The Reliability Leaderboard.

Data window

The last 122 days (~4 months) of incidents reported on each provider's official status page. Stawatch polls every tracked status page on a continuous schedule and stores start/end times locally. We re-aggregate every 10 days.

Uptime %

Counts only major and critical incidents — minor regional notices and informational entries providers leave open for weeks would otherwise inflate downtime in misleading ways.

uptime% = max(0, 1 − weighted_down / window_minutes) × 100

Per-incident cap

Each incident's contribution to weighted downtime is capped at 48 hours. Beyond that horizon the entry is almost always a long-tail monitoring state rather than active outage. Anything genuinely prolonged is still surfaced separately under Notable outages.

Severity weights

• critical × 1.0
• major × 1.0
• minor × 0.0
• none × 0.0

Total incidents

The incident count column reports every entry regardless of severity, so a provider with lots of minor regional issues is still visible in the data — even when their uptime % stays high.

Limitations

We can only see what providers publish. Internal incidents, silent regressions, and unreported degradations don't show up here. Provider-perceived severity is taken at face value.