Board Communications During Ransomware: A Practical Guide
On Thursday evening, 12 November, UK minister Ian Murray sent himself a test email. It reached him at about 2pm on Friday. The acknowledgement then bounced back. Imagine that email had been a request for a decision. For most of a working day, the sender would have had no way of knowing whether silence meant agreement, a delay at the other end, or a message still sitting in a queue.
That uncertainty mattered across hospitals, government departments and financial institutions. Outlook was where the failure was most visible. The wider exposure was how much work depended on the same Microsoft environment: finding the duty officer, inviting a partner to a Teams call, sharing a briefing and establishing who had authorised the next step.
As AI strategist Walter Pasquarelli told the BBC, for governments, businesses and critical infrastructure the outage had “put their vulnerabilities and dependence in plain sight”. For a crisis team, that dependence becomes visible when the same system carries the invitation, the briefing and the decision everyone is waiting for.
What went wrong inside Microsoft 365?
Microsoft’s preliminary investigation traced the failure to a network gateway, the infrastructure that accepts connections and routes them to services such as Exchange Online. A software change deployed in July had introduced a defect that only surfaced under particular traffic conditions. It had been running for more than three months. Microsoft’s tests had tended to use less powerful hardware, while the fault was more pronounced on newer servers with more processor cores.
Two things brought it to the surface. About 15 hours before the outage, an automatic switch moved Teams traffic onto the gateway already serving Exchange, changing the volume and mix of requests. Separately, nearly 14% of the gateway’s capacity was unavailable following an unrelated issue. As demand rose, processes on the remaining servers began waiting on one another. Requests accumulated and servers stopped responding.
The system redirected connections to healthy servers, which then encountered the same fault. At the worst point, more than 80% of the affected gateway fleet was unresponsive. Restarting servers brought only temporary relief because waiting traffic immediately overwhelmed them again. Engineers had to take groups of machines offline, reverse the defective change and restore capacity in stages. Microsoft recorded active impact from 16:20 GMT on Thursday 12 November to 16:10 on Friday 13 November. Processing the mail backlog continued after that.
What happened to Teams chat and conferencing?
Teams could divert traffic away from the affected gateway. Exchange Online could not do so at the scale required. According to Microsoft, Teams chat and meetings worked for most of the incident, although functions that relied on Exchange remained exposed. Experience varied by region, time and how people connected.
Consider a council incident lead who starts a Teams meeting using a link the core team already knows. Those colleagues can talk. The NHS representative invited by email ten minutes ago may have no idea the meeting is taking place. A director can approve a public statement in the chat while the communications officer waits for approval in Outlook. Both teams are working. They have different accounts of what has been decided.
This is the weakness in an instruction as simple as “switch to Teams if email goes down”. Can the incident lead reach everyone who needs to switch? Can someone joining the next shift find the latest decision? If chat or conferencing also becomes unavailable, does everyone already know where to reconvene?
That is where out-of-band (OOB) communications come in. YUDU Sentinel gives the organisation its own Single-tenant Out-of-Band Communications platform, hosted independently of Microsoft 365 and Azure. Responders can sign in without relying on corporate single sign-on through Entra ID. The duty lead can trigger the switch as soon as a critical group cannot be reached, even while Teams is still working for colleagues elsewhere.
How communications disruption can spread across critical sectors
Critical national infrastructure: coordinating a second incident
The UK Government reported no impact on water or wastewater treatment processes and no known disruption to mobile, broadband or landline connectivity. Transport services and borders remained operational. That distinction matters. An email failure at a utility is a different event from a failure of the infrastructure supplying the public.
Now consider a water main bursting during the same outage. The control room can dispatch engineers over its usual operational channels. But who tells the council which roads will close, briefs the customer helpline and confirms that the next shift has the revised repair estimate? If those exchanges rely on email and calendar invitations, crews may be making progress while the council and the public are still working to yesterday’s information. With Sentinel, the duty manager can send the revised estimate by two-way SMS to a predefined group of site leads and council contacts. Acknowledgements show whom to follow up by phone. That gives the control room a specific answer when the next shift asks whether the council has been told.
Government: agreeing what the public should be told
In the UK, officials at the Department for Digital, Culture, Media and Sport coordinated the response across departments with the Cabinet Office. Ministers also acknowledged organisational disruption in local government. Overseas, the BBC reported that the US House of Representatives lost email and court systems in Singapore experienced disruption.
For a council, even publishing a short service update can require three people: a service owner to confirm the facts, a duty director to approve the wording and a communications officer to publish it. If the approval is delayed in Outlook, does the officer wait, call the director or use an earlier draft? In a preconfigured Sentinel Space, the service owner can post the facts in Secure Chat and the duty director can approve the exact wording there. If a judgement call needs a conversation, they can open a Video Crisis Room. The communications officer can then post the agreed statement and the time of the next briefing in the same Space, ready for the incoming team.
Healthcare: reaching the next shift and the next provider
NHS staff reported difficulty accessing email and exchanging information across hospitals, GP practices and community services. Some used paper records. A&E departments remained open, and NHS services activated local contingency arrangements. The coordination challenge extended beyond an individual ward to organisations that routinely exchange information about the same patient.
