Skip to main content
Technology

Microsoft 365 was disrupted for 36 hours. Can your remote team work without it?

An outage communication plan for remote teams. What the 36-hour Microsoft 365 disruption showed, and the one-page backup plan to set up this week.

James Hitch
James Hitch· COO
Published Nov 19, 2026
36 min read
Microsoft 365 was disrupted for 36 hours. Can your remote team work without it?

Microsoft 365 experienced a major worldwide disruption on 12 November 2026, with Microsoft's incident record spanning nearly 36 hours before the company declared services restored on 14 November. Outlook, Teams, file access and administration tools were affected, leaving businesses struggling to carry out routine work.

For remote teams that depend on Microsoft's ecosystem for daily communication and collaboration, the incident exposed an uncomfortable weakness: what happens when the tools everyone relies on suddenly become unavailable?

The answer should not be to wait for Microsoft to fix the problem. Businesses need a practical continuity plan that allows employees to keep working when their primary systems fail. A secondary communication channel, an accessible contact list and a straightforward one-page recovery procedure can make a significant difference. However, these measures only work if they are tested before an outage occurs.

Key facts about the Microsoft 365 outage

QuestionAnswerSource
What happened?Microsoft 365 experienced a worldwide disruption beginning on 12 November 2026. Microsoft confirmed that services had been restored to all remaining customers and regions in the early hours of 14 November.Petri, The Register
How long did the outage last?Nearly 36 hours by Microsoft's own clock: 35 hours and 56 minutes, counted from its logged incident start at 18:26 UTC on 12 November to its all-clear at 1:22 AM ET (06:22 UTC) on 14 November.BleepingComputer, Petri
How long was traffic down?About 18 hours of downtime before traffic began returning, according to Kentik's Doug Madory.TechCrunch
How big was the reported impact?Downdetector showed more than 100,000 reported outages around 1:30 PM ET on 12 November, and the outage hit more than 70 countries.Global News, BBC
What caused the outage?Microsoft said the incident was introduced by an issue during an earlier network capacity upgrade. The company said it was not caused by a cyberattack or AI agents. A full post-incident report is still pending.Petri, BBC
What compensation does Microsoft offer?Exchange Online service credits can range from 25% to 100% of monthly fees, depending on the service's monthly uptime percentage and subject to a customer claim.Microsoft SLA

In this article

  • What happened to Microsoft 365 on 12 November 2026?
  • What is still not known about the cause?
  • Why does one vendor outage take down email, chat, files and logins at once?
  • What does a remote team lose first when its tools go down?
  • What belongs in a one-page outage plan?
  • How do you test the plan without disrupting a working day?
  • What does the service agreement pay back?
  • Where RocketDevs fits
  • Conclusion: your remote team needs a plan for the day the cloud goes dark
  • Frequently asked questions

What happened to Microsoft 365 on 12 November 2026?

Microsoft 365 began experiencing widespread failures on the afternoon of 12 November 2026. Services were not fully restored until the early hours of 14 November, according to Microsoft's own incident updates. Measured from the company's recorded start time to its final restoration notice, the disruption lasted almost 36 hours.

There is, however, some disagreement over exactly when the outage began. Users started reporting Outlook problems to Downdetector shortly after 12 PM ET, according to Wired. Microsoft's incident record, identified as MO1550821, lists the official start time as 1:26 PM ET, or 18:26 UTC, as reported by BleepingComputer. However, Petri reports an earlier start time of 16:17 UTC.

Using Microsoft's timestamp gives a duration of 35 hours and 56 minutes. Using Petri's earlier figure produces a disruption of approximately 38 hours. Either way, the outage lasted roughly a day and a half, far longer than most businesses can comfortably operate without their primary communication and productivity tools.

Wired headlined the outage as 24 hours long, and the BBC reported that services were beginning to recover on Friday afternoon UK time, nearly 24 hours after the problems first emerged. Microsoft's own incident record runs longer because it counts to the all-clear.

The following timeline outlines how the incident developed, based on Microsoft's updates as reported by Petri, The Register and BleepingComputer. All Eastern Time (ET) references are UTC minus five hours.

How the outage spread beyond email

The initial disruption was not limited to Outlook. According to BleepingComputer, Microsoft's service advisory covered Exchange Online, Teams, SharePoint Online, Copilot and Defender XDR.

TechCrunch reported that what initially appeared to be an Outlook and Teams problem developed into a much broader disruption affecting email, files and documents.

Microsoft's early updates indicated that servers were failing health checks and that an internal process had become overwhelmed. The Register reported that these problems reduced the capacity available to keep services running.

However, one of the most significant aspects of the incident was that customers' administrators, trying to work out what was wrong, could not open the tools they needed to investigate it.

When the people fixing the problem lose access too

The disruption created an additional challenge for IT administrators. Admins trying to investigate the outage found that some of their own management tools were unavailable.

