"Privacy by design" appears in laws and policies everywhere, but it's often left as a principle with no instructions. For engineering teams, it comes down to a handful of patterns that, built in early, make products compliant almost by default. Here are ten.
1. Collect less at the edge
Every field in a form is a liability. Before adding one, ask what decision it supports. Use derived values where you can: an age band instead of a full date of birth, a city instead of an exact address. Data you never collect can't leak.
2. Separate identity from activity
Keep directly identifying data (name, email, phone) in one store and behavioural data (events, transactions) in another, linked by an internal ID. Analysts get the activity data with the internal ID only. This is pseudonymisation, and it limits both breach impact and casual access.
3. Tag data with its purpose
Add a purpose tag to datasets and fields, for example purpose: billing, purpose: marketing_consented. Pipelines can then enforce rules such as "marketing jobs can only read marketing-tagged data", and you can answer "what do we use this for?" from metadata instead of meetings.
4. Treat consent as data
Store consent as structured records: who, what purpose, when, which notice version, how given, and when withdrawn. Services check consent through an API at the time of use, not just at sign-up. When someone withdraws, downstream processing stops automatically.
5. Make retention a default, not a project
Set a time-to-live when data is created: database TTLs, object-storage lifecycle rules, log retention settings. Deletion that is designed in happens; deletion that is planned for later rarely does. See building a data retention schedule.
6. Log safely
Logs are a classic source of leaks: tokens, passwords, full card numbers and personal data written in plain text. Use structured logging with a deny-list or allow-list of fields, mask values at the logger, and give logs their own retention period. Keep the access logs the law requires (India's DPDP Rules expect them to be monitored) without filling them with the data they are meant to protect.
7. Least privilege for data, not just systems
Role-based access to a database is a start. Go further with column- and row-level security, so a support agent sees the customer they are helping, not the whole table. Log who accessed which records and review unusual access.
8. Encrypt with keys you control
Encrypt in transit and at rest as standard, and use field-level encryption for the most sensitive values (national ID numbers, health data), so a database dump alone isn't enough to read them. Manage keys in a KMS with rotation and access logging. See cloud encryption and key management.
9. Build rights requests into the platform
Access, correction and erasure requests should not need an engineer to query ten systems by hand. Maintain a data map of where each person's data lives and expose internal APIs to export and delete by user ID. Test erasure end to end, including caches, search indexes, analytics and vendors.
10. Use privacy-safe test data
Copying production data into test environments is one of the most common privacy failures. Use synthetic data, or masked extracts that preserve formats but not identities. Lock down any environment that does hold real data to production standards.
Make it stick
- Add a short privacy checklist to your design review template.
- Run a DPIA for high-risk features, early.
- Give teams reusable components: a consent service, a logging library with masking, and a deletion API.
- Celebrate removed fields. Less data is a feature.
Learn more
Our free Data Protection & Privacy Fundamentals course covers privacy by design, privacy-enhancing technologies and privacy operations, with a final exam and certificate.
Written and reviewed by the DigiTrust Asia editorial team. How we research, review and correct articles.
General information, not legal advice. Laws and official guidance change; check the current text before acting.
Learn privacy properly
Data Protection & Privacy Fundamentals is a free, law-neutral course with a final exam and certificate.
Create a free account