
The Microsoft 365 outage is over and the dust has settled. A day without core services is survivable for many businesses; a week is not simply a longer version of the same disruption. It changes what stops, what quietly slips and what leaders may not be able to put right afterwards.
What the outage exposed
Microsoft's preliminary post-incident report, published on 15 November, puts the start at 16:20 UTC on Thursday 12 November and confirms customers were clear of active impact at 16:10 UTC on Friday 13 November, a little under 24 hours later. Its final customer update followed early on Saturday morning. About 430 million people use Microsoft 365 apps, and Downdetector logged reports from more than 70 countries.
The report points to a technical failure rather than a cyber incident or AI issue. A latent defect introduced in a July change caused lock contention inside a Windows process on network gateway servers, exhausting CPU and memory. The change had passed load testing, but Microsoft said its testing tended to focus on weaker hardware with fewer cores, while the problem was worse on newer, higher-core machines.
Two factors helped trigger the incident: Teams failed over to the primary gateway serving Exchange Online, while almost 14% of gateway capacity was already unavailable for an unrelated reason. At the low point, more than 80% of servers were unresponsive. A rollback was not possible because the defect had been in place for months, so engineers wrote a patch to return the code path to its previous known-good state. The preliminary report says a final one is due on 20 November.
The timeline is also worth noting without turning it into a criticism of the response. Detection flagged rising connectivity issues at 16:03 UTC, the first customer support case arrived at 16:33, engineers identified the cause at 00:30 the next day, and mitigation went in at 07:20. Roughly 15 hours passed between that first case and a fix that held. For a business depending on the service, that is a long time to operate without a clear recovery point.
Work that stopped
For healthcare, the cost of lost access was not just inconvenience. The BBC reported that Rachel Andrews, assistant manager at Orchard Medical Practice in Ipswich, said staff could not see hospital discharge summaries to check which medicines needed changing. “We cannot prescribe blindly,” she said. Mohamad Alom of Aqua Pharmacy in Ipswich said patients arrived to collect medicines, but prescriptions had not come through and the pharmacy could not email requests to surgeries. Nick Kaye, a superintendent pharmacist in Newquay, said community pharmacies generally coped but expected a backlog of emailed prescription requests.
Courts also had to improvise. At Minshull Street crown court in Manchester, staff looked for a fax machine; juries in some cases were sent home early and some prisoners were not brought to court, according to the Guardian. Police Scotland lost inbound email and warned of a backlog.
That was not the whole picture for emergency services. South Central Ambulance Service said email was affected, but its core 999 and 111 telephone systems and ambulance dispatch were not. It used temporary alternative communication measures, then stepped them down as email returned. That is a useful example of an organisation able to separate and describe its life-critical operations. The GMB union reported serious impact across ambulance and NHS trusts during the incident, but that should not be confused with a claim that this particular service had an ongoing patient backlog.
The pattern repeated well beyond the UK. Japanese government email went down and staff resorted to text messages, with the shipping firm Sagawa Express unable to take some online reservations. Government agencies in the Netherlands, Belgium and France reported problems, and a French municipality told citizens to phone instead of emailing. Australian councils, courts and schools lost inboxes, and the federal police and ASIO were affected at one level. The US House of Representatives lost a day of legislative work, and a Colorado detention facility could not take online bond payments.
Work that went wrong quietly
A message that cannot be sent is visible. A warning that never arrives may not be.
At Hotel Ling Bao, part of the Phantasialand theme park in Brühl, Germany, a fault in the pool’s automatic chlorine dosing system caused too much chlorine to be pumped into the water. The fault itself was independent of Microsoft. But the monitoring system sent an email warning, and staff did not receive it while Outlook was down. As reported by The Register, 15 guests needed medical attention, five went to hospital, and around 40 fire service personnel attended.
The lesson is not that email caused the dosing fault. It is that a safety alert relied on a channel that could fail. Matt Hull of NCC Group called the incident “a prime example of the importance of not only cyber resilience, but business resilience”. For an SME, that means asking whether any decision that protects people or operations depends on a single inbox.
Money and commitments that slipped
For two tenancies due to complete on a Friday, everything had been signed electronically. Residential lettings manager Phil Norgan, speaking to the BBC, said the parties could not prove the agreements had been signed while systems were unavailable. Some property completions also missed the Friday 4pm funds deadline.
The effects reached beyond smaller firms. Richemont delayed its half-year results and Workhorse Group moved its quarterly report. These examples show how a technology issue can become a missed commitment, not just a temporary pause in email.
Businesses reported more ordinary but still material problems too. Mark Ling of ICE Transport in Ipswich said the outage hit both his US principal office and his own operation. Earlier reporting described a fire safety adviser unable to send quotes, a cleaning supplies firm missing supplier order slots, and a 12-person Oxfordshire firm unable to access payroll data. As Jules Shorrock of the Suffolk Business Women Board put it to the BBC, the consequences were significant and businesses need a process in place.