The Register reported that administrators could not access Intune or the Microsoft 365 Apps Admin Center, both of which were affected by the wider disruption. TechCrunch also highlighted the difficulties caused by the loss of access to essential administrative services.

This illustrates an important weakness in relying on a single technology ecosystem. When the services that employees use to work are also connected to the systems administrators need to diagnose problems, an outage can become considerably more difficult to manage.

For remote businesses, the implications are particularly important. If employees cannot access their email, internal messaging, shared documents or IT support channels, they may have no reliable way to find out what is happening or when they can resume work.

The first problem for a remote team, therefore, may not be the outage itself. It may be finding a way to communicate that an outage has occurred.

How quickly did the disruption attract attention?

The scale of public interest was evident in several online activity measures.

Wikipedia's Microsoft 365 article recorded 5,271 page views on 13 November 2026, according to its page-view statistics. This was substantially higher than the approximately 1,057 to 1,185 daily views recorded between 9 and 11 November.

A Hacker News discussion linking to Microsoft's service status information also attracted considerable attention. The discussion reached 1,171 points and 993 comments.

These figures do not measure the number of people directly affected, but they demonstrate how quickly the incident became a major topic among technology professionals and businesses.

The outage hit more than 70 countries

The disruption was not confined to a particular region or industry. The BBC reported that more than 70 countries were affected.

In the United Kingdom, the outage disrupted services used by parts of the NHS, schools and police forces. In the United States, staff of the House of Representatives told Politico the outage stopped their work, and one said it had cost a day of legislative work, as the BBC reported.

The scale was also reflected in Downdetector's reporting figures. According to Petri, Downdetector logged more than 130,000 Outlook user reports at the peak of the disruption.

For businesses that depend on Microsoft 365 to manage daily operations, these numbers represent more than a temporary inconvenience. They highlight how much modern organisations depend on a relatively small number of cloud-based platforms.

A company might have employees working from several countries, communicating through Teams, sharing documents through SharePoint and handling customer enquiries through Outlook. If those services become unavailable simultaneously, the entire operation can grind to a halt.

The incident raises a question every remote business should consider: If your primary technology provider experiences a similar failure tomorrow, how much of your team could continue working?

What is still not known about the cause?

Microsoft has attributed the November 2026 outage to an issue introduced during an earlier network capacity upgrade. However, the company has not yet published its full post-incident report, meaning the precise technical cause remains unclear.

As Wired reported, Microsoft has promised a detailed explanation after completing its internal review. Until that report becomes available, businesses know broadly what went wrong but do not have enough information to understand exactly how the failure developed.

What Microsoft has confirmed so far

Microsoft's explanation evolved as its engineers investigated the disruption.

Initially, the company said it had not spotted any changes to its systems that might have sparked the incident, according to The Register.

It subsequently said the elevated load was not driven by an increase in traffic. Instead, a number of servers had become unexpectedly unavailable, leaving the remaining infrastructure struggling to handle demand. This was reported by BleepingComputer.

On 13 November, Microsoft provided a more specific explanation. Statements reported by Petri and Wired described the incident as introduced by an issue during an earlier network capacity upgrade that surfaced following a recent network change. Microsoft also said it was not related to a cyberattack, AI or agents.

This gives businesses a general understanding of the problem, but it does not establish the precise sequence of technical failures.

What remains unanswered?

Several important questions remain open:

  • Which network change triggered the failure? Microsoft has identified an earlier capacity upgrade as the source of the problem but has not publicly identified the specific change or configuration responsible.
  • Which systems were directly affected? The company has referred to network capacity issues, but the full technical explanation of the affected infrastructure has not been released.
  • Why did the disruption spread so widely? The outage affected multiple Microsoft 365 services, suggesting that the underlying problem had consequences beyond a single application.
  • Why did recovery take so long? Microsoft has not yet explained why initial recovery attempts failed to restore services reliably or why the disruption continued for almost 36 hours.

According to Petri, Microsoft's recovery efforts included freezing non-essential changes, throttling certain traffic and concentrating on patching an upstream authentication and routing component.

These details describe the reported recovery measures, not a confirmed explanation of the original root cause. Without Microsoft's post-incident report, it would be premature to treat them as a complete technical diagnosis.

Were there other Microsoft outages around the same time?

Two neighbouring incidents have attracted attention, but there is currently no confirmed connection between them and the November disruption.

The first involved a separate Teams incident, identified as TM1550488. According to BleepingComputer, the incident began at 9:45 AM ET on 12 November and was resolved by 10:53 AM ET. Microsoft has not confirmed whether it was connected to the much larger outage that followed.

The second was a multi-day Microsoft 365 disruption on 31 August and 1 September 2026. TechCrunch reported that Microsoft's status page attributed the earlier incident to a core authentication configuration used by multiple services. However, no evidence has established a relationship between that incident and the November failure.

