On November 12, 2026, a software defect that had sat quietly inside Microsoft's network gateway servers since July finally met the right conditions. Within two hours, more than 80% of those servers stopped responding, and Outlook users in 74 countries lost access to their mailboxes for most of a day.
For a risk leader, the Microsoft 365 outage raises a sharper question about your own supplier base.
How many of your critical vendors went quiet at the same moment, and could your team have pulled up that list the moment the outage hit?
The November 2026 Microsoft 365 outage began on November 12, when a dormant July software defect overwhelmed Microsoft's network gateway servers. Outlook, Exchange Online, Teams, and Copilot failed for users in 74 countries. Core service returned after about 24 hours, with full restoration on November 14.
Microsoft traced the outage to a defect in a July update to its network gateway servers. A routine traffic failover and capacity already lost the day before pushed those servers past the point where the bug took hold, and each failing server passed its load to the next healthy one. Microsoft said it found no sign of a cyberattack.
The disruption reached well beyond Microsoft's own customers. Shipping and stock exchange operators warned their customers about communication issues, some US court cases slipped, and UK pharmacies expected slower prescription processing.
Redundancy that shares a hidden dependency with your primary system fails at the same moment as your primary system. Many organizations hit on November 12 were somebody else's critical vendor, so their customers absorbed a fourth-party disruption they never contracted for.
Three gaps tend to surface after an event like this, and each one shows up in vendor programs at large enterprises far more often than it shows up in audit findings.
Your own contract with Microsoft sits in plain view. What usually stays invisible is how many of your Tier 1 vendors run email, ticketing, and client communication on the same provider, because standard questionnaires rarely ask vendors to disclose the platforms they depend on.
When your primary logistics provider and its backup both route customer updates through one email platform, you have a single point of failure on paper you believed was redundant.
Microsoft's own testing missed the defect because lab hardware differed from production hardware (The Register). Your vendors' annual failover tests can share that blind spot when they run at low volume or leave key dependencies out of scope.
A certificate or a test date tells you a test happened. You also need to know what load it ran under, which services it covered, and whether recovery times matched the targets in your contract.
Most contracts require vendors to report incidents within a set window, and most of those notices travel by email. On November 12, the notice channel and the failed service were often the same thing.
If a vendor's only route to you runs through the platform that just went down, your notification clause offers little protection during the event itself.
Regulators reached the same conclusion before this outage. Since July 13, 2026, UK financial regulators have directly overseen Microsoft Ireland Operations Limited as one of four designated critical third parties (Bank of England), and DORA applies similar oversight to critical ICT providers in the EU. Under both regimes, financial firms still own the resilience of their vendor relationships (PwC Legal).
Regulatory oversight of Microsoft leaves your accountability for vendor resilience exactly where it was: with your organization. Your supervisors will still ask how you identified, assessed, and monitored the vendors that depend on it.
Five steps cover most of the ground, starting with visibility and ending with accountability. Each one comes with a metric you can report to your risk committee and a fix for when progress stalls.
Add a short set of questions to Tier 1 intake and reassessments: which cloud, email, collaboration, and identity providers the vendor relies on, and which of them support the service it delivers to you. Combine the answers with what due diligence finds in public sources, then group vendors by shared provider.
Measure: The share of Tier 1 vendors with disclosed fourth parties, and the number of providers shared by three or more critical vendors
If it stalls: Vendors often skip open-ended questions, so offer a fixed list of common providers to tick
For each critical service with a backup vendor, compare the fourth parties of the primary and the backup side by side. Where they share a provider, treat the pair as one point of failure until you find an alternative or record a formal risk acceptance.
Measure: The share of critical services whose backup vendor shares no core provider with the primary
If it stalls: Where no independent alternative exists, document a manual fallback and assign an owner to it
Request the vendor's latest continuity or failover test report alongside any certificate. Check the test date, the load it ran under, the services in scope, and the recovery time against the targets in your contract.
