
🔔 Subscribe to ShipUniverse Weekly →
SHIPUNIVERSE / TECH · NOVEMBER 16, 2026 · FLEET RESILIENCE
One Microsoft outage disrupted shipping communications. 10 backups every fleet needs.
A vessel can be ready to sail while an instruction sits in an email queue nobody can open. That's what makes last week's Microsoft outage worth examining. Maersk warned of delayed customer emails and case creation; Lloyd's Register sent urgent survey requests to the telephone. The ships weren't reported to have lost their safety radios. The weak link was shore-side business communication, and a second copy of the same cloud app wouldn't necessarily have helped.
What actually happened
November 12–13Emails were not reliably reaching the case queue
On November 13, Maersk reported Exchange Online degradation affecting receipt of external emails and creation of cases in its Customer Experience platform. It directed urgent enquiries to telephone and live chat and warned about suspicious outage-related messages.
Urgent work moved to the phone
LR reported disruption associated with the global Microsoft outage. Customers needing a survey within 24 hours were told to call its technical care and planning teams. This shows why a published, independent route to a human matters.
Scope matters. Neither advisory demonstrates that navigation, shipboard machinery or GMDSS distress equipment stopped working. Maersk's separate November 11 legacy-system cybersecurity disclosure was unrelated to the Microsoft service outage. No verified fleetwide monetary loss has been published for this disruption.
How one missing inbox becomes an operational problem
Potential failure chainEmail is delayed
A terminal, agent, supplier or charterer sends a time-sensitive instruction.
No case gets opened
The service desk or shared inbox does not receive or route the instruction.
Approval stalls
Crew, spare, customs, cargo or survey arrangements await confirmation.
A delay may cost money
Extra handling is certain; vessel delay occurs only if the specific operation is affected.
This is a risk pathway, not a claim that all four steps happened at Maersk. The distinction is important when calculating financial exposure: do not convert every lost email hour into a full vessel off-hire day.
10 backup systems worth having before the next outage
Priority order for a fleetOrder reflects speed to deploy and ability to protect urgent operations, not a ranking of product brands. “Independent” means it survives the failure being tested, including identity, network and supporting services.
Independent voice and SMS escalation
Keep external-facing landline/mobile numbers, a duty roster and approved escalation contacts. Do not rely only on Teams Phone or a directory unavailable during the same outage.
Test: Call the duty officer and confirm a named human answers within the agreed response time.
True secondary email service
Maintain preconfigured addresses on an independent provider, with separate credentials, recovery contacts and authentication. A second mailbox in the same Microsoft tenant is not independent.
Test: Send and receive externally while deliberately denying access to the primary email platform.
Separate operational satellite path
For ships, maintain an approved business-communications fallback where needed, with a different failure path from the primary link. Check terminal, airtime, routing, coverage and power independence.
Test: Disable the primary data connection and transmit a short operational message shoreward.
Offline crisis contact pack
Maintain encrypted offline copies and tightly controlled printed essentials: agents, ports, charterers, ship contacts, class, flag, vendors and authority escalation numbers.
Test: Find the right contact without Microsoft 365, an active company login or office internet.
Independent case-intake queue
Provide a telephone/chat/manual queue that can capture case IDs and time stamps without Exchange-triggered workflows; reconcile the backlog after restoration.
Test: Create, assign and close a priority case while email automation is unavailable.
Second office internet path
Use a genuinely diverse ISP or wireless link for key response centres. It will not fix a cloud-provider outage, but it prevents a local circuit failure from blocking the backups too.
Test: Pull the primary link and verify the alternative reaches all emergency channels.
Recoverable document copy outside the cloud
Retain controlled, restorable copies of port instructions, ship contacts, critical certificates, charter orders and incident procedures outside the failed platform's trust boundary.
Test: Retrieve a current essential document without SharePoint or OneDrive access.
Break-glass identity and MFA
Have preapproved emergency access for independent systems. Protect credentials offline, use separate authentication factors and audit all activation; avoid a fallback that still requires the failed SSO.
Test: Access backup mail and the case queue without the primary identity provider.
Manual EDI, port and cargo fallback
Agree alternative procedures for time-critical bookings, arrival information, clearance and delivery orders. Use approved channels and retain evidence, not unverified personal messaging.
Test: Complete one simulated priority transaction and reconcile it into the primary system later.
Independent status and notification service
Publish outage instructions through a separately hosted status page or trusted alternate channel. Signpost verified contacts so spoofed “urgent payment” requests are easier to challenge.
Test: Issue and authenticate an external advisory with the normal mail and website staff login unavailable.
Five backups that look independent but aren't
Common-mode traps| Proposed backup | Hidden shared dependency | Better acceptance test |
|---|---|---|
| Another Outlook mailbox | Same Exchange Online service or tenant | Send through a separate provider with independent login and mail routing. |
| Teams calling instead of email | Shared identity, Microsoft infrastructure or endpoint | Call through a separately operated mobile/voice line. |
| Alternate email domain | Both domains' MX records still point to the same service | Configure and test independent inbound/outbound mail before an outage; DNS changes aren't instant. |
| Backup documents in OneDrive | Same cloud, credentials or access-control failure | Restore and open essential records from a separate controlled location. |
| Second satellite antenna | Same modem, network operator, power supply or shore gateway | Prove an end-to-end link under loss of the primary terminal and service. |
Safety boundary: SOLAS/GMDSS distress and safety systems have their own carriage and operational requirements. A business satellite subscription, WhatsApp group or backup mailbox is not a substitute for mandated GMDSS capability.
The first hour after cloud email goes dark
Exercise this quarterlyDeclare the incident
Confirm the failure, activate independent voice, identify priority movements and appoint one incident owner. Do not assume a cyberattack without evidence.
Move urgent work
Notify agents, terminals, customers and ships via verified channels. Log cases manually, authenticate payment changes and preserve timestamps.
Reconcile everything
Import alternate-channel messages, check for duplicate or missing actions, confirm order versions and document what failed in the workaround.
Governance: IMO's maritime cyber-risk guidance calls for identifying critical systems, protecting operations, detecting disruption, responding and recovering. That supports a business-continuity exercise; it does not prescribe ten particular products or make every optional backup mandatory.
What is one broken communications channel worth to your fleet?
Model extra handling and the chance of genuine operational delays. Then compare the expected reduction in those losses with backup running costs. These are hypothetical planning assumptions, not Maersk's costs or measured industry outage frequency.
Affected vessels = fleet × affected share (fractional fleet equivalents are used for expected-value modelling). Extra labor cost/event = affected vessels × handling hours × staff-hour cost. Expected delay cost/event = affected vessels × delay probability × incremental cost when a delay occurs. Annual avoided loss = (labor cost + expected delay cost) × disruptions per year × preventable share. Annual net = avoided loss − annual per-vessel backup costs − central annual cost. First-year net further subtracts one-time setup; three-year net assumes steady annual values and counts setup once. Break-even events/year = annual running cost ÷ avoidable loss per event. Outputs are scenario expectations, not audited cash savings, actual off-hire losses, service-level promises or cyber-risk probabilities. Operational resilience and safety may justify backup controls even if this financial model is negative.
We welcome your feedback, suggestions, corrections, and ideas for enhancements.
Please click here to get in touch