Skip to content
Overunity Let's get together

November 18, 2026

Your Email Shouldn't Go Down When Microsoft Does

By Overunity

Last Thursday, November 12, a lot of offices opened Outlook and got nothing. Mailboxes would not load. Mail that did go out often did not arrive. Calendar invites stalled, and Copilot could not see anyone's email. For many businesses it stayed that way for most of a day.

Microsoft's preliminary post-incident report puts the Exchange Online incident at 4:20 p.m. UTC on November 12 to 4:10 p.m. UTC on November 13. That is just after 9 a.m. Alberta time on Thursday until just after 9 a.m. on Friday. The BBC reported users were affected in at least 70 countries. Ars Technica reported that governments, hospitals, courts, and banks were scrambling, and quoted the former head of the UK's National Cyber Security Centre calling it possibly one of the biggest outages of all time.

Overunity Mail customers were not affected. Their mail does not run on Microsoft 365. When a giant platform has a bad day, your quotes, bookings, and client replies do not have to go down with it.

By the numbers

  • 23 hours and 50 minutes: the length of the Exchange Online incident, per Microsoft's preliminary report.
  • More than 80%: the share of Microsoft's gateway servers that had stopped responding at the worst point.
  • 450 million: the number of paid Microsoft 365 commercial seats Microsoft reported in January, per Ars Technica.
  • 48 hours: how long Microsoft extended email timeouts so delayed mail would not bounce.
  • About 43 minutes: the downtime a month that a 99.9% uptime promise still allows.

What actually happened

It was not hackers. On its status page, Microsoft said it had "no evidence of malicious activity." The cause was a chain of ordinary things lining up badly.

According to Microsoft's preliminary report, a software fix rolled out in July carried a hidden bug that could make the gateway servers in front of Exchange Online lock up under heavy load. It ran for more than three months without trouble. Then, about 15 hours before the outage, Teams traffic shifted onto that same gateway as part of normal operations. At the same time, nearly 14% of the gateway's capacity was already out of service from an unrelated problem the day before.

At peak U.S. traffic, servers started failing their automated health checks and dropping out. Their traffic moved to the servers that were left, which then failed too. Microsoft says the failure spread across its fleet in less than two hours. Recovery was slow because servers brought back online were swamped by waiting traffic and locked up again. Teams, OneDrive, and SharePoint could shift to other infrastructure and came back sooner. Exchange Online, which is the email part, could not.

Microsoft has apologized and says a final review is coming. None of this means its engineers are careless. It means a shared platform is a shared fate.

It is not only Microsoft

The other giants have had their own bad days in the past year or so.

  • Amazon Web Services, October 19 to 20, 2025: Amazon's own summary describes about 14 and a half hours of disruption in its busiest region, US-EAST-1, after a DNS failure cascaded through the region. Amazon.com itself was hit.
  • Amazon Web Services again, November 15, 2026: two days ago, AWS status notices described sign-in, sign-up, and connection errors in US-EAST-1 from 3:05 a.m. to 6:21 a.m. Pacific, a bit over three hours.
  • Microsoft Azure, October 29, 2025: Microsoft's incident review says a bad configuration change in Azure Front Door caused roughly eight and a half hours of global trouble, which also knocked out Microsoft 365 sign-ins and admin portals.
  • Cloudflare, November 18, 2025: Cloudflare says an oversized configuration file broke much of its network. Core traffic failed for about three hours, and full recovery took close to six.
  • Cloudflare again, February 20, 2026: a faulty cleanup task pulled about 1,100 customer IP ranges off the internet, in an incident Cloudflare says lasted six hours and seven minutes.
  • Gmail, April 8, 2026: Google reported 8 hours and 10 minutes of delays or failures sending and receiving email, especially to outside domains.

Microsoft 365 and Google Workspace both promise 99.9% uptime each month. That allows about 43 minutes of downtime a month, or about 8.76 hours in a year. Last week's outage used up well over two years of that allowance in one go. A service credit on next month's bill does not send the quote you missed or call the patient back.

When the problem is a person, not an outage

Uptime is only half the risk. The other half is one tired click on a lookalike message.

The FBI's 2025 Internet Crime Report recorded $3.05 billion in losses from business email compromise. In Canada, the Canadian Anti-Fraud Centre reported $67.9 million in spear-phishing losses for 2025. Outages also give scammers an opening, because people are expecting strange messages about their email.

When a mailbox is compromised, the useful question is how fast you can cut it off. On Overunity Mail, that work is local. We can lock one mailbox so the rest of the office keeps working, disable a whole domain if the problem looks wider, or block a single IP address or an entire country you do not do business with. We also set up a recovery email and a trusted IP address for home or the office before you need them.

This is not a claim that Overunity Mail never has a problem. Disks fill up. Passwords get mistyped. The difference is scope, and whether someone you can actually reach can fix it.

Getting more out of email, without living in it

Trades, clinics, and small offices want more out of email without spending more time in it. This winter, Overunity is rolling out a Grok integration so you can connect an assistant to your mailbox for help with writing, organizing, replies, and calendars, in a conversation you stay in charge of. It is coming soon, not a switch you have to flip today. The more immediate question is whether your mail can survive a bad day on someone else's platform.

Shopping for business email?

If you are comparing providers after last week, this is a practical list.

Look for

  • Local support you can actually reach when something breaks
  • Mail on your own domain (you@yourbusiness.ca), not a shared consumer address
  • Admin controls that can lock a mailbox, disable a domain, and block an IP or a country
  • A recovery option and two-factor authentication
  • Backups, and an honest record of uptime
  • Mail hosted in Canada, if that matters for your records and privacy

Avoid

  • Putting email, files, chat, and sign-in all on one giant vendor with no plan B
  • Using a free consumer inbox as the company address
  • Shared passwords and no recovery plan
  • A provider who can only point you to a status page when you need a mailbox cut off now

Overunity Mail is built around that first list: Alberta-hosted email on your own domain, with a local team that can make those changes. If you are shopping, it is worth a spot on the shortlist.

Sources

  • Microsoft 365 preliminary post-incident report for EX1550543 and MO1550821 (November 15, 2026)
  • Ars Technica, "Microsoft's global outage may be worse than Amazon's or CrowdStrike's, expert says" (November 13, 2026)
  • BBC News live coverage of the Microsoft outage (November 13, 2026)
  • UC Today, "Microsoft Traces Exchange Online Outage to a July Bug" (November 2026)
  • AWS post-event summary for October 19 to 20, 2025, and AWS Health Dashboard notices for November 15, 2026
  • Microsoft Azure post-incident review for Azure Front Door, October 29, 2025
  • Cloudflare blog incident reports for November 18, 2025 and February 20, 2026
  • Google Workspace Status incident, April 8, 2026
  • Google Workspace and Microsoft Online Services service level agreements (99.9% monthly uptime)
  • FBI IC3 2025 Internet Crime Report and Canadian Anti-Fraud Centre 2025 figures

Let's build it properly.

One crew for the site, the mail, and the search. Call us or send a note — we reply within one business day.

+1 (403) 250-2444