Skip to main content

When Microsoft Went Dark: What the November Outage Teaches Us About Service Continuity

On Thursday 12 November, starting around midday in the United States, Microsoft 365 began to fail for organisations around the world. Outlook would not load, email stopped flowing, Teams calls dropped and Copilot went quiet. Within an hour, outage reports were coming in from more than 70 countries and territories. Microsoft did not declare full recovery until Saturday.

For many IT teams, Thursday evening and Friday became an unplanned test of something most of us only discuss in workshops: what happens to our services when a supplier we cannot live without simply stops working? This article looks at what happened, why it matters even if your organisation was only mildly inconvenienced, and the practical continuity steps every IT team can take before the next one.

What happened

The disruption hit the core of Microsoft 365. Exchange Online mailboxes became unavailable for many users, so people could not send or receive email or even open their inboxes. Teams, Outlook and Copilot were all affected, and problems continued into Friday in many regions as recovery took longer than Microsoft had expected.

The impact went well beyond office productivity. According to Associated Press reporting, government departments in several countries lost email communications. In the UK, the government said it was working with Microsoft and had contingency plans in place, a trial at a Manchester court was halted, police were unable to receive online crime reports, and pharmacies reported delays because emailed prescriptions were caught up in the disruption. Banks, airlines and businesses across Asia, Europe and North America reported problems too.

Microsoft said the outage was not related to a cyber attack and that there was no indication of malicious activity. Its explanation was more ordinary, and for IT service management professionals more instructive: the problem was introduced by an issue during an earlier network capacity upgrade, which surfaced after a more recent network change. Services were then restored gradually, region by region.

Why it matters beyond Microsoft

It would be easy to treat this as Microsoft’s problem. It was their platform and their change. But the services that failed for your users were your services. When a GP surgery cannot receive prescriptions, a court cannot run a trial or a sales team cannot reach customers, nobody is asking which data centre was at fault. They are asking why the work stopped and when it will start again.

This outage also fits a pattern. Over the past two years, large incidents at cloud and security providers have repeatedly shown how concentrated our dependencies have become. Most organisations now rely on a handful of platforms for identity, email, collaboration, documents and increasingly AI. That concentration brings real benefits in cost and capability, but it also means a single fault can remove several services at once.

The question for IT leaders is not whether a major provider will have another bad day. It will. The question is whether your organisation can keep its most important activities going, at an acceptable level, while it happens.

Continuity is not the same as disaster recovery

Many organisations have a disaster recovery plan that describes how to restore systems after a failure: backups, secondary sites, failover procedures. That is valuable, but it assumes the recovery is in your hands. When the failing service is a global SaaS platform, there is very little for you to recover. You cannot fail Microsoft 365 over to your own data centre.

Service continuity management asks a different question: how do we continue to deliver what matters most while a service is unavailable, regardless of whose fault it is? In ITIL terms, it sits alongside availability management and information security management, and it links directly to the business continuity plans owned by the wider organisation. Standards such as ISO/IEC 20000 and ISO/IEC 27001 both expect continuity to be planned and tested, not just written down.

Question Disaster recovery Service continuity
Main focus Restoring systems and data Keeping critical activities going
Typical owner Infrastructure or operations team Service owners with the business
Works when the fault is Inside your own estate Anywhere, including suppliers
Key measures Recovery time and recovery point objectives Minimum service levels and maximum tolerable disruption
Example action Fail over a database to a secondary site Switch the service desk to phone and SMS updates

Seven lessons from the outage

1. Know which services depend on which suppliers

Many teams discovered on 12 November that far more depended on Microsoft 365 than they had realised. Approval workflows that run through Outlook, shared mailboxes that receive customer orders, Teams channels used for on-call coordination, Power Automate flows that move data between systems. If your service catalogue or CMDB does not show these dependencies, you cannot judge the impact quickly or prioritise sensibly.

