A retention rule cannot be enforced if the team does not know where the data lives. Start with an inventory that follows customer information beyond the primary database into files, exports, logs, derived outputs, backups, and external systems.
This is an engineering map for implementing the organisation's policy. It does not establish legal retention periods or replace advice about obligations in a particular jurisdiction.
Inventory data by purpose and lifecycle
For each category, record what it contains, why it is collected, who owns the decision, where it is stored, who can access it, and which event starts retention or deletion. Include whether copies or derived records exist elsewhere.
Distinguish an account closing, a study closing, a membership ending, and a deletion request. Those events may trigger different rules. Avoid a single vague “retain customer data” row that conceals several workflows.
Map a fictional voice-feedback study
Suppose a founder collects text and voice responses. The inventory includes study questions, participant answers, audio objects, private reviewer notes, exports, operational logs, billing references, and backups.
For each, ask whether it contains personal or confidential information and whether it must remain linked to a person. A generated export can preserve data after the source response is removed. A diagnostic log can accidentally retain answer text even when the application database follows the intended policy.
Choose a fictional response and trace its lifecycle from submission to review, export, closure, deletion, and recovery. Record which steps are automated, which require a person, and what evidence confirms completion. Use harmless fixtures for this exercise.
Make exceptions and dependencies visible
Backups may follow a different deletion and restoration policy from active records. External providers may have separate controls and retention behaviour. Document those boundaries with the responsible owner rather than implying the application can erase every copy instantly.
The PostgreSQL backup reference explains database backup approaches, but it does not define your retention policy. Verify the actual hosting and storage configuration as part of the inventory.
The tradeoff is operational usefulness versus retained exposure. Logs and exports help support and research, yet each becomes another store to secure and manage. Minimise unnecessary copies and keep enough context to operate the product responsibly under the chosen policy.
Review the inventory when adding a new provider, file type, AI-derived output, or support tool. A retention schedule that covers only the launch architecture will drift as the product grows.
The StackFlight product account supplies relevant study-workflow context. This inventory is illustrative and does not claim to describe its current retention implementation.
Questions about retention inventories
Should every data category have the same period?
Not automatically. Purpose, obligations, customer commitments, and operating needs may differ. Have the responsible policy owners decide the rules.
Does deleting the main record complete deletion?
No. Check files, derived outputs, exports, logs, external services, and backup policy. Track the result across the defined lifecycle.
What should the first inventory review produce?
Produce named owners, missing stores, unclear triggers, and testable cleanup tasks. Bring the data map to a SaaS lifecycle review.