Insight | Operating Resilience
The Illusion of Digital Resilience
What This Week's Tech Shifts Really Mean for Leaders
Resilience tests check whether your systems recover. Almost nobody tests whether your decisions do.
This week, a Microsoft 365 outage took Outlook, Teams and Copilot down across dozens of countries for more than 18 hours, and many organizations found out how much of their decision-making runs through a mailbox. Most have confused software workflows with operating resilience, and they are about to add AI agents on top.
This Insight sets out a quarterly 4-hour lights-out drill and a five-point scorecard that shows where your decisions stall when the software stops.
Three Signals, One Blind Spot
Three events defined the technology week:
- Compute is physical. SMIC's co-CEO said the chip supply chain for AI will stay tight through 2027. AI capacity is a constrained utility, not an endless cloud.
- AI is going public. Anthropic has set the price range for its Nasdaq listing, with trading expected to start on November 19. Generative AI is moving from subsidized experiments to enterprise unit economics.
- The spine is brittle. On November 12, Exchange Online started failing. Outlook, Teams, Copilot and a long list of dependent services went with it, in at least 74 countries. Eighteen hours later, Microsoft was still reporting recovery "progress".
The thread connecting them is simple: technology deployment is outrunning organizational architecture.
A Single Point of Failure in Modern Software
Companies are wiring AI agents into daily operations while their basic coordination spine (email, chat, calendar) has no fallback. During the outage, Copilot Chat could not reach email, calendar or files, because it depends on the same mailbox layer.
When your AI assistant goes dark because the mail server does, you do not have an autonomous enterprise. You have a single point of failure dressed up in modern software.
You cannot buy your way out of that with a better SaaS contract. The fix is structural: decision-making that does not depend on one vendor's uptime.
The Recommendation: A Quarterly 4-Hour Lights-Out Drill
Before deploying the next AI workflow, adopt one governance rule. Once a quarter, take one core business unit offline from email, chat and cloud AI tools for half a day. Announce it in advance, and keep a fallback for client-facing and regulatory duties. Then watch three things:
- Where do decisions stall? If approvals stop because Teams is offline, your delegation of authority lives in software permissions, not in clear management accountability.
- Who becomes the bottleneck? You will see the hidden "human middleware": managers who keep broken processes running by stitching them together with messages.
- Can operations run out of band? Resilient organizations keep a simple, independent command line (a phone tree, a short list of who decides what, manual overrides) so execution continues when the cloud goes dark.
Lights-Out Drill Scorecard
Five Measures for the Four Hours
Record these during the drill. Each one marked red is a decision that lives in software rather than with a person.
- Decisions deferred: how many decisions were postponed "until the systems are back"? Red if any of them sat within the unit's own authority.
- Time to decision: for the three most frequent approvals, how long did they take compared with a normal day? Red if more than double.
- Who decides what: could each team member name who approves what, without opening a system? Red if the answer depends on a workflow tool.
- Bottleneck people: who received most of the calls and visits? Red if one or two names carried most of them.
- Out-of-band command line: did the phone tree or contact list reach every decision-maker within 15 minutes? Red if it did not exist or failed.
Two or more reds: your delegation of authority depends on software. Fix that before the next AI deployment.
If four hours without software paralyzes your leadership team, AI agents will not fix it. They will scale it.
Conclusion: How Resilient Is Your Operating Architecture?
Most leadership teams find their single points of failure during a live crisis: a vendor outage, rapid growth, a post-merger integration. Each one exposes the same thing, blurred decision rights and hidden coordination drag.
Better to find them on your own schedule. Start with the 10-Minute Decision Drift Diagnostic™, which shows where your organization loses speed and clarity without switching anything off, or book a confidential Clarity Session for a top-down review of your setup.