2. Your communication channels probably share the same point of failure

The irony of a collaboration outage is that it removes the tools you would normally use to coordinate the response. If your major incident process starts with “open a Teams bridge” and your user updates go out by email, both fail at the same time as the service. Every organisation needs at least one out-of-band channel: a phone bridge, an SMS or messaging tool on a separate platform, a status page hosted elsewhere.

3. Identity is the hidden dependency

When a platform that provides sign-in struggles, the effect spreads to every application that relies on it. Check which of your critical systems depend on a single identity provider, and make sure you have tested break-glass accounts that do not.

4. Manual workarounds need to be planned, not improvised

The organisations that coped best had workarounds ready: paper forms, a printed contact list, an alternative way to receive orders, a documented process for urgent approvals. The NHS has long relied on exactly this kind of fallback during outages. Workarounds only help if people know they exist, know when to use them and know how to bring the data back into the system afterwards.

5. Communicate early, honestly and often

Users forgive outages much more readily than silence. A short message within the first fifteen minutes saying “we know, we are not the cause, here is what to do meanwhile” prevents hundreds of duplicate calls to the service desk. Agree in advance who can publish updates, through which channels, and how often.

6. Watch your supplier, not just your own monitoring

Your monitoring may show everything green while users cannot work, because the fault sits outside your estate. Subscribe the right people to supplier service health notifications, add key suppliers’ status feeds to your event management, and agree who checks them at the start of any widespread incident.

7. Recovery has a tail

Services did not come back for everyone at once. Mail queues had to drain, some users saw delays well after the official fix, and people had a backlog of work to catch up on. Plan for the recovery period as well as the outage: extra service desk capacity, clear guidance on delayed email, and checks that workaround data has been entered correctly.

What a useful continuity plan looks like

A continuity plan does not need to be long to be useful. In fact, the best ones are short enough to be read under pressure. For each critical service, it should answer a few practical questions.

  • What is the service and who owns it? Name a person, not a team.
  • What does it depend on? Internal components, suppliers, identity, networks and people.
  • How long can the organisation cope without it? Agree a maximum tolerable period of disruption with the business, not just IT.
  • What is the minimum acceptable service? Often a reduced service is enough for a day or two.
  • What are the workarounds? Step by step, including who switches them on and how.
  • How will we communicate? Channels, audiences, templates and approvers.
  • How will we return to normal? Including data reconciliation and a review afterwards.

Keep the plan where people can reach it during an outage. A continuity plan stored only in SharePoint is not much help when SharePoint is the thing that is down.

Test it, lightly and often

A tabletop exercise is the cheapest and most effective way to find gaps. Gather the service owner, the service desk lead, someone from communications and a business representative for an hour. Describe a scenario such as “email and Teams are unavailable across the organisation for 24 hours, starting at 2pm on a Thursday”, and walk through what each person would do. You will almost certainly find a missing contact, an untested workaround or a dependency nobody had noticed.

Key takeaways

  • The November outage was caused by a change, not an attack, and it removed core services for organisations worldwide.
  • Disaster recovery is not enough when the failure sits with a supplier. Service continuity planning fills that gap.
  • Map dependencies, including identity and automated workflows, so you can judge impact quickly.
  • Make sure at least one communication channel does not depend on the platform that might fail.
  • Plan and rehearse workarounds and the recovery period, not just the outage itself.

Five things to do this month

  1. Run a short review of the outage. What stopped, what coped, what did users do on their own? Capture it while memories are fresh.
  2. List your top ten services and their critical suppliers. A spreadsheet is fine to start with.
  3. Set up an out-of-band communication channel for major incidents and test it.
  4. Check your break-glass and emergency access accounts. Confirm they work and that the right people can find them.
  5. Book a one-hour tabletop exercise for your most critical service before the end of the year.

None of this requires a large programme. It requires someone to take ownership and a willingness to look honestly at how dependent your services have become. If you want to go further, our articles on building a resilient service desk and aligning OLAs with supplier SLAs cover two closely related areas.

Want to know how ready your services are for the next outage?

We help organisations map critical services and dependencies, design practical continuity plans and run tabletop exercises. Book a free scoping call to talk it through.

Book a free scoping call

Similar Posts