Connecting these events without further evidence would be speculation. They should be treated as separate incidents unless Microsoft provides information demonstrating otherwise.

What does this mean for businesses?

Microsoft executive vice president Charles Lamanna said the company's teams were reviewing the incident and examining how similar problems could be prevented, according to BleepingComputer.

Businesses should follow the findings of that investigation, but they should not wait for Microsoft to publish its report before reviewing their own continuity arrangements.

The important lesson is that organisations cannot always predict which technical failure will disrupt their cloud services. They can, however, prepare for the consequences.

A remote team should have alternative ways to communicate, access essential contact information and coordinate urgent work without relying on the same platform that has failed. Its recovery procedures should also be documented somewhere accessible outside the affected ecosystem.

The practical takeaway is simple: plan around the possibility of losing access to your primary systems, rather than building your entire recovery strategy around preventing one particular technical failure.

Why does one vendor outage take down email, chat, files and logins at once?

Modern cloud platforms may present email, messaging, file storage and administration as separate products, but they often depend on the same underlying infrastructure. These shared systems include identity and authentication services, network connections, traffic routing and server capacity.

When one of these foundational components fails, the disruption can spread across several applications simultaneously. Employees may find that Outlook is unavailable, Teams will not connect, SharePoint files cannot be opened and administrators cannot access their management consoles. Although these appear to be separate failures, they may all originate from the same underlying problem.

Research into Microsoft's own cloud infrastructure supports this explanation. A peer-reviewed study examining a year of Azure outage data describes outages as severe incidents that always affect multiple services, and notes that production cloud systems contain a large number of interdependent services. In other words, failures tend to arrive together, and having several applications does not necessarily mean having several independent ways to keep working.

Three underlying layers help explain why a single technical failure can disrupt an entire business ecosystem.

Identity and authentication

Identity services sit between employees and the applications they use. They verify credentials, manage access permissions and allow users to move between connected services without repeatedly signing in.

This makes authentication a critical dependency. If the underlying identity system fails, applications that are otherwise functioning may become inaccessible.

Microsoft's August to September 2026 outage provides an example. Microsoft attributed the disruption to a core authentication configuration used by multiple Microsoft 365 services. A problem with this shared component affected access across the wider platform.

For remote businesses, this creates a particular vulnerability. Employees may have working internet connections and functioning devices but still be unable to access the tools needed to do their jobs.

Network and routing

Network infrastructure determines whether information can travel between users and the services they need. Even when applications and servers are healthy, problems with routing can prevent traffic from reaching them.

Meta experienced a particularly significant example on 4 October 2021. According to its engineering post-mortem, a command intended to assess backbone network capacity unintentionally disconnected all backbone connections. The resulting disruption affected Facebook, Instagram and WhatsApp.

Microsoft attributed its November 2026 incident to an issue during an earlier network capacity upgrade. During recovery, Petri reported, the company froze non-essential changes, throttled certain traffic and worked on an upstream authentication and routing component.

Application infrastructure and server capacity

Cloud applications need enough healthy servers to process requests. When servers fail or become unavailable, the remaining infrastructure must handle additional demand.

Microsoft reported that servers were failing health checks and that an internal process had become overwhelmed during the November incident. This reduced the capacity available to support normal traffic.

Slack experienced a related problem on 4 January 2021. Its engineering post-mortem described an AWS Transit Gateway becoming overloaded and failing to scale quickly enough to handle demand.

The result can be a chain reaction: fewer healthy servers mean more traffic per server, which increases pressure on the remaining infrastructure and can cause further failures.

Why can recovery take longer than the original failure?

A second problem is that restoring cloud services is not necessarily as simple as restarting a server or reversing a software update.

Recovery depends on identifying the underlying problem, determining which systems are affected and implementing a fix without creating additional failures.

Gergely Orosz told the BBC he could not recall a business email outage ever taking this long to return to normal. Professor James Ferryman of the University of Reading also explained that services cannot simply be switched back on everywhere simultaneously.

Academic research helps explain why diagnosis itself can take time. A peer-reviewed study examining hundreds of Microsoft Teams incidents found that many incidents last a long time because existing techniques and practices fail to detect and mitigate them quickly.

Similarly, a EuroSys research paper examining Microsoft's incident data identified manual analysis of logs and traces as a labour-intensive, error-prone process that places considerable pressure on engineers responsible for responding to incidents.

The practical difficulty is that engineers must understand what failed before they can safely restore it. Acting too quickly can make the situation worse.

Why restarting everything at once can make matters worse

The November outage also demonstrated the difficulties associated with restoring services under heavy demand.

According to Petri, Microsoft's engineers found that previously unavailable nodes could become overwhelmed again when live traffic returned. Restarting individual components did not necessarily resolve the underlying capacity problem.

Microsoft said that, given the scale of demand, it restored services gradually, validating stability at each stage.

