Cloudflare’s Masterclass in Customer Support
Yesterday (November 18th, 2025) Cloudflare had the biggest system outage in over five years. Their entire service went down for multiple hours, leaving websites, apps, and other infrastructure that relied on them displaying cryptic 500 errors in an instant.
What followed was an absolute masterclass in customer communication and service.
First, their response was swift, even before they had understood the issue. Their status page—critically hosted off their own infrastructure and remaining operational (despite a brief overload when every sysadmin and developer worldwide hit it simultaneously)—immediately delivered the essential message: “We see it. It’s broken.”
Next, that status page was updated frequently, with no more than 20 minutes between updates, even if the update was simply, “We’re still working on it.” This is often overlooked: you can’t over-communicate during a crisis, as long as you stick to facts and stay calm.
Finally, the true masterclass began in this postmortem.
In those 600+ words they accomplished three things:
- Apologized and then demonstrated genuine remorse
- Provided a deep technical dive into what happened (and how they responded)
- Reinforced trust
Apologies work
Generally, support agents apologize too much. When one of my agents says, “I’m so sorry this has happened,” it’s the quickest way to earn a low score on Ownership (and likely Articulation) in my Customer Response Review.
Most of the time, agents apologize to seem empathetic. We do feel bad when someone’s upset! But most apologies are just words without substance unless you’re apologizing for something truly your fault.
In Cloudflare’s case, the outage was entirely their fault, and even the highest-paid engineers at dependent companies couldn’t have fixed it. An apology was necessary—and welcomed.
But it only landed because they clearly explained what went wrong and owned the mistake completely.
A Transparent Deep-Dive
The first apology (in the intro section, with another in the conclusion) would have felt hollow without the 200+ words detailing the issue.
In fact, when I revisited the post, I barely noticed the first apology. Like a recorded “your call is important to us” message, it could have been omitted. No one remembers it.
The second apology? Everyone believes it. They invested significant effort piecing together and publicly sharing the nitty-gritty details. The most curious engineers could dive into the wealth of technical information provided.
Your customers should leave every support interaction more educated about your product. Here, the postmortem shines. It reads like a choose-your-own-adventure novel, and no matter your technical level, you walk away knowing this team is competent.
Solidified Trust
Mistakes happen. How you respond defines trust. By the postmortem’s end, it’s clear that when Cloudflare says “we’re sorry,” they mean it.
To synthesize all that information into a coherent postmortem—on what was undoubtedly one of their most stressful workdays, and in mere hours after resolution—is remarkable.
I’ve never been prouder to host applications, proxy my website traffic, and defend my sites against bad actors using Cloudflare.
Sustainable Support works.
