If the internet stopped working for you on Friday 24 July, it was not your wifi, your phone or your laptop. At around 11:50am UK time on Friday 24 July, websites across the world simply stopped loading - and the common thread was Cloudflare, one of the biggest pieces of plumbing on the internet.
What actually happened
Cloudflare is the network that sits in front of a huge share of the world's websites, shielding them from attacks and speeding them up. When it has a bad day, the sites behind it do not crash - they become unreachable. That is exactly what the failure looked like from the outside: browsers could reach Cloudflare's network, but pages never answered. An endless spinner, not an error message.
At 11:53am we checked Cloudflare's official status page, which was reporting a service outage with partial outages listed against every region it monitors at once: Africa, Asia, Europe, Latin America and the Caribbean, the Middle East, North America and Oceania. It had already been a rocky morning - the same status page records a separate network performance incident in Frankfurt between 9:45am and 10:00am UK time.
Why event websites vanished with it
This magazine covers what's on, and at lunchtime a good chunk of what's on went dark: ticket pages, box-office checkouts, event listings and venue websites all ride on the same infrastructure. In the interest of transparency, that included us - Live Laugh Local and our sister site Spaces Please were both unreachable during the outage, while our monitoring watched requests go into Cloudflare's network and never come back.
On the last Friday before a huge festival weekend - with WOMAD mid-way through its first day at Neston Park and the August bank holiday three weeks out - twenty minutes of dead ticket pages at lunchtime is not a small thing for organisers.
How it ended
Recovery came in waves. By 12:09pm sites began answering again, and by around 12:10pm everything we monitor was back and healthy. Total disruption: roughly twenty minutes. At the time of writing Cloudflare has not yet published a full incident report or a cause, and press coverage is only starting to appear - we will update this article as the picture firms up.
What to do next time the internet "breaks"
It will happen again - the modern web leans on a handful of giant providers, and when one stumbles, thousands of unrelated sites stumble with it. The practical checklist: if several unrelated sites die at once, it is almost never your connection - check a status page or an outage tracker before rebooting anything. If you are mid-checkout for event tickets, wait a few minutes and try again rather than re-paying. And if you run events, keep your on-the-day announcements somewhere that does not depend on your website - because your attendees may not be able to reach it at the exact moment they need it.
Update, Saturday 25 July: no incident report for that window
A day on, Cloudflare's published incident history for 24 July records several separate problems, including email delivery delays, R2 storage errors in the Asia-Pacific region and a spell of GraphQL API rate limit errors. None of them covers the late-morning window when sites became unreachable for us, and no dedicated report for it has appeared.
To be precise about what we can and cannot stand behind: we verified first-hand that multiple unrelated sites, on different hosts and different domains, stopped responding at the same moment, that connections were reaching Cloudflare's network and getting nothing back, and that Cloudflare's own status page was showing degradation across every region it monitors when we checked at 11:53am. We did not see, and still have not seen, an official account of what happened. That is not unusual for a short-lived edge problem, but it is worth saying plainly rather than implying a cause nobody has confirmed.
