Microsoft 365 and Outlook Went Down for Ten Hours: What Orange County Businesses Should Change Before the Next One

For most of the working day on Wednesday 12 November 2026, a large share of U.S. businesses could not send or receive email. Exchange Online, the service behind Outlook, Microsoft 365 mail, shared calendars and contacts, lost capacity in the early afternoon Eastern time and did not reach stable recovery until close to midnight. Teams, SharePoint, Copilot, Defender, To Do and Purview were disrupted along the way. Outage-tracker Downdetector logged a peak of 128,874 Outlook problem reports at 1:09 p.m. ET and more than 47,000 for Microsoft 365, and reports were still arriving from India, Australia and Japan as Thursday began in those time zones.

This brief sets out what actually happened, what Microsoft said as it worked the problem, who was hit, and the part that matters for a business in Irvine, Santa Ana or Anaheim: the specific, inexpensive changes that decide whether the next cloud outage costs you an afternoon or a week.

What happened, in order

The first public signal came shortly after 1 p.m. ET, when Microsoft confirmed that connectivity to Microsoft 365 services “may affect” users in any region, not a single geography. Its early response was to redistribute connection traffic across its wider worldwide infrastructure to spread the load, and to pause all deployments so that nothing new was changing while engineers worked.

Through the afternoon, the picture Microsoft painted was of servers that could be restarted and returned to service but then fell over again under load. The company said it had found a core process on the affected servers becoming overwhelmed, causing capacity to drop. It validated an alternative health-check method that let a subset of servers come back cleanly, began applying it more widely, and built automated recovery to detect and restart unresponsive servers continuously. It also restricted which types of traffic could reach the affected infrastructure in some locations, and shifted traffic to alternate routes wherever it could.

By 5 p.m. ET, Outlook reports had fallen below 50,000 and Copilot was working for most users. By 7:45 p.m. ET, Outlook reports were around 31,000 and Microsoft 365 reports under 10,000. At 9 p.m. ET the company reported positive results from restarting infrastructure “in an isolated manner.” An hour later it said those changes were “not taking as expected” and that it was looking for alternative ways to add capacity against “sustained traffic management imbalances.” The last update before this brief was written, around 11:30 p.m. ET, described partial recovery across the regions where adjustments had been applied, but not sustained recovery, with more than 18,000 Outlook reports still open in the United States.

Microsoft’s service-health notice was precise about scope: features that interact with Exchange Online and email were affected; features that do not depend on Exchange Online kept working. That distinction explains why Teams chat often worked while calendar invitations did not, and why SharePoint files opened while the email notifications about them never arrived.

Who felt it

The disruption was not confined to office workers refreshing Outlook. In Southern California, communications problems at the LA County Sheriff’s Department and several other local agencies were reported by NBC4, and the Los Angeles County Medical Examiner’s Office posted a notice that its email was down because of the nationwide Outlook outage. In Arapahoe County, Colorado, the sheriff’s office said online bond payments for people held at the county jail could not be processed because the email notifications that drive the workflow were not arriving; anyone posting bond had to come to the jail in person.

Workhorse Group, an electric truck maker, pushed its third-quarter earnings release from 12 November to 16 November, citing business disruption from the outage. And the AI providers that many businesses now plug into Microsoft 365 were hit downstream: OpenAI reported that its Microsoft connectors were failing to make new connections and tools were misbehaving, and Anthropic traced a fault in its own Microsoft 365 connector to the upstream provider.

Microsoft did not immediately provide further comment beyond its service-health updates. This is the second major Microsoft cloud incident in a month, following the October Azure outage that took down the company’s productivity services and a long list of customer-facing sites worldwide.

Why one Exchange Online fault reaches so far

Exchange Online is not just a mailbox. In a typical Orange County business it is the identity hub for everything that happens by email: password resets, multi-factor prompts delivered by mail, invoice approvals, e-signature requests, calendar bookings, helpdesk tickets, CRM alerts, shipping notifications, and the automated messages that many line-of-business applications use to trigger the next step in a process. When Exchange loses capacity, every one of those workflows stalls, often silently. The Arapahoe County bond-payment example is a clean illustration: the payment system worked, the jail’s staff were at their desks, and nothing moved because a notification email did not land.