Measure: The share of Tier 1 vendors with test evidence less than 12 months old that names load, scope, and recovery time
If it stalls: Raise a finding with a named owner and target date, and add the evidence requirement to the next contract renewal
Annual reviews miss outages, ownership changes, and adverse news that land in the months between them. Set monitoring for the events that matter to each engagement, and route alerts to the person who owns that vendor relationship.
Measure: The time from a vendor incident going public to your team logging it
If it stalls: A flood of alerts usually means the monitoring scope is too broad, so narrow it to the risk domains each engagement depends on
Collect a phone number and a secondary channel for each critical vendor's incident lead, and write that channel into the notification clause of your contract. Test the contacts once a year.
Measure: The share of Tier 1 vendors with a tested secondary contact
If it stalls: Start with the vendors that share your most concentrated fourth party
Pro tip: Run a tabletop exercise built on a hypothetical scenario where your primary email platform stays down for 24 hours. List which Tier 1 vendors would also go silent, and you will usually find concentration that your vendor inventory never flagged.
Platform software has no control over a cloud provider's uptime, so ComplyScore® focuses on the parts your team does control. If you have tried mapping vendor dependencies in spreadsheets and still cannot see which fourth parties your critical suppliers share, the wall you have hit is scale.
ComplyScore®, the AI-powered risk and compliance management platform from Atlas Systems, supports the steps above at the scale of a large supplier base:
|
Step |
How ComplyScore® supports it |
|
Map shared dependencies |
Every due diligence report lists the supplier's disclosed fourth parties, and a supply chain map with concentration analysis flags the providers several suppliers share |
|
Check backup independence |
Concentration analysis shows where a primary and a backup supplier depend on the same provider |
|
Ask for real continuity evidence |
The platform recommends resilience controls for each engagement, domain owners refine them, and AI pre-fills supplier answers with a confidence level and the source page |
|
Monitor between assessments |
You choose the risk events to track for each supplier, and AI agents triage alerts and chase pending evidence |
|
Close the gaps |
AI marks each control as satisfied or not satisfied, drafts findings and remediations, and risk acceptance needs sign-off from a named owner |
Controls a supplier has already attested carry over to each new engagement, so you assess a shared vendor once instead of once per business unit.
Enviri, a 170-year-old enterprise, runs a vendor program of 45,000+ vendors across 30+ countries and three ERPs on ComplyScore®. Teams using the platform see sub-10-day assessment cycles and 70 to 80% less manual effort per assessment.
Microsoft will publish its final post-incident report, and your vendors will update their own incident logs. The more useful record is the one you build this month: a list of which critical vendors went silent on November 12 and which shared dependency took them down together.
Start with your Tier 1 vendors, map their fourth parties, and check whether their continuity evidence reflects real conditions. The organizations that recover fastest from the next outage will be the ones that can name their shared dependencies before it starts.
To see how AI third-party risk management software maps fourth-party concentration across a large supplier base, book a walkthrough of ComplyScore® built around your own supplier categories.
Microsoft traced the outage to a defect in a July update to its network gateway servers. A routine traffic failover and reduced server capacity triggered it, and the failure spread from server to server until Outlook and other Microsoft 365 apps went offline worldwide.
Map which cloud, email, and collaboration providers your critical vendors rely on, and check whether backup vendors share them. Then collect continuity evidence that reflects real load, monitor vendor events between assessments, and agree on a secondary incident channel with each critical vendor.
No. Microsoft said it found no indication of malicious or cyber activity and later linked the disruption to a latent software defect, a traffic failover, and reduced server capacity. The company also ruled out AI agents as a cause.
Fourth-party risk is the exposure your organization carries through your vendors' own suppliers. If several of your critical vendors depend on the same cloud or email provider, one failure at that provider can disrupt all of them, and you, at the same time.
Concentration risk appears when many critical vendors, or one vendor with no backup, depend on the same provider. Third-party risk programs manage it by mapping shared dependencies, testing continuity evidence, and monitoring those providers between scheduled assessments.