Since the evening of Thursday, 12 November 2026, several Microsoft 365 services have been disrupted worldwide. At the centre is Exchange Online: mailboxes cannot be reached and email arrives late. Because many other services build on Exchange, the disruption reaches into Teams, Copilot, Defender and Purview.

For many companies that means no Outlook, no calendar and no reply to the customer. This article covers what happened, what you can do straight away during an outage like this, and which fallback options exist for email. It ends with a contingency plan in six steps that small companies can manage too.

What happened on 12 and 13 November#

Microsoft tracks the incident under two IDs: EX1550543 for Exchange Online and MO1550821 for all services that depend on it. The timeline in Central European Time:

  • Thursday, 17:17. First report for Exchange Online. Users cannot open their mailboxes, first in North and South America, later in all regions.
  • Thursday, 19:26. Microsoft opens the broader incident MO1550821. Teams, SharePoint Online, Copilot and Defender are now affected as well.
  • Overnight. Microsoft pauses all pending changes, reroutes traffic and restarts servers. Several attempts do not have the expected effect. Servers that had returned to service fail again under load.
  • Friday morning. A patch for an authentication and routing component improves stability. Microsoft rolls it out region by region.
  • Friday, 10:03. Microsoft sees early signs of stabilisation. Full capacity has not yet returned.

On the cause there are only interim findings so far. According to Microsoft, a large number of servers became unexpectedly unavailable, which overloaded the remaining ones. On the affected servers, a process that sets up secure connections was overwhelmed. Traffic was not higher than usual. Microsoft sees no sign of an attack. It intends to name the actual cause in a post-incident report once the incident is closed.

What works and what does not, according to Microsoft:

AreaStatus according to Microsoft
Email (Exchange Online)Mailboxes are partly unreachable through Outlook on the web, Outlook desktop and mobile devices. Delivery is delayed
TeamsChats, calls, meetings and channel messages work. Calendar, scheduling and invitations are disrupted
SharePoint OnlineAccess to files works. Sharing with external users is partly disrupted
Security (Defender, Purview)Safe Links, quarantine, DLP and sensitivity labels work only to a limited extent
Copilot, To Do, PlannerDo not load, or load unreliably
AdministrationAdmin centre and Intune are partly limited

One detail matters for security. When Safe Links fails, rewritten links in emails can no longer be opened, and protection rules from Microsoft Purview apply only to a limited extent. An outage therefore also hits the protection you have put in place.

What you can do straight away during an outage#

  1. Confirm the outage. First check whether the problem is on Microsoft's side: at status.cloud.microsoft or in the Microsoft 365 admin centre under Health, Service health. The admin centre may be affected itself, the status page is independent of it.
  2. Do not rebuild anything on your side. Recreating Outlook profiles, reinstalling Office or resetting passwords does not help when the cause lies with Microsoft. Leave DNS, proxy and sign-in settings as they are. Changes like these often create the second problem that remains after the outage.
  3. Inform your staff. Briefly, and through a channel that works: what is down, what still works, how do we reach each other. In this incident, chats and calls in Teams kept working according to Microsoft.
  4. Do not send emails several times. Delayed mail usually arrives later. Sending everything twice causes confusion once the outage is over.
  5. Inform customers and partners. A notice on the website and a call to your most important contacts is usually enough. What matters is that you can still be reached by phone.
  6. Check deadlines. What has to go out today or arrive today? Quotes, deadlines, orders. For these you need a second route, more on that in the section on mail routes.
  7. Watch out for phishing. Attackers use well-known outages. Emails and calls asking you to “sign in again” or to “restore your mailbox” are almost always fraud.
  8. Keep notes. Start, end, affected users, impact. You need this for a service credit and for the review afterwards.

Is email lost during an outage?#

Usually not. Email was built for cases like this. If a server cannot deliver a message, it keeps the message and tries again later. The underlying standard (RFC 5321) recommends giving up only after four to five days.

In practice the periods are often shorter. By default, Exchange Online returns messages as undeliverable after 24 hours. In the current incident Microsoft has extended this window to 48 hours so that delayed mail does not expire.

For you this means:

  • The sender may receive a delay notification. The message is then still on its way.
  • Only when a non-delivery report arrives does the message have to be sent again.
  • After the outage, check everything you were expecting: orders, replies, deadlines. Ask actively for anything that is missing.