This is consistent with what Professor Ferryman told the BBC: you cannot turn it all back on at once, and have to bring it back bit by bit.

Meta encountered a similar challenge during its October 2021 outage. In its engineering report, the company explained that restoring all services simultaneously could have caused further instability.

GitHub's August 2026 availability report provides another example. A routine deployment to an internal GitHub Actions service briefly reduced capacity at one site, pushing the remaining sites beyond their limits. The failure cascaded across clusters, and GitHub put the impact at approximately nine hours.

These examples demonstrate why recovery is often a controlled process rather than an immediate return to normal operations.

How other major cloud outages compare

Microsoft is not the only technology company to experience widespread disruptions caused by shared infrastructure dependencies.

CompanyDateShared dependencyWhat happened
Meta4 October 2021Backbone network and DNSA command intended to assess backbone capacity unintentionally took down all the connections in the backbone network.
Slack4 January 2021AWS Transit GatewayThe gateway became overloaded and did not scale quickly enough.
Microsoft 36531 August to 1 September 2026AuthenticationMicrosoft attributed it to a core authentication configuration used by multiple services.
Microsoft 36512 to 14 November 2026Servers and network capacityMicrosoft attributed it to an issue introduced during an earlier network capacity upgrade.

Sources: Meta Engineering, Slack Engineering, The Register, BleepingComputer, TechCrunch and Petri.

Another incident worth distinguishing is the CrowdStrike disruption in July 2024. This involved a different failure mechanism: a faulty update affected Windows devices running CrowdStrike's security software. Microsoft estimated that approximately 8.5 million Windows devices were affected, although this was an estimate rather than a directly verified count of every affected machine. Microsoft's statement provides further detail.

Although the technical causes differ, these incidents highlight a common business risk: organisations can become dependent on systems they do not control and cannot immediately repair.

What remote businesses should take away from this

The important lesson is not that businesses should abandon Microsoft 365 or stop using cloud platforms. These tools offer substantial benefits, particularly for distributed teams.

The problem is assuming that using several products from one provider automatically creates redundancy. If those products depend on the same infrastructure, they may fail together.

Businesses should therefore separate their critical communication and recovery processes from their primary technology provider wherever practical. Employees need an alternative way to contact colleagues, managers need access to essential operational information, and IT teams need recovery instructions that remain available when the main platform is inaccessible.

You cannot make Microsoft's engineers diagnose an outage faster, but you can make your own team's fallback faster. That is the difference between a temporary technology disruption and a business continuity crisis.

What does a remote team lose first when its tools go down?

The first thing a remote team loses during a major technology outage is often the ability to communicate. When colleagues work in the same office, they can walk across the room, speak to a manager or gather around a desk to work out what is happening. Remote employees do not have that option. Their communication, documents and management tools are usually accessed through the same cloud-based platforms.

This creates a particular vulnerability. If the technology provider hosting those services experiences a major disruption, employees may lose not only the tools they need to complete their work but also the means of coordinating their response.

The October 2021 Meta outage demonstrated just how serious this dependency can become. Cloudflare documented how Facebook, WhatsApp and Instagram became unavailable simultaneously. Meta's own engineering report provides further insight into the internal consequences.

When communication and recovery systems fail together

Cloud-based email is already an established part of modern business operations. According to Eurostat, 36.65% of EU enterprises with between 10 and 249 employees paid for email as a cloud service in 2023.

The difficulty is that businesses often use the same technology ecosystem for both everyday communication and emergency coordination.

Meta's 2021 outage provides a particularly revealing example. The company explained that the loss of its DNS infrastructure made many of its internal tools unavailable, including systems normally used to investigate and resolve technical incidents. Engineers also struggled to reach the company's data centres because the networks connecting them were down.

According to Snopes, Meta's own employees could not access company email or internal communications platforms, and were locked out of their offices because their digital badges stopped working.

The problem was not simply that Facebook's services were unavailable to customers. The company had also lost access to some of the systems its own employees needed to coordinate the recovery.

Smaller incidents have exposed similar weaknesses. In its report on the 4 January 2021 disruption, Slack Engineering explained that its own dashboarding and alerting service became unavailable while engineers were investigating the incident. During Microsoft's November 2026 outage, customers' administrators trying to work out what was wrong could not open Intune or the Microsoft 365 Apps Admin Center, according to The Register.

These examples highlight an important distinction: a company needs systems for doing its work, but it also needs independent systems for managing what happens when those tools stop working.

Why some employees can work while others cannot

A major cloud outage does not always affect every service or employee in exactly the same way.

During the November Microsoft disruption, the BBC reported that many users could not send or receive email, while some were still able to use Teams messaging and meetings.

The Register also reported Microsoft's explanation that services not dependent on Exchange Online continued functioning.

This uneven access can create additional confusion for remote teams. One employee may be able to send messages through Teams, while another cannot access their mailbox or shared files. Managers may assume that the problem is isolated to an individual device or internet connection when the underlying issue is actually with the service provider.