The second amplifier is the connector economy. In 2026 a mid-sized firm commonly has a dozen SaaS tools reading and writing Microsoft 365 data through Graph API connectors: AI assistants, document management, backup, compliance archiving, phone systems, marketing automation. Those tools degrade together when the upstream degrades. OpenAI and Anthropic reporting connector failures on the same afternoon is what that looks like from the vendor side; from the customer side it looks like three or four unrelated tools all failing at once.

The third is behaviour under recovery. Microsoft’s own updates describe restarted servers that became unresponsive again as soon as traffic returned. That is the classic retry-storm pattern: every Outlook client, mobile device and connector that lost its connection retries aggressively the moment a server answers, and the surge knocks it over. It is why “the servers are back” and “your email works” were separated by hours, and why the company had to throttle traffic types and build automatic restart loops rather than simply rebooting.

What it cost a typical Orange County business

Put rough numbers on it. A 40-person professional-services firm in Irvine with an average fully loaded cost of $75 per hour per employee, where email and calendar underpin perhaps half of the day’s work, loses on the order of $1,500 for every hour the productivity hit lasts. Ten hours of disruption, even at reduced impact for part of that, lands somewhere between $8,000 and $15,000 in lost output, before counting missed deadlines, delayed proposals, and clients who could not reach you. A clinic with appointment reminders, lab-result notifications and referral emails flowing through Microsoft 365 has a patient-care exposure on top of the financial one. A distributor whose order confirmations go by email has customers who did not know whether their order was received.

None of that is Microsoft’s liability. The Microsoft 365 service agreement provides service credits against the subscription fee when monthly uptime falls below the committed level, and a ten-hour outage in a month is a credit worth a fraction of one month’s licence. The lost work is yours.

Seven changes to make before the next outage

Everything below is standard managed-IT practice and most of it costs little or nothing. The point is to decide now, while the memory is fresh, rather than during the next incident.

1. Keep a second channel that does not depend on Microsoft

Your team needs one place to coordinate that is not Teams and not Outlook. For many firms it is a phone-system group text, a Slack or Signal group kept dormant for exactly this purpose, or the chat function in your phone provider’s app. Write the fallback into the business-continuity plan, put the join link on a laminated card or in a password-manager note, and test it twice a year. The failure on 12 November for many businesses was not the loss of email; it was that nobody could agree where to talk about the loss of email.

2. Put a status feed where staff will see it

Microsoft publishes service health in the Microsoft 365 admin centre and on its public status page and social channels. Someone on your side, or your managed IT provider, should be subscribed, and the first minutes of any outage should go to confirming whether it is Microsoft, your internet carrier, or your own network. On Wednesday, helpdesks across the county spent the first hour answering “is it just me?” one ticket at a time. A one-line banner on your intranet or a group text that says “Known Microsoft outage, not us, no action needed” prevents that.

3. Make sure MFA and password reset do not live only in email

If your multi-factor codes or password-reset links arrive by email, an Exchange outage can lock people out of systems that have nothing to do with Microsoft. Move MFA to an authenticator app or hardware key, confirm the self-service password reset path has an SMS or phone option, and check that your identity provider’s emergency-access accounts are configured and the credentials are stored offline.

4. Identify the workflows that stall when email stops

Walk the Arapahoe County exercise through your own operations. Which approvals, payments, notifications or hand-offs are triggered by an email arriving? For each one, decide whether it needs a manual fallback (a phone call, a printed list, a shared spreadsheet) or whether it can be re-pointed to a channel that stays up. Invoice approvals and customer order confirmations are the usual first finds.

5. Back up Microsoft 365 independently

Wednesday’s incident was availability, not data loss, but it is a reminder that Microsoft’s retention is not a backup. A third-party backup of mailboxes, OneDrive, SharePoint and Teams gives you a copy you can search and restore from when the primary is unreachable, and it is the only defence against deletions, ransomware in synced folders, and mailbox corruption. If you do not have one, this is the week to ask your provider for a quote.

