Blog

When Microsoft 365 Stops Working: Business Continuity Lessons for Small Businesses

Nov 14, 2026

A frustrated office worker watches Outlook remain on its loading screen at his desk.
AI-generated illustration of a worker waiting for Outlook; not an actual incident screenshot.

An email outage quickly becomes a business problem when the next customer appointment, an urgent approval and the information needed to finish a job all sit behind the same service.

The Microsoft 365 outage of November 2026 illustrates that dependence. For an Australian business starting work on Friday, the practical question was how to keep serving customers while familiar tools were unreliable.

A disruption can reach beyond email into scheduling, document workflows, connected applications and processes that require a sign-in or an emailed verification code. Understanding those relationships is the first step towards a useful business continuity plan.

What happened during the November outage?

Microsoft tracked mailbox connectivity under EX1550543 and wider service effects under MO1550821. Its incident communications, reproduced by the University of Pennsylvania, record these milestones. AEDT is UTC plus 11 hours.

Milestone UTC AEDT
Exchange Online incident start 12 November, 16:17 13 November, 3:17 am
Wider Microsoft 365 incident start 12 November, 18:26 13 November, 5:26 am
Broad recovery; initial cause explanation 13 November, 17:57 14 November, 4:57 am
Sustained recovery; backlog monitoring 13 November, 22:00 14 November, 9:00 am

Incident snapshot — 9:00 am AEDT on 14 November 2026: Microsoft was monitoring recovery and queued email; the incidents remained open. This is the status at that verified update, rather than a live service-status report. The detailed post-incident report remained pending, so the initial cause explanation below should not be read as Microsoft's final analysis.

Mailbox access and delivery were affected, alongside Exchange-dependent Graph requests, Teams calendars, Copilot and some administration and compliance features. Microsoft said core Teams calls, meetings and messaging largely remained functional, as did direct SharePoint file access. This was not complete unavailability of every service worldwide. Microsoft incident updates via Penn.

Was this a network failure or a server failure?

Both descriptions can refer to part of the same incident. They describe different layers of infrastructure, so neither is a complete explanation by itself.

Servers process requests and run the software behind a service. A server can be powered on but unable to do useful work because a process is overloaded, blocked or responding too slowly.

Network infrastructure carries and directs traffic. Routing, load balancing, connection management and traffic distribution influence which systems receive requests and how connections reach them.

Application dependencies connect those systems to business functions. A calendar feature might need Exchange data, even though the user opens it inside another application.

What Microsoft reported and tried

Early reporting from Information & Data Manager described unavailable servers shifting load onto remaining infrastructure, rather than an overall increase in demand. It also recorded failed health checks, overwhelmed processing, a pause on deployments and early mitigation that had not delivered the expected relief.

The University System of Georgia's reproduction of Microsoft updates records an alternative health-check method returning some servers to service, plans for automated restarts, and traffic moving to alternative pathways. These were recovery actions; none independently established the initiating fault.

OneCall's provider timeline records the uneven recovery: some servers stopped responding again after traffic returned. It also describes a process establishing secure connections becoming overwhelmed and preventing servers from handling requests. Microsoft tested connection-handling changes, reversed a recent network change, added capacity and applied traffic controls. A patch to upstream authentication and routing components improved stability, followed by staged restoration. OneCall summarises the updates; it is not Microsoft's final technical report.

Microsoft subsequently attributed the incident to an issue introduced during an earlier network capacity upgrade that surfaced after a recent network change. It said the event was technical, unrelated to cyberattack or AI agents. That is an initial causal explanation, with the detailed post-incident review still outstanding. Microsoft's 13 November update via Penn.

A change to network infrastructure can trigger failures in server-side processes without the underlying fault being a conventional routing or connectivity failure. This is why describing the event simply as a network outage can obscure how the different layers interact.

Microsoft's later statement establishes a connection to network changes, but it still does not identify a particular routing fault, failed piece of hardware or complete sequence of software failures. Nor does adding capacity mean Microsoft simply ran out of servers.

How a service can become trapped in a recovery loop

Consider this explanatory scenario, rather than a confirmed reconstruction of every step in Microsoft's incident:

  1. A service component becomes overloaded and responds slowly.
  2. Automated health checks time out, so traffic management removes servers from service.
  3. The remaining servers receive more work, increasing pressure on the same component.
  4. Restarted servers initially look healthy, then struggle when ordinary traffic returns.

A health check is a test of readiness to serve requests. Failure does not necessarily mean broken hardware. Equally, a successful check immediately after a restart does not prove that a server can remain stable under load.

This explains why recovery may require several coordinated measures: repair the underlying behaviour, control incoming connections, restore capacity gradually and verify that the improvement lasts. Moving traffic alone cannot fix a software problem shared by all the destinations receiving it.

Why one cloud outage can affect several business functions

A business may think it has separate systems for email, appointments, documents, customer management and security administration. Those systems may depend on common services underneath.

