Microsoft 365 Outage Today: What Happened and What QA Teams Should Test

Rows of servers in a data centre, illustrating the infrastructure behind cloud workplace tools such as Microsoft 365

Photo: The National Archives (UK) — “A view of the server room at The National Archives”, CC BY 3.0, via Wikimedia Commons.

When Outlook, Teams and Copilot go dark together, it is a live fire drill for every team that has built its work and its incident plan on a single cloud vendor.

A global Microsoft 365 outage began on Thursday evening, 12 November 2026, and was still not fully restored as of Friday morning, 13 November. Here is what is confirmed, what is still unknown, and a checklist for QA and engineering teams. New here? Start with Start Here. Related reading: third-party risk after the Amadeus incident, and how to check if a website is safe before you click a “restore your account” link.

Update (15 November 2026): Microsoft has given a preliminary root cause. In its final service-health update for incident MO1550821, posted at 04:17 UTC on 14 November (9:47 AM IST), it said the incident “was caused by a failure in the systems responsible for establishing connections to the Exchange Online service.” Traffic shifts during standard network maintenance and upgrades “exposed a latent bug,” connection requests queued, and affected servers exhausted their processing capacity. Microsoft said this was “a technical issue not related to cyber or AI / agents,” and that the incident “was introduced by an issue during an earlier network capacity upgrade that surfaced following a recent network change.” It dated the impact from 16:20 UTC on 12 November to 16:10 UTC on 13 November (9:50 PM IST Thursday to 9:40 PM IST Friday). The same record’s header lists a start of 10:05 UTC (3:35 PM IST); Microsoft has not explained the difference. A preliminary post-incident report was promised within two business days of that update, and a final report within five. As of 15 November those reports had not been published. We cover the updated cause, the wider pattern of recent outages, and a fuller QA list in Why the Internet Keeps Breaking: What the Microsoft 365 Outage Means for QA.

What we know so far

Downdetector first detected problems just after 16:15 GMT on Thursday, 12 November (9:45 PM IST), the BBC reported. Reports spread to 74 countries in about an hour. Outlook reports peaked at nearly 140,000 around 18:00 GMT (11:30 PM IST). The Guardian put the peak around 1pm ET at nearly 130,000 reports, covering Word, Excel, PowerPoint, Outlook, Teams, SharePoint, OneDrive and Copilot. Those are user reports, not an official count.

As of about 10:30 AM IST on Friday, 13 November (05:00 GMT), Microsoft’s public service health page still listed a Microsoft 365 service degradation. Admins can follow incident IDs EX1550543 and MO1550821, or use the public status page if the admin centre will not load.

Exchange Online was at the centre. Users may be unable to reach mailboxes in Outlook on the web or desktop, and email may be delayed. Microsoft has said queued email is expected to arrive, just late. Copilot may fail to load or to reach work data. Features that depend on Exchange — Teams calendars, some SharePoint features, Planner, To Do, and parts of Purview, Defender and the admin centre — may also fail. Microsoft has said core Teams messaging and online meetings remain largely functional, and that files stored directly in SharePoint Online remain accessible. OpenAI and Anthropic reported problems with Microsoft 365 connectors. Service NSW asked customers to delay in-person visits where possible, and Workhorse Group postponed an earnings release, citing the outage.

What caused it: confirmed versus still unknown

Microsoft has been clear on one point: this was not a cyber-attack. A spokesperson told the Guardian the company “had no indication of a cyber-attack.” Charles Lamanna, Microsoft’s executive vice-president of Copilot, wrote that there was “no indication this is the result of malicious or cyber activity.” Microsoft’s public Azure status also showed no Azure issues at the time of writing.

Microsoft’s account of the fault has evolved, and a formal post-incident review has not been published. It first pointed to “elevated load” across multiple regions, then said the load was not from a surge in user traffic but from servers “being unexpectedly unavailable.” Some servers were failing automated health checks. A later status update said a process “responsible for establishing secure connections becomes overwhelmed, causing the servers to stop processing requests.” A University of Queensland status page, relaying Microsoft’s briefing, said the issue was confined to the Microsoft 365 network gateway and that workloads had been failed over to other gateways.

Recovery was uneven. Microsoft paused deployments, moved traffic, tested an alternative health-check method, restarted servers, and limited traffic in some locations. It began validating a configuration change on a subset of servers, and started reverting a recent network configuration change that “correlates with the start of impact.” Servers returned to service “can become unresponsive again once they begin receiving traffic.” By early Friday, Microsoft said those changes were “not taking as expected.” It has not said whether the network configuration change caused the outage, contributed to it, or only coincided with it. Until a post-incident report lands, treat expert comparisons as hypotheses for your own tests, not as confirmed facts.

What QA and engineering teams should take from this

You cannot patch Microsoft’s gateway. You can test how your product behaves when a vendor of this size fails, and apply the same discipline to the changes you ship.