Alternative mail server: which fallback options exist for email#

The most common question on days like these: can we not simply put a second mail server next to it? You can, but not every variant helps. The options at a glance:

OptionHow it worksAssessment
Email continuity serviceA service in front of Microsoft 365 accepts all mail and passes it on. If Microsoft is down, staff keep working in an emergency webmailThe most complete solution. Costs per mailbox and needs a clean setup
Emergency mailboxes with a second providerA few mailboxes under a separate domain, independent of MicrosoftCheap and quick to set up. Covers only the essentials
Second MX record (backup MX)A second server accepts mail when the first one does not respondRarely helps during Microsoft outages and opens a door for spam
Own Exchange server (hybrid)A server on your premises runs alongside Microsoft 365Too much effort for most mid-sized companies, with its own maintenance and security workload
Independent backupMailboxes and files are backed up outside MicrosoftDoes not keep mail flowing, but gives access to existing content
Local copiesOutlook and OneDrive keep data on the deviceCosts nothing, it only has to be switched on

Email continuity service. Providers such as Hornetsecurity, Mimecast, Retarus or Barracuda place their service in front of Microsoft 365. The MX record of your domain points permanently to the provider. It filters the mail and delivers it to Exchange Online. If Exchange Online cannot be reached, it holds the mail and makes it available in a webmail. Staff read, reply and send there with their usual address. Once Microsoft is back, the mailboxes are synchronised.

There are three points you should know. Microsoft's spam protection has to be aligned with the upstream service, otherwise it assesses incoming mail incorrectly. Sending through the emergency service has to be authorised for your domain (SPF, DKIM), otherwise your mail ends up in the recipient's spam folder. And the sign-in to the emergency webmail must not depend on Microsoft. If sign-in through Entra ID ever fails, you would not get in there either.

Emergency mailboxes with a second provider. The pragmatic variant for smaller companies. You set up a few mailboxes under a second domain with an independent provider, for example for management, sales and support. During an outage you write from there, and customers learn the address through the website and by phone. Mail to your normal addresses keeps waiting at Microsoft. For urgent matters this is often enough. Important: set it up beforehand, test it beforehand, and do not store the credentials only in Microsoft 365.

Second MX record. It sounds obvious but rarely solves the problem. A backup MX is only used when the first server does not accept a connection. In many outages Microsoft keeps accepting mail, and only the mailboxes behind it cannot be reached. The second server then receives nothing. On top of that, spam is sent to backup servers on purpose, because there is often no filter in front of them. And everything that arrives there has to get back into the mailboxes later. We advise against it as the only measure.

Own Exchange server. Running email yourself trades one risk for others: hardware, updates, vulnerabilities, power supply, staff. The waves of attacks on on-premises Exchange servers since 2021 have shown how much work that is. For regulated companies with their own IT department, a hybrid setup can make sense. For most mid-sized companies it does not.

Independent backup. You need a backup outside Microsoft anyway. Microsoft is responsible for the platform. Protecting your data against accidental deletion or encryption is your own job. During an outage the backup has an added benefit: you can get to last week's quote or the contract in an attachment even when the mailbox does not load.

Local copies. Classic Outlook keeps a copy of the mailbox on the device in cached mode. Existing mail, calendars and contacts stay readable even when the server does not respond. The same goes for files that OneDrive syncs to the device. Check whether important folders are set to “Always keep on this device”.

More than email: the second channel#

Email is only one part. In this incident, chats and calls in Teams kept working. You cannot rely on that next time. A contingency plan therefore needs a channel that has nothing to do with Microsoft 365:

  • Phone list. Mobile numbers of staff and of your most important customers, suppliers and service providers. Also on paper or stored locally on the device.
  • Messenger group. A prepared group for emergencies in a service the company has approved.
  • Website. A prepared notice that someone can switch on in five minutes. For that, the website should not depend on Microsoft 365.
  • Phone system. If telephony runs through Teams, it depends on Microsoft too. Then you need call forwarding to mobile numbers that can be switched on without Teams.

And the plan itself? If it lives only in SharePoint, it may not be reachable when you need it. A copy belongs in a second place.

Cloud sovereignty: a fair debate, the wrong conclusion#

After every major outage the call for cloud sovereignty gets louder. The debate is justified. Who has legal access to our data? How dependent are we on a single provider? How would we get out again? Every company should answer these questions for itself.

An outage does not answer them, though. Disruptions are part of IT, with every provider:

  • Microsoft publishes the worldwide uptime of Microsoft 365 per quarter. In the first quarter of 2026 it was 99.526 percent. That works out at around ten hours of downtime in three months.
  • Amazon Web Services was partly down for around 15 hours in its US-East-1 region on 20 October 2025. The trigger was a fault in the name resolution of a database service.
  • Microsoft Azure was disrupted for a good eight hours on 29 October 2025 after a faulty configuration change. Microsoft 365 was affected as well.
  • Cloudflare had an internal configuration error on 18 November 2025. For hours, many large websites could not be reached.

A European provider or your own server has different risks, not none. If your own server fails, only your company is down, and nobody but you is working on the fix.

Something else matters more. Whoever is responsible for IT cannot hand that responsibility to a provider. What is outsourced is the operation. The question of how the company keeps working when the operation stalls stays in house. You have to be able to keep the impact small, and you need a plan B that works.

