top of page
davydov consulting logo

DAVYDOV CONSULTING BLOG

HOME  >  NEWS  >  POST

The Microsoft 365 Outage: How to Keep Your Business Running When the Cloud Goes Down

10 hours ago
5 min read

On Thursday 12 November, millions of people around the world opened Outlook and found nothing there. No new mail, no calendar, and in many cases no way to sign in at all. The Microsoft 365 outage that began that afternoon spread from the Americas to at least 74 countries and dragged on, in one form or another, for the better part of two days.

And it was not just email. Teams calls dropped, OneDrive and SharePoint files became unreachable, Copilot stopped answering, and the Intune admin dashboards that IT teams would normally use to investigate were down as well. In the UK, even local councils and police forces reported problems reaching their systems.

Microsoft has apologised and confirmed there was no cyberattack behind it. Still, if your whole working day runs on Microsoft's cloud, an apology does not bring back the lost hours. So let's look at what actually broke, and what you can do so the next outage hurts a lot less.

An office worker rubbing his face in frustration at a laptop as email and cloud services stop responding during the Microsoft 365 outage

What actually happened on 12 November

The first reports appeared around 12:30pm Eastern time on Thursday. Within an hour, Downdetector was showing roughly 120,000 simultaneous reports in the US alone, and the count for Outlook later peaked above 131,000. What started as a problem in the Americas spread across the globe during the evening as more and more of Microsoft's servers dropped out of service.

Microsoft's first explanation was vague: the company pointed to elevated load. Later it gave a more useful picture. Servers were failing their automated health checks and being pulled out of rotation, which pushed their traffic onto the servers that remained. Those then struggled under the extra load and started failing checks too. A classic cascade, in other words.

Recovery was slow and uneven. According to Doug Madory of the network analytics firm Kentik, traffic to Microsoft's services only began returning after roughly 18 hours of disruption. On Friday morning some regions saw a second surge of failures before things settled, and by Friday afternoon Microsoft was reporting encouraging progress, with email queues approaching normal levels. To avoid losing queued messages, the company extended Outlook's delivery timeout window to 48 hours, which is why many people saw two days of mail arrive in waves.


Why one technical fault reached 74 countries

Microsoft says the root cause was a fault introduced during an earlier network capacity upgrade, which only surfaced after a more recent network change. It has been explicit that the incident was not related to a cyberattack, and not related to AI agents either. A full post-incident report is still pending.

The uncomfortable detail is that the safety mechanism did the damage. Health checks exist to protect users by removing a sick server from service. But when the check itself starts misjudging healthy machines, automation removes good capacity at enormous speed, far faster than any human operator could react. The protection became the amplifier.

There is also the matter of concentration. The same shared infrastructure serves consumer and business accounts across the planet, so a fault in one layer can touch everyone at once. Worse, many of the tools administrators needed to respond lived inside the very platform that was failing. Some IT teams could not open Intune, and some users could not even reach their cloud-hosted Windows desktops. When the cloud is down, the controls for the cloud are down with it.


The third major cloud failure in just over a year

If this story feels familiar, it should. In October 2025, a DNS fault in Amazon's US-EAST-1 region took a large slice of the internet offline for most of a day. A few weeks later Cloudflare, which sits in front of a huge share of the world's websites, had its own global wobble. Microsoft itself suffered a multi-day incident in September that it blamed on a core authentication configuration problem. Now this.

Notice the pattern. None of these events were attacks. Every one of them came from an internal change or configuration fault that slipped past review and was then spread everywhere by the same automation that normally keeps these platforms fast and cheap.

The practical conclusion for a business owner is simple: a cloud outage is not an if, it is a when. A handful of providers now carry most of the business internet, and even the best of them will fail occasionally. Your plans should assume it.


Network cables connected to racks in a data centre, the kind of infrastructure that failed during the Microsoft 365 outage

What the outage actually cost businesses

The direct losses are easy to picture: sales enquiries sitting unanswered, invoices stuck in outboxes, support inboxes going dark, meetings missed because nobody could see their calendar. For a company with 20 people on Microsoft 365, a working day of disruption is roughly 160 paid hours spent improvising.

The second wave is less obvious. When delivery resumed, two days of delayed mail landed in bursts and out of order. Teams answered questions that customers had already resolved by phone, duplicate orders appeared, and threads made no sense. Several businesses told reporters the cleanup took longer than the outage itself.

And then there is trust. Your customers do not care whose server failed. They emailed you, and you did not reply. For firms in regulated sectors, or public bodies like the affected UK councils, silence of that kind carries real consequences.


Practical takeaways: an outage plan you can write in an afternoon

Start with communication. Agree on one fallback channel that does not share infrastructure with your main suite, whether that is a messenger on a different platform, a shared phone tree, or simply SMS. Export your key contact lists somewhere reachable without your mailbox, and decide in advance who posts the we-are-aware message to customers.

Next, remember that the cloud is not a backup. Microsoft stores your mail and files, but if you cannot reach them, you do not have them. An independent third-party backup of mailboxes and critical documents costs a few pounds per user per month, and a weekly local export of your most important files costs nothing but discipline.

Then write the one-page playbook. Who declares an incident, what gets switched to manual, and how. Can you take orders by phone and invoice later? Can your team keep working on local copies? If you build digital products, the same thinking applies to your stack: put your status page on different infrastructure than your product, design features to degrade gracefully, and do not let a single provider sit under both your app and your recovery path.

Finally, map your dependencies once a quarter. List every system your revenue passes through, from website hosting to checkout to email, and mark which ones share a provider. You will usually find one or two cheap changes that remove a surprising amount of risk.


A business team gathered around a laptop working on a continuity plan for the next cloud outage

Does this mean the cloud was a mistake?

No. For most small and mid-sized companies, self-hosted servers would fail more often than Microsoft does, and with nobody famous to blame. The cloud is still the right call on cost, security and convenience. The mistake is treating it as infallible and building your operations with no slack at all.

Think of resilience as a design requirement rather than an insurance product. The preparations above cost a few hours and a small subscription or two. One day of global downtime, as 12 November showed, costs considerably more.


Final notes

Microsoft's full post-incident report is still to come, and it will be worth reading, because it will say a lot about how the company plans to stop health-check cascades from happening again. We will cover the conclusions when they are published.

In the meantime, the most useful thing you can do is spend one afternoon this month on your own single points of failure, before the holiday trading season puts them to the test. The next big outage is already somewhere in someone's deployment pipeline. The only question is whether your business will shrug it off or lose a day to it.


 
 
 

Comments


​Thanks for reaching out. Some one will reach out to you shortly.

CONTACT US

bottom of page