Without an independent communication channel, establishing the extent of the disruption becomes considerably harder.

The four things a remote team loses during an outage

The consequences generally extend beyond the initial loss of access. Four areas of business operations are particularly vulnerable.

Communication

Employees can no longer reliably contact colleagues through the tools they normally use. Managers struggle to distribute instructions, teams cannot coordinate tasks and employees may not know whether others are experiencing the same problem.

Visibility

IT administrators and managers may lose access to the dashboards, status information and management consoles they need to understand what is happening. Without reliable visibility, it becomes difficult to distinguish between a vendor-wide outage, a local network problem and an individual device failure.

Customer communication

Businesses may struggle to respond to customer enquiries, provide updates or manage urgent requests. If customer correspondence is handled through the affected email platform, even communicating that services are temporarily disrupted can become difficult.

Business continuity plans

Emergency procedures are not particularly useful if employees cannot access them. When recovery instructions, contact details and operational documents are stored exclusively in SharePoint, OneNote or another affected service, the team may lose access to the very information it needs to recover.

This is why business continuity documents should be available independently of the primary technology platform.

How businesses improvised when Microsoft 365 stopped working

The November outage forced some organisations to return to older communication methods and manual processes.

According to Tom's Hardware, which cited Japanese reporting, some workers reportedly resorted to fax machines when their usual digital tools became unavailable.

A South African winemaker told the BBC his entire business had effectively been brought to a standstill, including communication with exporters.

In the United Kingdom, Petri reported that several NHS trusts and regional police forces had to fall back on phone lines.

These examples demonstrate that business continuity does not always require sophisticated technology. Sometimes, the ability to make a telephone call, access a printed contact list or send a basic message through an alternative service is enough to keep essential operations moving.

Is maintaining a second technology platform realistic?

There is an important distinction between having a complete replacement for Microsoft 365 and having a basic alternative communication system.

Jos Poortvliet, co-founder of Nextcloud, a Microsoft competitor, told The Register that when everything hangs on a single vendor, companies cannot switch providers quickly or keep a second setup as a backup.

Matt Saunders of the technology consultancy Adaptavist made a similar point to the BBC. Classic IT disaster planning suggests using a second service for resilience, he said, but in practice it is almost impossible for a business to function without tools such as Microsoft.

Both perspectives highlight the practical challenges of technology redundancy. Maintaining two complete productivity ecosystems can be expensive, complicated and difficult to justify for smaller businesses. However, that does not mean organisations should have no alternatives.

A remote team does not necessarily need a second email platform, document management system and administrative environment ready to take over immediately. It needs a smaller, independent way to communicate, access essential information and coordinate urgent work.

For example, a separate messaging channel, a telephone contact sheet and an offline copy of the team's emergency procedures could provide enough support to keep essential tasks moving until the primary services return.

The real lesson is that remote businesses do not need to duplicate their entire technology infrastructure. They need to ensure that losing one provider does not also eliminate their ability to communicate, make decisions and respond to customers.

Hire the top 2%.

Vetted developers, part-time or full-time, remote and ready, from $9.99/hr.

What belongs in a one-page outage plan?

A remote team does not need an elaborate disaster recovery strategy to handle every technology disruption. It needs a practical, accessible document explaining what employees should do when their primary communication and productivity tools become unavailable.

A good one-page outage plan should answer five questions: Who declares the outage? How does the team communicate through a different provider? How are customers contacted? What should employees work on while services are unavailable? And who decides when normal operations can resume?

The document should also be available somewhere the outage cannot reach. Keeping it exclusively in SharePoint, OneNote or another cloud-based platform that depends on the affected provider defeats the purpose. A printed copy, supported by an offline digital version, gives employees access to instructions when they need them most.

This approach follows principles outlined in the National Institute of Standards and Technology's (NIST) contingency planning guidance, SP 800-34 Rev. 1. The guide describes call trees as a common manual notification method and recommends accounting for primary and alternative contact methods, including what to do when someone cannot be reached.

It also says personal email may be useful if the local network is down, and that copies of the notification procedures can be kept securely at alternate locations. Although the guidance was developed for US federal information systems, a smaller business can adapt its principles.

The following template provides a practical starting point. Each alternative should operate independently of the primary technology provider wherever possible.

Six rules that make the plan work

Having a document is a starting point, but the arrangements it describes must work in practice. Several straightforward rules can help prevent a relatively minor technology disruption from becoming a larger operational problem.

Choose an alternative provider, not another version of the same dependency

A backup communication platform is only useful if it remains available when the primary provider experiences an outage.

For example, if a company uses Microsoft Teams as its main communication channel, its alternative should not depend on Microsoft authentication or infrastructure. Employees should already have the secondary application installed, registered and tested on their personal devices.

