is no single policy for the whole dataset: each field is configured with its own method. Names and emails can be replaced with fake data while a phone number keeps its format.
- Fake data based on the field label
- Masking with the format preserved
- Random shuffling, replacement from a list, clearing or deletion
- JavaScript functions for custom logic
- Locale settings for consistent names and formats
Dictionary
The relationships between tables are preserved
The dictionary decides whether the same source value always receives the same substitute. With global scope, consistency is kept across every entity and field, and also between successive runs.
- Scope by label, by field name, by entity or global
- Foreign keys stay consistent after the transformation
- Same mapping between runs for comparable datasets
- Option to skip the dictionary when no correspondence should remain
Synthetic data
When no link to the original should remain
Synthesis generates new records from the schema, not from existing data. It is useful to scale volume for load testing or to work without any production source.
- Configurable volume, above or below the original
- Fake data by label, lists, sequences and ranges
- Custom JavaScript functions
- Can be combined with the anonymization rules
The four capabilities
Questions about anonymization
Does anonymization break the relationships between tables?
No, if the rule uses a dictionary. With a dictionary of global scope, the same source value always receives the same substitute across every entity and field, so the keys keep matching between tables.
Can I apply my own transformation logic?
Yes. Besides the predefined techniques you can use JavaScript functions for logic specific to your data model, such as national formats or dependencies between fields.
Is anonymization reversible?
It depends on the technique and the dictionary. Fake data, clearing or deleting a field without a persistent dictionary do not allow the original value to be reconstructed. Masking with format preserved keeps part of the value and offers a different level of protection, which should be assessed against the applicable regulatory requirement.
Does an anonymized dataset fall outside the GDPR?
Data anonymized irreversibly stops being personal data. Whether it is depends on the configuration: techniques with no recoverable mapping versus masking that keeps part of the value. The legal assessment is up to your legal team or your DPO.
What is the difference between anonymizing and synthesizing?
Anonymizing starts from real data and replaces its values. Synthesizing generates new records from the schema, without starting from previous real data.
Try it on your own schemas
We show you anonymization on an architecture like yours, with your relationships between tables intact.