Config changes need the same care as code

The public updates keep returning to configuration, including a change that did not take “as expected.” Treat infrastructure and flag changes as production releases. Use staged rollouts and canaries, and practise rollback in staging, including the case where it only partly applies.

Single-vendor dependency is a testable risk

If email, chat, files, calendars, identity and admin tools all sit in Microsoft 365, one gateway problem becomes an organisation-wide halt. Map those dependencies, decide which workflows must survive a day without Outlook or Teams, and test the alternative in a drill.

Resilience, failover and chaos engineering

Microsoft failed work over to other gateways and still saw recovered servers wilt when load returned. In environments you own, schedule controlled tests: disable a dependency in staging, fail a region, or block an API. Rehearse your own degraded modes, such as a login that still works if a notification email never arrives, before a real outage forces the rehearsal on you.

Monitoring and status-page communication

If the official page is slow or itself affected, unofficial trackers fill the vacuum. Your status page should not share a fate with the system it describes. Write the customer message before you need it, and make sure staff know where to look.

Retries, backoff and the thundering herd

Restored servers became unresponsive again under traffic. When a service flickers back, millions of clients can retry at once and knock it down again. In your own apps, test exponential backoff with jitter, a cap on retries, and a circuit breaker. Load-test the “service is back” moment, not only the happy path.

Checklist: what to do during a SaaS outage — and what to test afterwards

Use the first list while the vendor is down, and the second in your next sprint, on systems you own or are authorised to test.

During the outage

  1. Confirm the incident on official channels. For this event: status.cloud.microsoft/m365, incidents EX1550543 and MO1550821, and @MSFT365Status.
  2. Open a fallback channel that does not depend on the same vendor: a phone tree, SMS, a separate chat tool, or a status page hosted elsewhere.
  3. Do not mass-resend email. Microsoft has said queued mail should still arrive, just delayed.
  4. Freeze non-essential changes that depend on the affected platform.
  5. Record impact as you go: which workflows failed, which clients and regions, what still worked.
  6. Tell staff and customers what is down, what still works, and where official updates will appear.
  7. Ban unapproved apps and remote-support tools offered “to keep work moving.” Use only pre-approved channels.

What to test afterwards, in your own environment

  1. Inventory every critical workflow that has a single SaaS vendor underneath it, and name a tested fallback for those that cannot wait.
  2. Put configuration and flag changes through the same staged rollout, canary and rollback drill you use for code.
  3. Run a degraded-mode test (email down, calendar down, chat up) and check that the product fails in a controlled way.
  4. Verify client retries use backoff and jitter, and that a recovering endpoint is not stampeded.
  5. Check that your status page still loads if the primary vendor does not.
  6. Add a yearly tabletop or chaos exercise for “the productivity suite is gone until tomorrow.”

Phishing alert: scammers exploit outages

Outages are a gift to phishing crews: people are anxious, official inboxes may be delayed, and “IT support” messages suddenly sound plausible. The BBC reported that banks and universities were already warning users. ING Australia told customers to be wary of unexpected messages, especially those with a link to download an app or join a call. Old Dominion University asked students to watch for phishing emails and texts claiming to offer IT support.

Treat any unexpected message that asks you to reset a Microsoft password, “verify your account to restore Outlook,” or download a remote-access tool as hostile until you have checked it. Do not use the link in the message. Type the official address you already trust, or use the company’s own app. Our guide on how to check if a website is safe walks through the two-minute checks. If you entered a password on a page you now distrust, change it on the real site and review recent sign-ins.

FAQ

Is the Microsoft 365 outage a cyber-attack?

Microsoft says no. As of Friday morning, 13 November 2026, spokespeople and Charles Lamanna said there is no indication of malicious or cyber activity.

Is the outage over?

Not fully, as of about 10:30 AM IST on Friday, 13 November 2026. Microsoft’s public status page still showed a service degradation. Check the official status page for the latest.

Will delayed emails still arrive? Should we resend everything?

Microsoft has said queued email is anticipated to be delivered, albeit delayed. Wait, and do not mass-resend. If a message is time-critical, use a fallback channel instead of retrying Outlook in a loop.

What should QA actually test? We cannot break Microsoft 365.

You test your own dependencies, clients and process: canary and rollback drills, degraded-mode behaviour, retry/backoff, and fallback communications. The Security Testing for QA Engineers guide is a wider starting point.

Short close

As of Friday morning, this was an infrastructure failure in Microsoft 365’s network path — not a cyber-attack — with recovery still incomplete and early fixes struggling under returning traffic. The final root cause is not yet on the record. Treat configuration like code, rehearse failover under load, and keep a comms path that does not die with the vendor. Run one degraded-mode test this sprint. Then subscribe so the next breakdown lands in your inbox: Start Here · follow.it/hackup-s.

Comments

Popular posts from this blog

How to Check if Your Email or Password Was Leaked (and What to Do Next)

Types of Hash & Encryption

Burp Suite Tutorial for Beginners: A QA Tester's First Web App Security Test