Dear {{ first_name | reader }},
At some point on Tuesday 19 November 2024, tens of thousands of organisations discovered the same thing at the same moment: they could not send email. Teams was down. SharePoint was unreachable. And the channel they had nominated as a fallback for exactly this kind of situation was running on the same infrastructure that had just failed.
The Microsoft 365 global outage lasted around nine hours. For most organisations, it passed without lasting damage. But a subset — those in the middle of an active crisis, those whose stakeholders were already watchful, those whose communications function ran almost entirely through the Microsoft stack — found themselves effectively silent at precisely the moment they most needed to speak.
That is what this edition is about.
Not the outage itself — Microsoft's post-incident review covers the technical sequence well enough. This is about what the outage revealed: a structural blind spot that most crisis communication plans carry quietly, waiting for the wrong moment to surface.
This newsletter includes my full case study for download a well as a link to my guide called the Fallback Protocol.
Happy reading.
WAG THE DOG NEWSLETTER | WEEK 47, 2026
FIVE THINGS TO TAKE FROM THIS
Most crisis communication plans nominate a fallback channel without checking whether it shares infrastructure with the primary. When Microsoft 365 failed, organisations that had listed Teams or Outlook as their backup discovered they had not actually created redundancy — they had created an illusion of it.
The outage exposed a scale-of-dependency problem that most organisations have never mapped. Email, calendar, document storage, internal messaging, and external stakeholder communication often run through a single vendor. A single point of failure takes all of them down together.
Nine hours of silence is long enough to damage a narrative you do not control. Stakeholders who cannot reach you do not wait patiently — they form their own picture from whatever information is available, and that picture is rarely flattering.
The window in which an organisation can shape the narrative opens immediately and closes fast. Every hour without a credible signal from an organisation is an hour the story is being written by someone else — a journalist, a competitor, an anxious employee posting on LinkedIn.
A fallback protocol is only useful if it has been tested when the tools it depends on are unavailable. Paper-based, phone-tree-based, and out-of-band communication systems atrophy without practice. The organisations that coped best in November 2024 were those that had rehearsed communication without digital infrastructure, not just communication through it.
Table of Contents
What happened
On 19 November 2024, a configuration change deployed to Microsoft's Azure Active Directory authentication infrastructure caused a cascading failure across multiple Microsoft 365 services. Exchange Online, Microsoft Teams, SharePoint Online, and OneDrive were all affected to varying degrees.
The incident began around 03:00 UTC and was not fully resolved until approximately 12:00 UTC — nine hours in which a significant portion of the global workforce could not access email, could not reach internal messaging, and could not retrieve documents stored in SharePoint.
Microsoft's post-incident review, published in the days following, was detailed and technically honest. The root cause was clearly identified, the fix was documented, and the review acknowledged the breadth of the impact. For a technical post-mortem, it did what post-mortems are supposed to do.
What it did not address — because it was not written to address it — was what the outage revealed about how organisations communicate when the infrastructure they depend on disappears.
The dependency trap
Most large organisations run their internal and external communications through a small number of platforms. Microsoft 365 is the dominant one. Email through Exchange Online, messaging through Teams, document storage and collaboration through SharePoint, calendar coordination through Outlook — for many organisations, all of this sits on a single vendor's infrastructure.
The dependency is not irrational. Consolidation reduces cost, simplifies administration, and makes cross-platform collaboration easier. These are real advantages. The problem is not the consolidation itself; it is the assumption, rarely examined, that the consolidated platform is effectively indestructible.
When organisations nominate a fallback communication channel in their crisis plans, they typically mean a secondary channel within the same ecosystem. The backup to email is Teams. The backup to Teams is a group chat. The backup to SharePoint is a shared OneDrive folder. None of these is a fallback — they are all expressions of the same dependency. When the infrastructure fails, they fail together.
The November 2024 outage made this visible in a way that most tabletop exercises do not. It is easy, in a simulation, to assume that communication tools are available. It takes an actual infrastructure failure to discover that your fallback protocol was never tested against the conditions it was designed for.
The channel problem
Out-of-band communication — channels that operate independently of your primary infrastructure — is not a new concept. Military and emergency management doctrine has required it for decades. The principle is simple: if your primary communication system fails, you need a backup that does not share its failure mode.
In practice, most corporate crisis plans treat out-of-band communication as an afterthought. Phone trees are listed, then not maintained. Personal mobile numbers are collected, then not updated. Signal groups are created for a specific exercise, then forgotten. The result is that when an infrastructure failure occurs, the out-of-band options either do not exist or exist only on paper.
The organisations that maintained communication during the November 2024 outage shared a common characteristic: they had genuine alternatives that did not depend on Microsoft infrastructure. Some used SMS broadcast systems. Some had maintained WhatsApp or Signal groups for crisis communication specifically. Some fell back to direct phone calls. A small number had invested in dedicated crisis communication platforms that run on separate infrastructure entirely.
None of these alternatives is perfect. SMS is limited in what it can carry. Personal messaging apps create data governance complications. Phone calls do not scale. But imperfect out-of-band communication is vastly preferable to no communication at all.
Download the full case study below.
The timing problem
Crisis communication has a timing dimension that infrastructure failures make brutally visible. The first hour after a significant incident is the period in which an organisation has the most influence over the narrative that will form around it. Stakeholders are still forming their understanding. Media coverage is still thin. The information vacuum is not yet filled.