The second wave: phishing that used the outage as its lure
From 13 November, Microsoft also reported a high-volume phishing campaign using the outage as a lure. Fake service-health emails referenced the real incident ID EX1550543 and claimed held messages would be deleted within 24 hours unless recipients clicked a link. Some used adversary-in-the-middle techniques to capture credentials and multi-factor authentication sessions.
Microsoft says it blocked the campaign’s infrastructure. Its advice is to check service incidents in the Microsoft 365 admin centre, not through links in emails. It is a reminder that disruption can create an opening for a second problem, especially when staff are expecting urgent updates.
OpenAI and Anthropic also reported problems connecting to Microsoft services during the incident. There were other technology and service disruptions in the same week, including an FAA ground stop in the US, but officials said there was no evidence they were linked. There is no need to turn separate events into one larger story.
If it had lasted a week
Dr Rory Hopcraft of the University of Plymouth described the day as a “true test” of business continuity and disaster recovery plans. The harder question is what happens when a technology service you rely on remains unavailable. For a business of 10 to 250 people, the answer often unfolds in stages.
First 48 hours: promises become hard to confirm
The first problem is not always that work stops. It is that the business cannot confirm anything with anyone.
A 40-person manufacturer may still be able to make goods, but be unable to confirm a changed delivery time or get approval for a substitute part. A family-run distributor may know what is on the lorry, but not be able to confirm an order change with the customer. A 25-person professional services firm may be able to do the work, but unable to access the latest file or confirm which version a client has approved.
Most SMEs will find that activity continues for a while. Their promises are what become uncertain: the yes, the revised time, the agreed price, the proof that something was signed.
Days three to five: the quiet losses gather
As the days pass, gaps accumulate. Invoices may not be issued. Remittances may not be matched to bills. A customer waiting for an answer may decide to pay another supplier first. A week-long disruption crosses from “we will catch up” into a payroll run: that 12-person firm which could not access payroll data for a day may face a different problem if it cannot access it for a week.
Customer trust can erode in the same quiet way. Silence is often read as indifference, and a competitor who answers the phone may win the order. Staff may work around the problem brilliantly for two days; by the second week, they may start to wonder whether the business is equipped to support them.
The decisions a director has to make
These are practical choices, not a governance exercise. Which customers need to hear from you first, and what can you honestly tell them? Do you pause new orders if you cannot confirm delivery dates? If payroll software and records are unavailable, do you arrange to pay people manually from the bank, and who checks that decision? Do you authorise spending on a tool you had not planned to buy? Who is allowed to speak for the company when the usual channels are down?
A clear answer does not remove every problem. It does stop people making conflicting promises or waiting for permission that is not coming.

What may not be recoverable
Most disruption can be worked through after services return. But some consequences cannot simply be replayed. A statutory or contractual deadline may have passed. A safety or clinical decision may have been made without information that mattered. A completed transaction may be impossible to undo. A customer may already have moved to another supplier; a client may have learned you were unavailable from somebody else.
That is worth planning for, but fear is a poor basis for a plan. Dr Alex Taylor of the University of Exeter has noted that each failure is slightly different, making resilience a continuous learning process. He also suggested bringing some data services back on premise to diversify storage. For many SMEs, the first step is not a large technology project. It is identifying where relying on one route leaves the business exposed.
What good looks like for an SME
A 200-page business continuity plan is overkill for a 15-person firm. A one-page note that people can use is more valuable. It should cover:
- The three or four activities that cannot wait a week.
- A way to contact key staff and customers that does not depend on the affected provider.
- A named decision-maker and a stand-in.
- An agreed way to pay people and receive or record payments.
- Where the essential contact details, current orders and key records can be found if the usual system is unavailable.
The South Central Ambulance Service update is a useful example of clearly stating what was affected, what was not, and which temporary measures were being used. For a small business, the same clarity can be a short note rather than a formal manual.
This is a different question from the first-hours response covered in our earlier article, Supplier Risk Management for Small Business: What the Microsoft 365 Outage Reveals. Here, the focus is how losses build if the disruption lasts. A practical business resilience approach can help you make that distinction without adding unnecessary process.

Frequently asked questions
Was the Microsoft 365 outage caused by a cyberattack?
No. Microsoft’s preliminary report described a technical defect in network gateway servers and said the outage was not related to cyber activity or AI.
What did Microsoft say caused the outage?
A latent defect introduced in a July 2026 change led to lock contention inside an internal Windows process, exhausting CPU and memory on gateway servers. The report says the impact combined with Teams failing over to a primary gateway and reduced server capacity.
What should a small business do if its cloud email is unavailable?
Start by deciding which work cannot wait, how staff and customers can be reached by another route, and who can make decisions. Then agree how key records, payments and commitments will be managed until access returns. Keep the plan short enough for people to use.
Does an SME need a full business continuity manual?
Not necessarily. For a small firm, a one-page plan covering essential activities, contacts, authority and payment arrangements may be more useful than a long document. The right level depends on what the business needs to keep doing.
If you want to see where a longer outage would put pressure on your organisation, a Rapid Resilience Diagnostic can help identify practical priorities without adding pointless bureaucracy.
Philip Alcock, CEO, Real Risk Solutions
philipalcock@realrisksolutions.com | +44 (0)1223 230205