Take a hospital discharge team waiting for a community provider to confirm a handover. A message in a sent folder gives the team little basis for deciding whether someone has taken responsibility. A Sentinel alert can ask both duty leads to confirm who owns the outstanding handover. If the procedure has been downloaded to the Sentinel mobile document app in advance, the incoming lead can consult it even without access to SharePoint. Patient information can then be exchanged through the approved clinical route. When the next shift arrives, it needs to know which handovers still require action, especially as delayed emails begin to appear.
Finance: keeping decisions authorised and attributable
On the morning Aviva was due to report its results, it issued alternative dial-in numbers for its earnings calls and used Bluetooth to transfer recordings. The insurer had exercised a comparable outage about two weeks earlier. Euronext moved enquiries from listed companies to the phone and verified callers’ identities. Reuters also reported that HSBC and Barclays activated incident teams, even though most client services continued.
Euronext’s experience illustrates why simply exchanging mobile numbers leaves part of the problem unsolved. When a caller asks a team to act on sensitive information, staff still need to establish who is speaking and what that person is authorised to request. A bank’s incident team also needs a record of the decision, particularly if an instruction later arrives by email and appears to be new. A bank using Sentinel can limit access to the relevant response team and have the authorised officer record the decision there. Its tamper-evident audit log gives reviewers a timestamped record of activity to reconcile with messages that arrived after the decision.
A brief timeline of earlier outages
The November outage brought several dependencies into view at once. Earlier incidents show how losing even one link, such as access to chat or a shared calendar, can delay a response.
-
26 January 2024 – Teams access. A networking issue disrupted Microsoft Teams. Some users could not access the service while others had limited functionality. Microsoft moved traffic to backup systems. A colleague already in Teams could therefore have a different picture of who was reachable.
-
19 July 2024 – Windows devices. A faulty CrowdStrike update affected an estimated 8.5 million Windows devices. This third-party incident exposed a separate dependency for crisis teams whose only way to reach a healthy cloud service was a laptop that would no longer start.
-
12 September 2024 – network access to Microsoft 365. Some US users had difficulty reaching Microsoft services through AT&T networks. For an incident lead, that distinction matters: successfully opening Microsoft 365 at head office tells you little about whether a site using a different network can receive your instructions.
-
25 November 2024 – Exchange Online and Teams calendars. A Microsoft 365 disruption affected email and Teams calendar functionality together. For teams arranging a crisis call, the invitation itself became a point of failure.
-
1 March 2025 – Outlook and Microsoft 365. Thousands of outage reports centred on Outlook as Microsoft investigated problems across several services. The disruption fell on a Saturday, when reaching a small on-call team can matter more than reaching the wider workforce.
-
9–10 July 2025 – Outlook across devices. Users had difficulty opening their mailboxes through Outlook.com and the mobile and desktop apps. Switching from a laptop to a phone still left users dependent on the same mailbox.
-
29–30 October 2025 – Azure Front Door. A configuration change to Azure Front Door disrupted connections to multiple services for eight hours and 24 minutes. Some customers also had difficulty reaching the Azure management portal.
-
22–23 January 2026 – North American Microsoft 365 services. Outlook was among the services disrupted when part of Microsoft’s North American infrastructure stopped processing traffic as expected. Microsoft redirected traffic and confirmed recovery on 23 January. A global incident team could still need a fallback to include colleagues in one affected region.
-
31 August 2026 – Exchange Online and authentication. Microsoft identified a fault in an authentication configuration used by several Microsoft 365 services. Users could have difficulty connecting to Exchange Online or using its features. It is a reason to test whether responders can sign in to their backup channel independently.
-
12–13 November 2026 – Exchange Online and connected services. Microsoft recorded 23 hours and 50 minutes of active impact. Teams chat and meetings worked for much of that period, but email delays left responders without a reliable way to distribute fresh invitations and confirm who had received an update.
What does your crisis team need to do without Microsoft 365?
A practical way to use Sentinel’s three levels of operational resilience is to walk through a shift change during a prolonged outage. The outgoing lead has three jobs: reach their replacement, brief them on the decisions already taken and put the information they need for the next shift in their hands.
-
Level 1 – OOB Communication. The duty lead sends a two-way SMS to the incoming supervisors: “Email is delayed. Please confirm who is covering each site until 18:00.” They can see which sites have acknowledged and use voice to follow up with anyone missing. A working mobile network becomes a usable response channel because the contact groups and instructions are already in place.
-
Level 2 – OOB Collaboration. Each supervisor reports outstanding issues in Secure Chat. Where two sites need the same contractor, their leads can convene in a Video Crisis Room, agree which job takes priority and record who will contact the contractor. The next shift can see the decision and its owner without searching an inbox or chasing a meeting invitation.
-
Level 3 – OOB Service Continuity. The incoming supervisor also needs the duty rota and contractor call-out schedule. Where those records are synchronised into a designated Sentinel Space through a configured API, the team can consult the last successfully synchronised copies. The supervisor can check when the rota was current and confirm any subsequent changes with the outgoing lead.
Try that handover as part of your next Microsoft Teams fallback exercise. Give the duty lead ten minutes to reach an external partner, bring the decision-makers together and issue an approved update without Outlook or Teams.
Then hand the incident to someone who missed the meeting. Can they tell you what was decided, who still needs to act and when the next update is due?
18 Nov 2026