Every hour that passes without a credible signal from the affected organisation is an hour in which that vacuum fills with something else: speculation, rumour, competitor commentary, social media posts from frustrated employees or customers. By the time communication infrastructure is restored, the narrative may already be set.
Nine hours is a long time. For an organisation already under scrutiny — facing a regulatory inquiry, managing a product recall, in the middle of a labour dispute — nine hours of silence is not a neutral absence. It is interpreted. And the interpretation is rarely charitable.
The practical implication is that the trigger for activating out-of-band communication cannot be the restoration of normal infrastructure. It has to be the failure itself. If your crisis communication protocol requires you to send an email to your crisis team to tell them the email system is down, the protocol has failed before it started.
The narrative window
There is a concept in crisis communication practice — not always named, but widely understood — that relates to the period in which an organisation can shape public understanding of an event. Call it the narrative window. It opens at the moment an incident becomes visible to external stakeholders and closes, gradually, as coverage and commentary solidify into a received account.
Infrastructure failures affect the narrative window in two ways. They delay the organisation's ability to communicate, which narrows the window from the inside. And they tend to generate immediate external commentary from affected users — on social media, in industry channels, in direct conversations — which narrows the window from the outside.
The organisations that navigated November 2024 most effectively did not wait for Microsoft to restore service before communicating. They used whatever channels remained available — website updates, SMS, pre-prepared social media posts, direct calls to key stakeholders — to establish a presence in the narrative before the narrative formed without them. The message did not need to be detailed. It needed to exist.
Acknowledge. Signal. Redirect. Those three steps — acknowledging that something has happened, signalling that you are aware and responding, redirecting stakeholders to whatever alternative channel is available — do not require a functioning email system. They require preparation.
The PIR asymmetry
Post-incident reviews in infrastructure failures tend to focus on the technical sequence: what failed, why it failed, what fixed it, what will prevent recurrence. This is appropriate and necessary. The technical learning matters.
What is less systematically captured is the communication sequence: who learned about the incident and when, what communications were attempted and failed, what stakeholder relationships were strained by the silence, what narrative formed in the absence of organisational communication, and what it cost to correct it.
This asymmetry is not unique to infrastructure failures, but it is particularly pronounced in them. When the failure is technical, the post-incident review is typically owned by IT or operations. The communication dimension is treated as secondary, or not treated at all. The result is that organisations learn how to prevent the technical failure from recurring without learning how to handle the communication failure that accompanied it.
A complete post-incident review of any significant infrastructure outage should include a communication audit: a timeline of when key stakeholders were reached, what channels succeeded and failed, what was communicated and when, and what the narrative looked like at each stage. Without this, the communication lessons do not survive the incident.
Three things to do now
None of this requires a large investment. It requires specificity.
First, audit your fallback channels against infrastructure independence.
For each channel listed as a fallback in your crisis communication plan, identify what infrastructure it depends on. If the answer is the same infrastructure as your primary channel, it is not a fallback. It is a redundant expression of the same dependency.
Second, test your out-of-band protocol under realistic conditions.
This means running a communication exercise in which Microsoft 365 — or whatever your primary platform is — is assumed to be unavailable. Not degraded. Unavailable. Can your crisis team convene? Can they reach key stakeholders? Can they issue a holding statement? If the answer to any of these is no, you have identified a gap before an outage does it for you.
Third, pre-draft your infrastructure failure messaging.
The holding statement for a major platform outage — acknowledging the situation, signalling your awareness, redirecting stakeholders to alternative contact points — should exist before the outage occurs. Drafting it in the middle of an infrastructure failure, using tools that may also be affected, is not a process. It is improvisation under pressure, which is exactly when improvisation performs worst.
Templates, guides and simulation scenarios

Sponsor
Lowest error rate on real dictation. Ahead of Google, OpenAI
Speech models love clean lab audio. Your real prompts don't sound like that. So Wispr Flow tested its in-house speech model, Canto, on 10 hours of real dictations from 2,300+ speakers.
The result: the lowest word error rate of every model tested, including models from Google, OpenAI, AssemblyAI and Deepgram. It also tied for the lowest error rate on LibriSpeech, a standard public benchmark.
For you, that means fewer fixes and longer, richer prompts. Speak your context, constraints and examples into ChatGPT or Claude, and Wispr Flow turns it into clean text in any app. Works on Mac, Windows, iPhone and Android.
THE RADAR: WHAT I’M TRACKING
[GUIDANCE]
Are we ready? Building a culture of resilience[TOOL]
Best AI Agent for Non Techies[ARTICLE]
Protecting information flows in times of crisis[REPORT]
The Distance Trap – Alder reputational risk research 2026
LET’S MEET
🇪🇬 NOV 24-26 | CAIRO
2027
🇰🇼 JAN 25-26 | KUWAIT CITY
🇦🇪 MAY 10-14 | ABU DHABI
How valuable was the strategic insight in this edition?
Transparency & Disclosures
AI Transparency: In alignment with EU AI Act requirements, please note that AI technology was used in the research, drafting, and/or image generation for this edition. All strategic analysis, professional opinions, and final editorial oversight are conducted exclusively by the author. Affiliate Disclosure: Some links in this briefing may be affiliate links. I only recommend tools and services I use personally or have vetted for professional efficacy. Professional Advice: This newsletter is for educational and informational purposes only and does not constitute legal or professional crisis management advice. © 2026 RiskComms FZCO. All rights reserved.


