Support tiers & SLAs
Response times, uptime commitments, and what the SLA actually covers.
Response times #
| Severity | Definition | First response | Coverage |
|---|---|---|---|
| P1 — Critical | Production down or unusable for most users | 15 minutes | 24/7 |
| P2 — Major | Significant degradation, workaround exists | 1 hour | 24/7 |
| P3 — Minor | Limited impact, single function affected | 4 business hours | Business hours |
| P4 — Request | Questions, changes, improvements | 1 business day | Business hours |
Uptime commitment #
Managed systems carry a 99.95% monthly uptime SLA, measured by external synthetic checks — the same numbers you can see, not a private dashboard. Scheduled maintenance inside approved windows is excluded; everything else counts, including our mistakes.
If we miss the SLA, service credits apply automatically to the next invoice. You do not need to detect the miss or file a claim — the monthly report states it.
What the SLA covers #
The SLA covers systems under our management: monitoring, patching, backups, incident response, and the infrastructure itself. It does not cover third-party outages we cannot control (an upstream SaaS going down), though we still respond and mitigate — the distinction only affects credits, never effort.
How to raise an issue #
- P1/P2 (something is down or badly degraded): call the emergency line in your runbook, or use the /page command in the shared channel. Both page the on-call engineer directly — even if you are not sure it qualifies.
- P3/P4: the shared channel or the support email. Tickets are acknowledged within the SLA window with a named owner.
- Not sure of the severity? Pick the higher one. Down-grading a false alarm costs nothing; a P1 sitting in a ticket queue costs the SLA its meaning.
Escalation matrix #
| If | Escalate to | How |
|---|---|---|
| On-call has not acknowledged in 5 min (P1) | Secondary on-call | Automatic — no action needed from you |
| Response feels insufficient mid-incident | Incident commander | Say so in the incident channel; the IC role exists for this |
| Pattern of issues across weeks | Account principal | Email or the quarterly review — both work |
| Anything at all, relationship-level | Head of Delivery | Contact is in your runbook; using it is never an overreaction |
Maintained by the delivery team · updated quarterly