Policy
Risk Management Policy
How risks to customer data are identified, rated, treated and reviewed.
- Version
- 1.0
- Last reviewed
- 23 October 2026
- Owner
- Security Lead, security@etropo.com
1. Purpose
Etropo manages security by deciding, on the record, which risks to customer data matter most and what is done about each. This policy sets the method.
2. When
- A full assessment at least once a year.
- A targeted assessment when a new vendor, integration, data class or architectural component is introduced, and after any P0 or P1 incident.
3. Method
- Identify the assets (customer budget data, personal data, credentials and keys, the production platforms, the source code) and the threats to each: unauthorised access, data loss, service loss, vendor failure, insider misuse, supply-chain compromise.
- Rate each risk for likelihood (rare, possible, likely) and impact (minor, serious, severe) and combine them into Low, Medium, High or Critical.
- Treat each risk: mitigate with a control, transfer (insurance, contract), accept with a reason, or avoid by not doing the thing. High and Critical risks are not accepted without the Security Lead's written reasoning.
- Assign an owner and, for mitigations, a date.
- Record the register in the repository and summarise it to the other founders.
4. The register
Each entry has: asset, threat, existing controls, rating, treatment, owner, date, status. The Controls page on this site reflects the treatments that are in place, partial or planned.
5. Standing high-priority risks
Named here so customers can see what the programme is working on:
- Single-person dependencies. A small team concentrates production access and knowledge. Treatment: two owners on every system, written procedures, this Trust Center as the record.
- Limited independent assurance. Treatment: annual penetration test (most recent October 2026, findings tracked to remediation) and a SOC 2 Type II audit.
- Database recovery point of 24 hours. Treatment: evaluate point-in-time recovery; restore from backup tested every year (2025 and 2026).
- Session replay without text masking in product analytics. Treatment: enable masking. Done 2026-10-22; text and inputs are masked.
- AI features that read customer data. Treatment: provider terms excluding training; same access rules as the user; data sent only on user action.
6. Review
Reviewed annually. Version 1.0, October 2026.