Minimal data / architecture
A smaller data footprint is easier to protect and recover
- Default
- Collect only what the task needs
- Fixtures
- Fictional, not live identities
- Decision
- Client and qualified reviewers
Separate identity from utility
Many workflows can operate with less identity than initially requested. Ask whether full names, government identifiers, exact location, financial details, political information, health data, biometrics, or documents are truly necessary. Use role or reference identifiers when they satisfy the task and the responsible owner approves them.
Know where every copy lives
Map the browser, application, database, logs, analytics, email, help desk, backups, exports, AI tools, and developer access. Document geography, account owner, encryption, privileges, retention, deletion behavior, and provider restrictions. A system diagram should expose shadow copies rather than treating the main database as the whole flow.
Design recovery without overexposure
Backups and exports support continuity but can also multiply sensitive data. Encrypt and limit them, verify restore, retain only as approved, and ensure an authorized person can recover the system without relying on one developer or inaccessible account.
Reduce the data map before choosing a platform
List the user task, proposed fields, systems, vendors, access roles, countries, retention needs, and sensitive categories that require review.
Prepare the project brief