Data Minimisation: Stop Collecting Data You Cannot Defend
A field-level minimisation process for forms, KYC, analytics, support, HR and vendors.
Every extra field increases storage, security, rights-request and breach impact. ‘We may use it someday’ is not a defensible requirement.
The DPDP framework is phased. The 14 November 2025 commencement notification brought specified institutional and enabling provisions into force immediately; section 6(9), section 27(1)(d) and rule 4 follow after one year; most operating duties and rules follow eighteen months after Gazette publication. As of 22 June 2026, readiness should distinguish current law from future-state DPDP controls.
The Act ties consent to specified purpose, and the Rules require itemised data and specified purposes when operative.
Minimisation should assess collection, later use, sharing, inference and retention.
Mandatory and optional fields should be technically and visually distinct.
What the organisation should understand
- The DPDP framework is phased. The 14 November 2025 commencement notification brought specified institutional and enabling provisions into force immediately; section 6(9), section 27(1)(d) and rule 4 follow after one year; most operating duties and rules follow eighteen months after Gazette publication. As of 22 June 2026, readiness should distinguish current law from future-state DPDP controls.
- The Act ties consent to specified purpose, and the Rules require itemised data and specified purposes when operative.
- Minimisation should assess collection, later use, sharing, inference and retention.
- Mandatory and optional fields should be technically and visually distinct.
- Derived scores, profiles and predictions can create personal-data risk even if raw identifiers are hidden.
The five-point review
| Check | What to examine |
|---|---|
| Field | Exact value collected or inferred. |
| Purpose | Current feature or legal need. |
| Necessity | Why less data will not work. |
| Audience | Who receives it. |
| End | Retention and deletion trigger. |
Practical example
A newsletter asks for date of birth, city, employer and income though only email is needed. Removing the extra fields reduces compliance and fraud risk without reducing service quality.
How to apply the framework
Run a field-challenge workshop before launch.
Review APIs and SDKs because products often minimise visible forms while transmitting background data.
Operating workflow
Define the processing or incident precisely
Identify the people, data, system, purpose, owner, vendor and transaction or event. Review field, purpose and necessity together. Do not start from a policy template or software feature; start from what the business and system actually do.
Separate current duties from future-state DPDP readiness
Apply the 14 November 2025 commencement notification provision by provision. Continue complying with currently operative IT, CERT-In, telecom, banking, insurance, employment, consumer, contract and criminal-law requirements. Build the future DPDP process now, but do not describe a scheduled rule as already legally operative.
Preserve proof and improve the system
Keep the approved decision, notice or workflow version, access or event logs, vendor evidence, user communications and remediation record. Update product design, role access, retention, support scripts or incident playbooks so the same weakness does not recur.
Implementation checkpoint
Before closing the review, assign a named owner, a completion date and a live-system test that proves the control works. A policy statement is not enough when the product, vendor, support team, payment process or access configuration behaves differently. Preserve the test result, exception approval and remediation ticket so management can distinguish an operating control from an intention that has not yet been implemented.
Action checklist
- Create field inventory.
- Challenge necessity.
- Make optional fields optional.
- Limit derived profiles.
- Reduce vendor payloads.
- Delete unused historical fields.
Evidence to keep
- Field-purpose matrix
- Product decisions
- API samples
- Schema changes
- Deletion logs
Warning signs
- Future analytics as sole purpose
- Mandatory field with no dependency
- Full ID where last four digits suffice
- Vendor gets entire database
- Old fields remain after UI removal
Finin2min takeaway
Privacy governance is an operating system, not a policy PDF. The data map, purpose, access, vendor, retention, user workflow, incident response and evidence file must all tell the same story.