The same principle applies to email, video meetings and file storage. The objective is not to purchase a duplicate version of every service but to ensure that the most important communication channels do not share the same point of failure.

Every remote team should have a contact sheet containing employees' names, personal phone numbers, personal email addresses and instructions explaining who contacts whom.

This information should be stored somewhere accessible to more than one authorised person. If the only copy is held by an employee who is unavailable, the business has created another single point of failure.

This is closely related to the concept of a bus factor, which describes how vulnerable a business becomes when essential knowledge or responsibilities depend on too few people. Our guide to the bus factor in small teams explains why distributing critical knowledge is important.

Define what happens during the first hour

Employees should not have to invent an emergency procedure while services are unavailable. The plan should establish who declares the incident, who contacts employees, who handles customer communication and who monitors updates from the provider.

A simple sequence is usually sufficient:

  • The team lead confirms that the problem is affecting the wider service rather than an individual device.
  • The team is notified through the alternative communication channel.
  • The operations owner checks that every employee has received the message.
  • The account owner contacts customers who may experience delays.
  • The IT lead monitors the provider's updates and advises when normal operations can resume.

This gives everyone a clear responsibility instead of allowing several employees to duplicate the same work or assume someone else is handling it.

The same principle of assigning responsibilities and working to a defined response window applies to cybersecurity incidents. Our guide to responding to a critical vulnerability in 48 hours explores how smaller teams can organise an effective response.

Tell employees what not to change

An outage can create pressure to start troubleshooting immediately. However, employees and administrators making unnecessary changes can introduce additional problems while the provider is still restoring its services.

During the November incident, Petri reported advice for administrators to avoid rebuilding Outlook profiles, changing local DNS records or modifying tenant routing while services were recovering.

The outage plan should therefore include a clear instruction: do not make significant configuration changes unless authorised by the designated IT lead.

This matters because a business should not turn a vendor-side failure into an additional internal problem through unnecessary troubleshooting.

Do not repeatedly resend delayed messages

Another important consideration is that an outage can delay messages rather than permanently prevent them from being delivered.

During the November disruption, Microsoft extended the email timeout window to 48 hours, allowing delayed messages more time to be delivered as services recovered, according to BleepingComputer.

Employees should therefore avoid repeatedly resending the same email simply because they have not received an immediate response. Duplicate messages can create confusion, particularly when the original eventually arrives.

The plan should explain when employees should wait, when they should use an alternative communication method and how urgent messages should be prioritised.

Establish how the team will identify an outage

Employees need a reliable way to distinguish a widespread vendor disruption from an individual connectivity problem.

Checking the provider's official status page is a sensible first step. However, businesses should also consider an independent monitoring service that does not rely on the same infrastructure being monitored.

One example is Uptime Kuma, an open-source, self-hosted monitoring tool released under the MIT licence. Its repository describes it as an easy-to-use self-hosted monitoring tool with multiple status pages. The project is widely used, with more than 94,000 GitHub stars as of 15 November 2026.

The important consideration is where the monitoring system is hosted. If it runs entirely within the same cloud environment as the services it monitors, it may become unavailable during the same incident.

Does a small business really need a formal continuity policy?

Not every business needs an extensive business continuity management programme.

ISO 22301 is an international standard covering business continuity management systems. It provides a framework for organisations that need structured processes to prepare for, respond to and recover from disruptive events.

However, a five-person remote business does not need to pursue certification before putting basic safeguards in place.

The Server Fault business continuity tag, for example, has only 14 questions as of 15 November 2026. That is a rough signal at best, but it suggests that smaller technical teams will find little crowd-sourced practice on the topic.

The more useful starting point is to document the essential arrangements, assign responsibility and test whether employees can follow them.

A backup is not the same as a business continuity plan

Businesses should also understand the difference between protecting their information and keeping their operations running.

A data backup protects information against loss, corruption or accidental deletion. It allows a business to restore files if something happens to its primary storage environment.

An outage plan addresses a different problem. It explains how employees communicate, make decisions, contact customers and continue essential work while their normal systems are unavailable.

Our guide to protecting data outside your cloud region examines the importance of maintaining independent data backups. That is an essential part of disaster recovery, but it does not automatically provide a way for employees to keep working during an outage.

Businesses need both arrangements because restoring information and maintaining daily operations are two separate responsibilities.

An outage plan is not an exit strategy

Finally, experiencing a major disruption does not automatically mean a business should abandon its existing technology provider.

Microsoft 365 remains a widely used productivity platform, and a temporary outage does not necessarily justify the expense and disruption of migrating an entire organisation to another ecosystem.

A continuity plan is designed to help a business remain operational while its primary provider resolves a problem. An exit strategy, by comparison, explains how an organisation would permanently move away from a vendor if its long-term requirements changed.

Our guide to creating a vendor lock-in exit plan explores the latter.

For the November incident, the immediate priority is simpler: maintain an independent communication channel, protect access to essential information and give employees clear instructions.

