The outage that took the status page down with it

What happened

On November 18, 2025, Cloudflare had a widespread outage. Cloudflare acknowledged the problem at 11:48 UTC. Its dashboard, API and core network services were partly down, and many sites that rely on it for content delivery, DNS or DDoS protection were disrupted. Reports named X, ChatGPT, Perplexity, Canva, Spotify, Discord and Shopify among those affected.

Partial recovery began about half an hour later, but the work continued well into the day. The reporting attributes it to an internal service degradation producing HTTP 500 errors, and doesn't give a deeper technical cause.

One detail stood out. At the peak, Cloudflare's own status page was inaccessible, so customers couldn't get real-time updates from the source. Its support portal also had a separate problem with a third-party provider.

This is an older event, and Cloudflare later published a postmortem with more detail. We're using it for the lesson, which hasn't aged.

So what for your business

Most companies depend on a few shared providers for DNS, content delivery or sign-in. When one has a bad day, many unrelated services have the same bad day at once.

You can't prevent that. You can decide how well you handle it. Customers forgive an outage that's explained quickly far more readily than one that's met with silence.

The failure mode here is circular: the thing that tells people about a problem depends on the thing that has the problem. If your status page, monitoring and alert delivery all sit behind the same provider as your product, then your first sign of trouble may be customers telling you.

What to do this month

1. Host your status page elsewhere. On a different provider, with a different DNS path, from your product.
2. Monitor from outside. Use an external checker that doesn't share your provider's network.
3. Know your dependencies. Make a list of what relies on each shared provider: DNS, CDN, sign-in, email.
4. Write a fallback for the worst one. Even a manual runbook, like how you'd change DNS to reach your origin directly, is better than improvising.
5. Prepare the message. A short template for "we're aware and investigating", and a named person who sends it.