For example, a booking tool might read calendars through Microsoft Graph. A customer workflow might send approval messages through a mailbox. A document-sharing process might need a notification or a protection service before the recipient can use it.

Those are examples to investigate in your own environment, rather than a claim that every such workflow failed during this outage.

The useful question is: what else stops if this service becomes unavailable?

An application sign-in failure also needs careful interpretation. If an application cannot complete an authorisation step because a dependency is unavailable, that does not establish that the entire identity platform has failed.

The same thinking applies outside Microsoft 365. Our discussion of Australian website resilience and the INDIGO cable faults examines why the full dependency chain matters. A service's own infrastructure is only part of its availability story.

What should small businesses do before the next outage?

Start with the work that must continue for the next few hours. For a consultancy, that might mean contacting clients and finding today's appointments. For a trades business, it might mean accessing job addresses, safety information and urgent customer requests.

Maintain emergency contacts somewhere independently accessible. An encrypted local copy or a securely stored printed list may be enough. Keep it current and limit it to the information needed during disruption.

Choose an approved alternative communications channel. Staff should know when to use phone calls, SMS or another agreed service, who can authorise its use, and what information may be shared. Check whether that channel relies on the same sign-in service as the system it is meant to replace.

Make essential records securely accessible offline where appropriate. Confirm that files have actually downloaded, that authorised staff can open them, and that device encryption and access controls remain in place. A cloud backup is useful for recovery, but it is not automatically an immediately usable working copy during an outage.

Give someone responsibility for monitoring Microsoft's public status page or Health → Service health in the Microsoft 365 admin centre. Staff also need a way to receive that person's updates when company email is unavailable.

Define recovery as well as continuity. Record work completed through alternative channels, reconcile it once normal service returns and check for missed requests or duplicate transactions. Reopening an inbox is only one part of resuming normal operations. OneCall's recovery updates noted that mailbox access could improve while queued email was still being processed.

A practical business continuity checklist

  • Identify which business activities would stop if Microsoft 365 were unavailable, including connected applications.
  • Keep emergency staff, customer and supplier contacts independently accessible and current.
  • Establish an approved alternative communications channel and rules for its use.
  • Confirm that essential documents can be accessed securely during an outage.
  • Document how urgent customer requests, appointments, approvals and invoices will be handled.
  • Ensure administrative recovery does not depend exclusively on the unavailable service; securely maintain approved recovery procedures and credentials.
  • Maintain MFA, access controls and data-handling rules throughout the incident.
  • Test the plan periodically, then reconcile temporary records and actions after recovery.

Test with a realistic exercise: assume email is unavailable for half a day and ask staff to find tomorrow's appointments, contact a customer and handle an urgent request. Record the obstacles. A plan stored only in the inaccessible system cannot guide the people who need it.

Authentication needs its own dependency check

Email-delivered one-time codes create a direct dependency on email delivery. If the mailbox cannot be reached or messages are delayed, the code may be unavailable when needed.

Authenticator applications, FIDO2 security keys and passkeys can reduce dependence on emailed codes. These are ways to prove your identity; the identity provider is the service that checks that proof and completes the sign-in. Alternative authentication methods cannot compensate for an unavailable identity provider. An authenticator may still produce a code while the service needed to accept it is unavailable.

Review recovery arrangements before an incident, including who can access approved emergency accounts and how those accounts are protected and monitored. Our article on stronger MFA for small businesses discusses authentication options. Continuity planning should preserve those protections.

The security risks created by improvised workarounds

An availability incident can create secondary security and privacy risks even when its cause is technical.

Under pressure, someone might forward confidential material to personal email, upload customer records to an unapproved file-sharing service or ask for security controls to be disabled. Those actions can move information outside the organisation's usual access, retention and audit arrangements.

Set the boundaries in advance. Identify which tasks may continue through approved alternatives, which information can move and which activities should wait. Staff need a workable procedure and a person to contact when the procedure does not cover their situation.

Do not disable MFA, weaken spam protection or remove document protections simply to make work move faster. Verify unusual payment instructions or requests for urgent access through a known, independent contact method.

The lesson also connects to outsourced IT and third-party security risk: a supplier's role does not remove the need to understand your own exposure and response responsibilities.

Does this mean businesses should abandon Microsoft 365?

No. Cloud services can provide substantial operational and security advantages, including managed infrastructure, collaboration and centralised administration. Availability is still a design and planning concern.

Moving to Google Workspace, another cloud provider or self-hosted email would introduce a different set of dependencies and responsibilities. It would not remove the need for continuity planning.

For a small business, the proportionate response is to understand critical dependencies, make essential information accessible during disruption and practise an approved way to keep urgent work moving.

A business does not need to eliminate every technology dependency. It needs to know which ones are critical and establish how essential work can continue when one becomes unavailable.

Share on LinkedIn