The short answer: claim what you are owed, check that nothing important was lost, and then decide which parts of your business must keep running the next time Microsoft 365, or any provider, stops. From the afternoon of 12 November 2026 (UTC), users could not reach their mailboxes in Outlook, and the failure spread to Teams calendars, Copilot, Defender, Intune and many other Microsoft 365 services. Microsoft marked the incidents restored early on 14 November, after more than a day of disruption. Business customers on Microsoft’s service level agreement can claim a service credit until 31 December, but the credit only refunds part of the subscription for the service that failed. The larger cost, the orders, replies and meetings that did not happen, stays with you, and the only protection against it is a plan made before the next outage.
What happened between 12 and 14 November
Microsoft tracked the disruption as two incidents in the Microsoft 365 admin centre: EX1550543 for Exchange Online and MO1550821 for the wider set of services. The times below are taken from Microsoft’s status updates as reported by the trade press, in UTC.
| When (UTC) | What happened |
|---|---|
| 12 November, 16:17 | Microsoft opens EX1550543: users cannot connect to their mailboxes in Outlook on the web, desktop or mobile |
| 12 November, 18:26 | Microsoft opens MO1550821 as more Microsoft 365 services fail |
| 12 November, evening | Changes are frozen, traffic is moved and servers are restarted, but restored servers fail again under live traffic |
| 13 November | Microsoft apologises, deploys a fix to an upstream component and brings services back in stages; email delivery times are extended to 48 hours so delayed messages are not returned |
| 14 November, about 04:17 | Both incidents are marked as restored |
The scale was large. Downdetector logged more than 130,000 reports on 12 November from users in more than 70 countries and territories, and Microsoft says more than 430 million people use its Microsoft 365 apps. Government email in Singapore went down, the Supreme Court of the Philippines saw some of its operations disrupted, including replies by email, Air India moved its customer email elsewhere to keep queries flowing, and one US company postponed its quarterly earnings call. Not everything stopped: Teams chats, calls and meetings largely kept working, files stored directly in SharePoint stayed reachable, and mail already cached in the desktop Outlook app could still be read.
On the cause, Microsoft’s explanation changed as it learned more. It first reported that a large number of servers had become unavailable without any rise in traffic, then that a process which sets up secure connections on those servers was being overwhelmed. Its final update traced the incident to a problem introduced during an earlier network capacity upgrade, which surfaced after a later network change. Microsoft said it was not a cyberattack and not related to AI or agents, and it has promised a post-incident report with the full root cause. On 15 November that report had not yet been published.
What the service credit pays, and what it does not
Microsoft’s Service Level Agreement for Online Services promises 99.9% monthly uptime for business plans of Exchange Online, SharePoint Online and Teams. When uptime falls short, the customer can claim a credit against that service’s fee for the month:
| Monthly uptime | Service credit |
|---|---|
| Below 99.9% | 25% |
| Below 99% | 50% |
| Below 95% | 100% |
Uptime is counted in user minutes, so a partial outage counts for less than its length suggests. November has 43,200 minutes. A company whose every user lost Exchange Online for eight hours would score about 98.9% and fall in the 50% band; one that lost it for the full 30 hours that admins have recorded for EX1550543 would score about 95.8%, still in the 50% band. The Teams definition of downtime is narrower and covers messaging and meetings, which mostly kept working, so a Teams claim may not qualify.
Four conditions limit what the credit is worth:
- It is not automatic. Customers who buy directly from Microsoft file a request through Help and support in the Microsoft 365 admin centre. Customers who buy through a Cloud Solution Provider ask their partner, who files one claim per tenant in Partner Center.
- It has a deadline. For a November incident the claim must reach Microsoft by 31 December 2026, and Microsoft says it typically takes up to 45 days to decide.
- It needs evidence. The incident IDs, the times and duration of impact, the number and location of affected users, any support tickets and the steps taken.
- It is the only remedy. The credit reduces a future bill for the affected service. It does not pay for lost sales, overtime or missed deadlines, and personal and family plans have no service level agreement at all.
Five things to do before 31 December
- Record the impact while it is fresh. Note which teams and how many users were affected, from when to when in UTC, what failed and what error messages appeared. Screenshots of the Service health page help.
- File the claim. Directly in the admin centre, or through your partner. Quote EX1550543, and MO1550821 if other services were affected.
- Find the mail that matters. Microsoft says delivery was delayed, not lost, and most sending servers keep retrying for days. Run a message trace for your largest customers, suppliers and payment providers, look for bounce notices, and check that orders, invoices and approval requests sent during the outage actually arrived.
- Answer the enquiries that waited. A customer who wrote on 12 November and heard nothing may already have gone elsewhere. Our article on how buyers choose before they get in touch explains why the speed of that reply matters.
- Read the post-incident report when it arrives. It will say what Microsoft is changing. Ask your IT team or partner whether anything in your own setup should change as well.
Warn staff, too, about emails that pretend to come from Microsoft and ask them to restore a mailbox or reset a password after the outage. Microsoft does not need anything from users to restore service.
Before the next outage: decide what has to keep working
Australia’s home affairs minister said Microsoft likened the event to the faulty CrowdStrike update of July 2024, and every large provider has had days like it. Leaving Microsoft 365 is rarely the answer, because any other suite carries the same risk. The useful question is narrower: which of your processes stop when email and calendars stop, and what does each one fall back to?
| Process | What it depends on | A fallback to set up now |
|---|---|---|
| Customer enquiries and support | A shared Outlook mailbox | WhatsApp Business, a phone line and a support channel that does not run through Exchange Online |
| Order, invoice and payment emails sent by your systems | Exchange Online or Microsoft Graph | A separate transactional email service for system mail, so customer notifications keep going out |
| Staff communication | Outlook and Teams | A phone tree or messaging group agreed in advance, with a named person who declares an outage |
| Approvals and scheduling | Outlook and Teams calendars | An agreed manual approval route for payments and urgent decisions |
| Telling customers what is happening | A status note on your website and templates ready to send on other channels |
Write these fallbacks into a short runbook and keep a copy outside Microsoft 365. Turn on email alerts from the Service health page so the right people hear from Microsoft directly, and add simple automated checks on your own critical flows, such as an order confirmation reaching a test inbox, so you learn about a failure before your customers tell you. Then rehearse it: a half-hour walk-through twice a year shows whether people know what to do and whether the fallbacks still work.
Where SoftwareBiz fits. Our architecture and performance audit maps the single points of failure in your systems, including every process and application that depends on Microsoft 365, and tests that the fallbacks behave as designed. Our managed cloud and DevOps service runs the monitoring and alerting, the incident runbooks, the backups and the post-incident reviews that prevent repeats, with response times agreed in a service level agreement. When email goes quiet, our customer support team keeps answering your customers by phone, live chat and WhatsApp, in your brand’s voice.
What to do this week
Start with the claim and the mail trace, because both have a deadline or a cost of waiting. Then take one hour with the people who run sales, support and finance and fill in the table above for your own business. Most of the fallbacks cost little to set up. What makes them expensive is setting them up during the next outage.



