Glossary
Data Retention Policy
The document that says, for every kind of data you hold, how long you keep it and what happens when the clock runs out. Regulators set the floor, privacy law sets the ceiling, and your systems have to enforce both.
General definition
Data Retention Policy is the written policy that assigns a retention period and a disposal method to each class of data an organisation holds. In practice it is a table: data class, retention period, legal or business basis, deletion or archival method, system where it lives, and the owner who signs off. It is one of the first documents an auditor, a regulator or a GDPR data subject asks for, because it shows whether data is being kept deliberately or just accumulating.
In a chat or messaging product the classes usually include:
- Message bodies, attachments and the message archive that preserves them for compliance
- Audit, access and security logs, which have their own log retention rules
- Backups and snapshots, which copy everything above and must expire on their own schedule
- Account and profile data, analytics events, support tickets and billing records
The periods pull in two directions. Sector rules set minimums: HIPAA requires documentation of policies and required records to be kept for six years (45 CFR 164.316), SEC and FINRA rules keep broker-dealer communications for three to six years, and tax law keeps financial records for years after the transaction. Privacy law sets maximums: GDPR Article 5(1)(e) says personal data may be kept only as long as a purpose justifies it, and Article 17 lets a person request erasure. A legal hold overrides both by suspending deletion for data relevant to litigation or an investigation. A good policy records the exception path as carefully as the default.
Enforcement is where most policies fail. The periods have to become time-to-live settings in databases, lifecycle rules on object storage, index rollover in log platforms and scheduled jobs that actually run, and deletion must propagate into backups or the backups must expire on a schedule that respects the policy. Test it: pick a record, wait for its period to end, and confirm it is gone. Review the policy when you add a region, a product line or a new regulator, because retention and data residency are decided together.
In the Ethora ecosystem
Ethora keeps chat history in a server-side archive with configurable retention, so each deployment can set the period a regulator expects or a shorter window where data minimisation matters more. Per-user erasure handles GDPR requests, and Trust and Safety rules can remove accidentally shared sensitive data immediately or after a delay. The compliance audit trail is exported as immutable logs to S3 in your account, where your own lifecycle and Object Lock settings decide how long those records live.
On dedicated and self-hosted deployments the chat archive, the audit exports, the backups and the operational logs from Grafana and Prometheus all sit in your account, so one retention policy can govern the whole stack and your team enforces it with the tools it already uses. Retention is set per deployment through the admin panel and API, and the same choices carry into the healthcare, finance and insurance SDK deployments where the retention period is a contract term.