Website outages recover in two minutes through summer

A new analysis from Oh Dear found that website outage recovery times stayed steady through the summer holiday season. The median time to resolve outages held at about two minutes in each of the past 17 months.

The analysis covered 1.8 million website outages across tens of thousands of monitored sites, tracking monthly resolution times over a period that included two summers. August ranked among the fastest months in the dataset rather than the slowest, while summer as a whole recorded slightly better resolution times than the rest of the year.

The figures suggest continuity in website operations during a period often associated with reduced staffing and out-of-office absences. For smaller development teams and agencies in particular, website reliability can depend on a limited number of people rather than formal round-the-clock rotations.

Oh Dear said it expected to find signs that websites stayed offline longer in summer because the people responsible for fixing them were away. Instead, the month-by-month median recovery line remained broadly unchanged throughout the period examined.

"Europe goes on holiday. The websites don't, and neither does the recovery time," said Mattias Geniar, Co-Founder of Oh Dear. "Half of all outages are gone in under two minutes, every single month, right through August."

Some outages are resolved without direct human intervention because hosting platforms can restart processes or reroute traffic automatically. Others still require a person to act, such as reversing a configuration change, renewing a certificate, or stopping a failed deployment.

Median measure

A key part of the analysis is its use of the median rather than the average. Oh Dear said averages can shift sharply from month to month if a single long-running incident distorts the overall figure, while the median gives a clearer picture of a typical outage.

That distinction matters when assessing operational performance because outages vary widely in duration. A relatively small number of severe incidents can make one month appear materially worse even if most faults are fixed quickly.

The same dataset underpinned Oh Dear's State of Website Uptime report, which found that around seven in ten outages begin outside the 9-to-6 workday. That earlier finding suggested many incidents start when small teams have nobody formally assigned to monitor systems.

The seasonal analysis extends that argument by focusing on summer, when holiday schedules can add pressure to already thin staffing arrangements. Larger organisations may cover these periods through follow-the-sun support models or structured on-call rotations, but smaller operators often rely on a single person to keep services available.

Geniar framed the summer pattern as evidence that those responsible for uptime remain reachable even when they are away from the office. "The out-of-office replies pile up in summer. The downtime doesn't get any worse for it," he said.

Holiday resilience

The findings cut against a common assumption in IT operations that holiday periods increase the risk of prolonged service disruption. While websites continued to experience outages in July and August, the data suggests they returned to service at roughly the same speed as in other parts of the year.

For businesses that depend on web services, the result may offer reassurance about the resilience of digital operations during peak leave periods. It also highlights the extent to which automated recovery tools and informal human intervention now work side by side to keep websites online.

Oh Dear describes itself as a website monitoring service founded in Belgium by Mattias Geniar and Freek Van der Herten. It monitors uptime, performance, SSL certificates, DNS, scheduled jobs, and status pages for development teams and agencies using European infrastructure.

System Administrator Appreciation Day, marked on the last Friday of July, provided the backdrop for the analysis and its focus on the people who restore sites after failures. The data showed those efforts continue through the holiday season, with the midpoint outage still resolved in about two minutes.