SLA and OKStatus's own status page
The commitment
Section titled “The commitment”Paid plans (Pro, Max, Enterprise) come with a 99.9% monthly availability commitment — about 43 minutes per 30-day month. The Free plan carries no availability commitment.
The commitment covers three components, measured separately:
| Component | Scope |
|---|---|
| Alerting pipeline | Detecting a monitor state change and handing the notification to your channel’s provider |
| Check execution | Running your monitors at their configured interval from at least one selected region, and recording the result |
| Dashboard and API | Authenticated access to the dashboard at okstatus.eu, the public API at api.okstatus.eu, and serving your public status pages |
Measuring them separately means a dashboard outage does not trigger a credit for the alerting pipeline if alerts kept flowing, and the other way round.
The full contractual text lives in the SLA Annex, which forms part of the Terms of Service.
Service credits
Section titled “Service credits”Credits are reserved for the Max and Enterprise plans. On Pro, the 99.9% figure is a commitment of means without financial compensation.
| Monthly availability | Downtime (30-day month) | Credit |
|---|---|---|
| ≥ 99.9% | ≤ 43 min | none |
| < 99.9% and ≥ 99.5% | 43 min – 3h36 | 10% |
| < 99.5% and ≥ 99.0% | 3h36 – 7h12 | 25% |
| < 99.0% and ≥ 95.0% | 7h12 – 36h | 50% |
| < 95.0% | > 36h | 100% |
Credits are a percentage of the month’s excl.-tax fee, issued as a credit note against the next invoice. They are not automatic: email sla@okstatus.eu within 30 days of the end of the month concerned.
Where the numbers come from
Section titled “Where the numbers come from”This is the part most vendors are vague about, so to be explicit:
- status.okstatus.eu runs outside the infrastructure it reports on. It is a separate OKStatus instance hosted at a different provider, in a different country, behind a different network — so it stays up during an outage of the platform it measures.
- A component is only declared down after three consecutive failed checks 60 seconds apart, and only when a majority of probes agrees. That majority must include at least one probe hosted outside the measured infrastructure, so a single-probe network blip is not counted as downtime.
- 90 days of history are published there in the open — including the months where the commitment was missed. For a monitoring vendor, hiding a bad month would be the worst possible advertisement.
Exclusions in brief
Section titled “Exclusions in brief”Not counted as downtime: planned maintenance announced at least 48 hours ahead (capped at 4 hours a month, scheduled 22:00–06:00 CET), emergency security maintenance, force majeure and third-party network failures, anything caused by your own configuration or quota, account suspension, and beta features.
Article 5 of the SLA Annex is authoritative.
Related
Section titled “Related”- Status & uptime — how a monitor’s status and uptime percentage are computed for your monitors
- Maintenance windows — suppressing your own alerts during your own maintenance