6. Map your connector dependencies

List every application that authenticates through Microsoft 365 or reads data from it, including AI tools. For each, note what breaks when the connection drops and whether the application has an offline mode. The list is usually longer than anyone expects, and it is the list you will need during the next incident to answer “why is the phone system not showing calendar availability?”

7. Decide whether email continuity service is worth it

Several security vendors offer email continuity: an emergency mailbox that keeps receiving and sending during a Microsoft outage, then synchronises when service returns. For a firm where an hour without email is a real cost (law practices, medical offices, logistics, anyone with same-day customer commitments) it is often the cheapest insurance available. For a firm that can tolerate a half-day, it may not be. Make the decision on numbers, using the hourly cost calculation above.

What this means for the cloud question

Two Microsoft incidents in four weeks will prompt some owners to ask whether they should be in the cloud at all. The honest answer is that an on-premises Exchange server would have had worse uptime over the past year, not better, and would have carried patching and security burdens that most Orange County businesses are right to have handed off. The lesson of 12 November is not “leave the cloud.” It is “design for the cloud being unavailable for a day,” which is a planning exercise, not a migration. The businesses that coped best this week were the ones with a second channel, an independent backup, authentication that did not route through email, and a provider who told them within minutes that the problem was upstream and what to do in the meantime.

What Technijian clients saw

For businesses on a managed service agreement with us, the outage was confirmed as a Microsoft-side incident shortly after the first reports and communicated through the out-of-band channel each client has on file. Where email continuity is in place, mail continued to flow through the emergency mailbox. Where it is not, we logged the hours of impact so each client can make the decision in item 7 with their own numbers. If you are not a client and want the same for your business, a short continuity review is the place to start.

Frequently asked questions

Was any data lost?

Nothing in Microsoft’s updates indicates data loss. The incident was a loss of capacity and connectivity on Exchange Online infrastructure. Mail sent to you during the outage was queued by sending servers and should have been delivered as service recovered, though some senders’ servers give up after a set number of hours; if a specific message is missing, ask the sender to resend.

Was it a cyberattack?

Microsoft has not said so and nothing in its updates suggests it. The company’s descriptions, a core process becoming overwhelmed, servers failing under returning traffic, and traffic-management imbalances, point to a capacity and recovery problem rather than an intrusion. We will update this brief if Microsoft publishes a post-incident report that says otherwise.

Our Outlook still shows “disconnected.” What should we do?

Close and reopen Outlook, or sign out and back in on the mobile app. Clients that lost their connection during the outage sometimes do not reconnect on their own even after service returns. If it persists across several users, check Microsoft’s service health page before raising an internal ticket.

Do we get money back from Microsoft?

Possibly a service credit, if monthly uptime for the affected service fell below the committed level. Credits must be claimed through the admin centre within the window set in the service agreement, and they are a percentage of the monthly fee for the affected service, not compensation for lost business. Your managed IT provider can file the claim.

Does this change the case for Copilot and AI connectors?

It changes the planning, not the case. Any tool that reads your Microsoft 365 data will be unavailable when Microsoft 365 is. Treat AI assistants as part of the dependency map in item 6 and make sure nobody’s critical daily process runs only through them.

Is this related to the October Azure outage?

They are separate incidents on different parts of Microsoft’s infrastructure: October’s was an Azure platform issue that took down customer websites and services broadly, Wednesday’s was Exchange Online capacity. The pattern that matters to you is the same in both: a single upstream provider, a multi-hour impact, and the businesses with a plan recovering faster than those without.

What to watch next

• Microsoft’s post-incident review, which usually follows within five business days and should name the root cause of the overwhelmed process.
• Whether service-credit eligibility is confirmed for November, and the claim deadline.
• Any residual Outlook connection problems on Thursday morning as U.S. clients reconnect at the start of the day.
• Connector vendors, including OpenAI and Anthropic, confirming full recovery of Microsoft 365 integrations.

Technijian will update this brief when Microsoft publishes its root-cause report.

Source: USA TODAY, “Is Microsoft down? Updates as users report issues with Outlook, 365”, Mary Walrath-Holdridge, Fernando Cervantes Jr. and Thao Nguyen, 12 November 2026 (updated 13 November). Outage figures from Downdetector as reported there; Microsoft statements from its service-health notices and social updates.