Oh Dear’s analysis of 1.8 million outages across 38,544 sites finds downtime near-universal, usually brief — and most often striking outside work hours.
BERLAAR, Belgium — Website monitoring service Oh Dear today published The State of Website Uptime, a data report drawn from 1,817,403 confirmed outages across 38,544 monitored sites between March and December 2025. The full report is free to cite with attribution at https://ohdear.app/uptime-statistics.
Key findings:
- Downtime is near-universal but usually brief. 61.4% of monitored sites went down at least once, and the average site logged 47 outages — yet half of all outages were resolved in under two minutes, and 98.6% were back within the hour.
- The damage lives in the long tail. The slowest 0.3% of outages dragged on past six hours, pulling the mean time-to-resolve (21.9 minutes) more than ten times above the median (1.9 minutes).
- Outages don’t keep office hours. More than two-thirds (68%) began outside the 9-to-6 workday, on weeknights and weekends — roughly 32 a year for the average site.
- “Three nines” isn’t what it sounds. A 99.9% uptime target still permits nearly nine hours of downtime a year.
“The number that surprised us wasn’t how often sites go down — it’s the gap between the typical outage and the worst one. Half are over in under two minutes, but a small tail runs for hours, and that tail is where the real cost sits. Most teams build their alerting around the average outage. The data says build it around the rare bad one — and assume it starts at 2am, because two out of three of them do.”
— Mattias Geniar, founder, Oh Dear
Methodology: All figures are aggregate, anonymized telemetry from Oh Dear’s own monitoring, March–December 2025: 1,817,403 confirmed outages across 38,544 sites. An outage is measured from the first failed uptime check to the first subsequent success; sub-minute blips are excluded. No figure identifies a customer, domain, or site.
About Oh Dear: Oh Dear monitors websites for uptime, performance, broken links, mixed content, scheduled-task (cron) failures and certificate expiry, alerting teams the moment something breaks. Learn more at https://ohdear.app.
Press contact: Deian Isac · deian@ohdear.app
# # #
Additional figures for journalists
All from The State of Website Uptime, Oh Dear telemetry, March–December 2025. Free to cite with attribution.
Outage resolution time
- Resolved under 1 hour: 98.6%
- Between 1 and 6 hours: 1%
- Over 6 hours: 0.3%
- Median: 1.9 min · Mean: 21.9 min
Reliability baseline
- Average uptime across all monitored sites: 98.71%
- Sites with at least one outage: 61.4%
- Average outages per site over the period: 47
Response speed (full request: DNS + connect + TLS + server + download)
- Median (p50): 159 ms
- 90th percentile: 801 ms
- 95th percentile: 1,244 ms
Beyond uptime (share of sites affected)
- Scheduled tasks (cron) that failed or missed a run in 30 days: 23.3%
- Broken links on the average site: 2.9%
- HTTPS sites serving at least one insecure (mixed-content) asset: 2.9%
- Domains that came within 10 days of expiring: 0.7%
Average-site quality scores (Lighthouse, last 6 months)
- Performance: 78/100 · Accessibility: 89/100 · SEO: 91/100 · First paint (median): 836 ms
What each uptime target actually allows
| Uptime target | Downtime per year | Downtime per month |
| 99% | 3.65 days | 7.2 hours |
| 99.9% | 8.8 hours | 43.8 minutes |
| 99.95% | 4.4 hours | 21.9 minutes |
| 99.99% | 52.6 minutes | 4.4 minutes |
| 99.999% | 5.3 minutes | 26 seconds |