A one-page outage plan will not prevent a cloud provider from failing. It can, however, prevent that failure from bringing your entire remote team to a standstill.

How do you test the plan without disrupting a working day?

A 30-minute drill can test an outage plan without interrupting normal operations. Use a tabletop exercise and a timed call tree. Do not switch off working systems or deliberately disrupt live services.

NIST defines a tabletop exercise as a discussion-based test in which participants work through a simulated disruption. For low-impact systems, this can be sufficient, provided the exercise covers the main points of contact. NIST also gives testing the call tree as an example of a test, to check whether people can be reached within prescribed time limits (NIST SP 800-34 Rev. 1). The 30-minute schedule below is our suggested format, not a timetable prescribed by NIST.

  • Minutes 0 to 5: set the scenario. The exercise lead announces: “Microsoft 365 is down, and the status page is slow.” Everyone treats the scenario as real and stops relying on Microsoft 365 for the exercise.
  • Minutes 5 to 15: test the call tree. Start the clock. Each person must confirm that they have received the message through the alternative communication channel. Record who responds, the order of responses and how long the full call tree takes.
  • Minutes 15 to 25: check the fallback arrangements. Work through each part of the outage plan. Can someone access the printed copy? Is the customer holding message ready? Does the contact sheet contain outdated numbers? Can the team reach the alternative meeting platform?
  • Minutes 25 to 30: assign the fixes. Record every gap and give each one to a named person with a deadline. A problem that has no owner is unlikely to be fixed before the next outage.

Run the exercise once, then repeat it whenever someone joins or leaves the team. Contact details change, responsibilities shift and plans become outdated. If you are building a team, establishing this habit early is useful. Our guide to what to do the morning after a production mistake is a helpful next read.

The drill may uncover problems that seem minor until an actual outage makes them urgent. The contact sheet might still contain a former employee's number. Nobody may have installed the alternative chat app on their phone. The only copy of the plan might be saved in the shared drive that is currently inaccessible.

These are simple problems to fix when the business is running normally. Finding them during a real disruption is a different matter. Test the plan on a quiet Tuesday, not when your team is already trying to recover from a difficult day.

What does the service agreement pay back?

Very little, and only if you submit a claim. Under Microsoft's Online Services SLA (October 2026 edition), Exchange Online service credits are 25%, 50% or 100% of the applicable fees when monthly uptime falls below 99.9%, 99% or 95%, respectively. These credits are deducted from service fees. They are not cash refunds.

Two details determine how the agreement applies. First, Microsoft defines Exchange Online downtime as a period when users cannot send or receive email through Outlook Web Access. Availability is measured in user-minutes. Second, customers must submit a claim within the specified deadline. For services other than Azure, Microsoft must receive the claim by the end of the period following the month in which the incident occurred. Microsoft says it typically processes claims within 45 days.

Consider the November calculation. The month has 30 days, or 43,200 minutes.

Monthly uptime thresholdDowntime allowed in NovemberService credit if uptime falls below the threshold
99.9%43.2 minutes25%
99%7.2 hours (432 minutes)50%
95%36 hours (2,160 minutes)100%

Now apply those thresholds to the reported outage. If every user had been unable to send or receive email for the full 35 hours and 56 minutes, monthly uptime would have been approximately 95.01%. That is just above the 95% threshold for a 100% credit, leaving the incident in the 50% credit tier.

Using Kentik's estimate of 18 hours of downtime instead would produce uptime of approximately 97.5%. The result would still be a 50% credit. However, whether Microsoft measures the incident in either of these ways is for Microsoft to determine under the agreement. This is an illustrative calculation, not legal advice or a guarantee of the credit a customer would receive.

The more important question is what the lost working time costs the business.

The US Bureau of Labor Statistics reported average hourly earnings of $38.00 for private-sector employees in October 2026, a preliminary figure. If a team of 20 loses one eight-hour working day, the wage cost alone is:

20 employees × 8 hours × $38.00 = $6,080

That figure excludes lost sales, overtime, delayed projects and the cost of redoing work. A 50% service credit would equal that wage cost only if the business paid $608 per user per month for the affected service. The calculation is $6,080 divided by (50% × 20 users), which gives $608 per user per month. Check your own invoice, but most businesses will not be paying anything close to that amount.

Wages are only one part of the financial impact. Uptime Institute's Annual Outage Analysis 2025 reports that 54% of respondents to its 2024 survey put the cost of their most recent significant, serious or severe outage above $100,000, while one in five reported costs exceeding $1 million. This survey covers data centre outages generally, not Microsoft 365 specifically, so it should not be treated as a measure of the November incident. It does, however, illustrate how quickly serious outages can become expensive.

The practical lesson is that the service agreement limits the vendor's financial exposure, while your business carries most of the operational risk. Claim any credit you are entitled to, but do not mistake it for compensation for everything the outage cost you. A working fallback plan is likely to be worth considerably more.

