Notesfoundations🌱 seedling
Data Retention
Also: Retention Policy · Storage Limitation · Retention Period
Data retention is deciding, per dataset, how long personal data is kept and enforcing deletion when that period ends. Data that has served its purpose isn't harmless; it is pure risk, and the retention limit caps what any breach can take.
Why it exists
Systems default to keeping everything forever. Old data rarely earns its keep but stays exposed to every future breach, subpoena and misuse. GDPR's storage limitation principle requires keeping personal data no longer than its purpose needs.
How it works
- A written retention period per dataset, tied to a purpose (e.g. "two years after the account closes", "90 days for request logs").
- A scheduled job that deletes or anonymizes expired data; a policy without a job is "forever".
- Storage-level TTLs where available (object lifecycle rules, database TTLs, log retention settings).
- Backups expire on their own schedule; restores must re-apply deletions made since the backup.
- Legal holds and statutory retention (tax, accounting) as explicit, documented exceptions.
Trade-offs
| Gains | Costs |
|---|---|
| smaller breach blast radius | historical analysis is lost |
| less to search and erase | building and monitoring deletion jobs |
| lower storage cost | conflicts with "keep it in case" instincts |
In practice
- Deutsche Wohnen: fined about €14.5M in Berlin (2019) because its archive could not delete tenants' old payslips, bank statements and tax data; in June 2026 a court cut the fine to €900,000 (not final), still finding intentional breaches of minimization and storage limitation.
- Blackbaud (FTC, 2024): kept data "far longer than was necessary", widening its 2020 breach; ordered to delete it and publish a retention schedule.
- Google documents deletion from active systems taking about two months (including up to a month of recovery window), with backups retaining data up to six months.
Pitfalls
- Retention set on the primary database but not on logs, analytics copies and backups.
- Soft deletes that hide rows but keep them forever.
Connections
- Part of
- Data Ethics
Recall
These cards are also in your review deck, scheduled with spaced repetition.