Data minimisation

Data minimisation is the GDPR principle that personal data must be adequate, relevant and limited to what is necessary for the purposes for which it is processed.

Part of the Learn hub and the customer data glossary.

What it means in practice

The principle is in GDPR Article 5(1)(c). Article 25(2) applies it by default: only personal data necessary for each specific purpose should be processed, which covers the amount collected, the extent of processing, the storage period and who can access it 1.

In a customer data platform the principle becomes design choices. Collect fewer events and properties. Reject unexpected fields at ingestion. Keep sensitive attributes out of marketing profiles. Use field-level allowlists for each destination. Set retention limits on raw events. Prefer derived or aggregated signals to raw detail. Give an AI agent only the fields its task needs.

US healthcare has a related but separate rule. HIPAA's minimum necessary standard asks covered entities and business associates to make reasonable efforts to limit protected health information to the minimum necessary for the intended purpose, with exceptions such as disclosures to a provider for treatment (45 CFR 164.502(b)) 2.

Hashing an identifier is not minimisation. A hashed email is still personal data; the control is to need fewer identifiers.

Example (health insurance): an appointment reminder needs a name, a contact channel and a time. It does not need diagnosis codes. Sending the diagnosis to an email or advertising destination would breach minimisation and could also be an impermissible disclosure of protected health information.

Related terms

• Purpose limitation: defines what is "necessary".
• Consent state: limits what may be used.
• Protected health information: the US health data it protects.
• Destination: where field allowlists apply.
• AI agent: context should be minimal.

Where it matters

• HIPAA marketing rules explained: authorisation, exceptions and pixels.
• Healthcare marketing pixels after HHS guidance: what to stop sending.

Back to the customer data glossary or the Learn hub.