Where RocketDevs fits

A remote team is only as reachable as its contact sheet. When you bring on a developer, add their phone number, personal email address and alternative communication channel on day one. Include them in the next outage drill so they know how to reach the rest of the team if your usual tools become unavailable.

RocketDevs vets each developer for 6-8 hours and accepts only the top 2% of applicants. The 14-day risk-free trial gives you two weeks to assess how a new developer works with your team, including whether they understand and follow your outage procedures, before you commit.

Hire vetted developers through RocketDevs and build reliable communication practices into your remote team from the start.

Microsoft will restore its network on its own schedule. Your team controls how quickly it can reconnect with one another. If everyone can reach each other through a second channel, using information available offline, a disruption that might otherwise cost a full working day may be contained to a much shorter delay.

Conclusion: your remote team needs a plan for the day the cloud goes dark

The Microsoft 365 outage exposed an uncomfortable truth about remote work: a business can have talented people, reliable internet connections and expensive software, yet still be unable to function when one provider fails. When email, chat, files and administrative tools disappear together, the problem is no longer technical. It becomes an operational crisis. People cannot reach colleagues, customers cannot get answers, deadlines slip and an ordinary working day can turn into a costly exercise in damage control.

You cannot control when Microsoft, Google, Slack or another cloud provider experiences its next major disruption. You cannot make a vendor investigate faster, restore capacity sooner or guarantee that its next update will work. What you can control is how dependent your business becomes on that single point of failure.

A printed contact sheet, an alternative communication channel, offline access to essential information and clearly assigned responsibilities may seem like basic precautions. They are also the difference between a team that spends hours trying to find one another and a team that knows exactly what to do when its usual tools stop working. The plan does not need to be complicated. It needs to exist, be accessible and be tested before anyone needs it.

For founders building remote businesses, resilience should be part of how the team operates from its first hire. Every developer should know how to communicate during an outage, where to find critical information and who makes decisions when the normal systems are unavailable. Hiring capable people matters, but building a team that can keep working when conditions are far from ideal matters just as much.

The next major cloud outage is not a question of whether your provider will have a bad day. It is a question of what happens to your business when it does.

Your software may be in the cloud. Your ability to keep the business running should never depend on it.

Frequently asked questions

What caused the Microsoft 365 outage in November 2026?

Microsoft attributed the outage to a technical issue introduced during an earlier network capacity upgrade. It ruled out a cyberattack and AI agents as the cause, but the precise technical mechanism has not yet been made public. Microsoft has promised a full post-incident report once its investigation is complete, as reported by the BBC.

How long was Microsoft 365 down?

Counted from Microsoft's logged start (18:26 UTC on 12 November) to its all-clear (1:22 AM ET on 14 November, which is 06:22 UTC), the incident ran nearly 36 hours. Kentik's Doug Madory said traffic began returning after about 18 hours of downtime, as reported by TechCrunch. The two figures measure different things: Microsoft's window runs to the all-clear, while Kentik measured when traffic was down.

What should a team do when Microsoft 365 is down?

Switch to your alternative communication channel, use the printed contact sheet to reach colleagues and avoid changing your own system settings while the provider investigates. During the November incident, Petri reported that IT administrators were advised to avoid rebuilding Outlook profiles, changing local DNS records or modifying tenant routing rules while Microsoft worked on recovery. Check the provider's status feed from a device that does not depend on the affected service.

Does Microsoft compensate customers for outages?

Microsoft offers service credits under its Online Services SLA rather than direct cash compensation. For Exchange Online, credits are 25%, 50% or 100% of the applicable fees when monthly uptime falls below 99.9%, 99% or 95%, respectively. Customers must submit a claim by the end of the period following the month in which the incident occurred. The credit is applied to the service bill, not paid as a cash refund.

Should a small team pay for a backup chat tool?

Not necessarily. What matters is whether the alternative uses a different provider and whether every team member has already installed, signed into and tested it on their phone. A backup tool that nobody has configured before an outage is unlikely to be useful when the primary service fails. Even a simple alternative channel can make a difference if the whole team knows how to use it.

James Hitch, COO at RocketDevs.LinkedIn

Sources

Serious devs. Serious value.

The top 2% of applicants, rigorously vetted, from $9.99/hr. Part-time or full-time, dedicated to your team.

  • Top 2% of applicants
  • 6–8 hours of human vetting
  • 14-day risk-free trial
James Hitch

Written by

James Hitch

COO

James Hitch is the COO of RocketDevs, where he runs sales, recruiting, and the vetting operation that accepts only the top 2–3% of developer applicants. He cares about putting accessible, elite engineering talent within reach of founders and startups worldwide, at a fair price. He writes about technical hiring, building AI-native engineering teams, and how startups can access elite developers affordably.

Share this article

Help others discover this content