Business continuity management: a contingency plan in six steps#

Business continuity management (BCM) sounds like something for large corporations, but at its core it is simple: decide beforehand what happens in an outage. In November 2025, Germany's Federal Office for Information Security (BSI) published a dedicated brochure on this for small and medium-sized companies. For companies that fall under NIS2, the topic is mandatory. In Germany, section 30 of the BSI Act requires measures to maintain operations, including backup management, disaster recovery and crisis management.

A sensible start in six steps:

  1. Identify critical processes. What has to run for the company to work? Order intake, customer service, invoicing, deadlines.
  2. Write down dependencies. Which process depends on which service? This often shows that more depends on a single service than expected. Sign-in through Microsoft is one such case: if it fails, every application that signs in through it is affected too.
  3. Define tolerable downtime. How long can each process manage without the service? An hour, a day, a week? This decides how much preparation is worthwhile.
  4. Set up fallback routes. A plan B for each critical process: emergency mailbox, phone list, template for customer information, access to the most important documents without the cloud.
  5. Keep the emergency manual reachable. Short, up to date and in a place that does not depend on the disrupted service. It includes who decides in an emergency and who informs whom.
  6. Practise and review. Once a year, play through an hour without email. And after every real outage, record what went well and what was missing.

The effort is manageable. For a company with 20 to 50 workplaces, a first contingency plan fits on a few pages. It belongs in day-to-day IT operations and is updated with every change to the environment.

SLA and service credits: what Microsoft owes you#

Microsoft commits to a monthly uptime of 99.9 percent for Exchange Online, Teams and SharePoint Online. That allows around 43 minutes of downtime per month. If uptime falls below that, customers are entitled to a service credit:

Monthly uptimeService creditDowntime of more than
below 99.9%25%43 minutes
below 99%50%7.2 hours
below 95%100%36 hours

Four things matter here:

  • The credit does not come automatically. It has to be claimed, through a service request in the admin centre or through the partner you buy your licences from.
  • The deadline is the end of the month after the outage. For November 2026 that is 31 December 2026.
  • Only the share of the licence that belongs to the affected service is credited, offset against a later invoice.
  • Uptime is measured per customer, by affected users and minutes. So note the start, the end and the number of affected users.

This does not cover the damage. The credit replaces part of the licence costs. Lost revenue and lost working time stay with you. That is one more reason to have your own contingency plan.

Frequently asked questions#

Where can I see whether Microsoft 365 is having problems?
At status.cloud.microsoft, publicly and without signing in. Administrators find details in the Microsoft 365 admin centre under Health, Service health. Microsoft also reports larger incidents through the @MSFT365Status account on X.

Is email lost during an outage?
Usually not. The sending server retries delivery for hours or days. Only when the sender receives a non-delivery report does the message have to be sent again.

Can I point the MX record to another server during the outage?
Technically yes, but it rarely makes sense. Depending on your DNS settings, the change takes hours to reach everyone. The new server has to know all addresses and filter spam, and the mail received there has to be moved back into the mailboxes later. Without a prepared and tested target, it does more harm than good.

Does a backup help during an outage?
It does not keep mail flowing. If it sits outside Microsoft, though, you can get to existing mail and files while the service is disrupted.

Do I get money back for the outage?
If monthly uptime falls below 99.9 percent, you are entitled to a service credit on the licence costs of the affected service. It has to be claimed by the end of the following month.

Does a small company need a contingency plan?
Yes. Small companies in particular often depend on a handful of services and have no IT department of their own to step in. A plan of a few pages is enough to start with.

Conclusion#

The outage of November 2026 is not the first of its kind and will not be the last. It shows how much depends on a single service and how well a company is prepared.

Most of the preparation costs little. A phone list, local copies, an emergency mailbox and a short plan absorb most of a day like this. Anyone who depends more heavily on email adds a continuity service.

Whether IT operations are better kept in house or with a service provider is the subject of our article MSP or in-house IT admin.

Want to know how your company would get through a day like this? Talk to us. We go through which services your processes depend on and set up the right fallback routes.

Sources#

  • Microsoft, Service Health, incidents EX1550543 and MO1550821 (as of 13 November 2026, 10:03 CET)
  • Microsoft Learn, Service health and continuity, worldwide uptime per quarter
  • Microsoft, Service Level Agreements for Online Services
  • BleepingComputer, Massive Microsoft 365 outage disrupts Outlook, Teams, Copilot worldwide
  • IT-Administrator, Microsoft 365: Störung legt zahlreiche Dienste lahm (German)
  • RFC 5321, section 4.5.4.1, sending strategy and retries
  • Microsoft Learn, Enhanced Filtering for Connectors in Exchange Online
  • BSI, Business Continuity Management (BCM) für KMU (German)
  • Gesetze im Internet, section 30 BSIG, risk management measures (German)