Insights/Guides

The Microsoft 365 outage of November 2026 and what support teams should take from it

What happened from 12 to 14 November, how customers and other brands reacted in public, what Microsoft could have done, and a first hour plan.

Thinkiyo team··14 min read

Microsoft 365 went down on Thursday 12 November 2026 and took more than a day to recover. Three things from it apply to any software company. Customers went public within the hour, and most of what they posted was not addressed to the vendor. Microsoft published frequent, dated updates without a fix time, which held up, and published early theories about the cause, which did not. And many companies found that their own support inbox ran on the thing that was down. Plan your first hour around those three facts. There is a fourth lesson for whoever runs your social accounts: other brands joined the conversation, and the company at the centre of it mostly did not.

What happened

From news reports we opened on 14 November. The sources disagree on some times, so each is given with its publisher.

Start. NBC News reports that Microsoft put the start of the incident at 11:17 a.m. US Eastern time on Thursday 12 November, which is 16:17 UTC. Elsewhere the same page says about 12:50 p.m. Eastern. ABC News says Microsoft's first public post, about users having trouble reaching their mailboxes, came shortly before 1 p.m. Eastern. Tom's Guide says user reports began at around 12:00 p.m. Eastern. Customers were reporting trouble before the vendor's first public post.

Services. NBC's summary of Microsoft's list, on Friday morning: Exchange Online, meaning mailbox access through Outlook and delayed email delivery. Copilot chat. Intune device enrolment and check-in. In Teams, calls, meetings and chats mostly worked, while calendar functions that depend on Exchange and the creation of private or shared channels did not. Azure showed no issues. PCWorld lists the consumer side as Outlook.com, OneDrive, OneNote, Microsoft To Do, Teams and Microsoft 365 for the web. Tom's Guide's copy of the status page also lists the admin centre.

Scope. ABC reports that Microsoft first described an access issue in the Americas and later said it "may affect Microsoft 365 service connectivity for users in any region."

Duration. More than a day. At about 3:05 p.m. Eastern on Friday 13 November, almost 28 hours after Microsoft's stated start, NBC quotes the company saying: "We do not have an ETA for full recovery yet." The Associated Press, as carried by ABC7 Chicago, reports that Microsoft said on Saturday 14 November that it had "restored service to the remaining impacted customers and regions." None of the pages we opened gives an exact end time.

Cause. A full root cause has not been published as far as we could find. What Microsoft has said, quoted by ABC, NBC and PCWorld: "This was a technical issue not related to cyber or AI / agents. The incident was introduced by an issue during an earlier network capacity upgrade that surfaced following a recent network change." Tom's Guide quotes Microsoft promising a full post-incident report once its review is complete.

We could not open Microsoft's own posts on X, so every Microsoft statement here is quoted from the news pages linked. Its status page showed all products operational when we loaded it on 14 November.

How customers reacted in public

Fast. ABC reports over 100,000 reports on Downdetector within an hour of Microsoft's first post. Windows Central saw nearly 130,000 at around 1:15 p.m. Eastern. NBC, citing Downdetector, gives a peak of 19,650 reports a minute across five Microsoft services, against fewer than 15 on a typical Thursday, and reports from 74 countries and territories within about an hour.

Downdetector counts reports that users submit. It does not count affected users, and the Associated Press says the number affected was not clear. Read the figures as a measure of how quickly people speak up.

Mostly not to the vendor. Tom's Guide describes Downdetector's comment section being used as a message board, full of jokes with the occasional useful status report. NBC quotes Reddit users swapping workarounds ("Having meetings on WhatsApp now instead") and X users joking about a day off, and reports brands such as Air India and Burger King joining in. Customers talk to each other in places you do not run, and a status page they never visit does not reach them.

With the wrong name on it. Tom's Guide noted users filing Downdetector reports against Gmail, Zoom and Yahoo Mail during the outage. Windows Central says spokespeople for AWS and Cloudflare got in touch to say their services were not down. Expect to be blamed for a provider's outage, and to be reported as down when you are not.

Then your customers have to explain it. NBC quotes Baltimore County telling residents that staff could not send or receive email and that 911 "and other essential County services are NOT affected." It reports IndiGo saying its email support was disrupted, and Air India rerouting its email so customer queries could still get through. Each had to write its own message about a vendor's failure.

How Microsoft communicated

What it didSourceTake for your team
Posted dated updates through the night. NBC logs them at about 2:30, 4:15, 6:00, 7:00, 8:00 and 9:00 a.m. Eastern on FridayNBC NewsA steady rhythm is possible even with no fix. Promise the next update time
Refused to give a fix time, and explained that sign-in could recover before mail flow didNBC NewsSay what will come back first. It stops a second wave of "it's still broken"
Said early and repeatedly that it saw no sign of malicious activityABC News, NBC NewsAnswer the question customers are most afraid of as soon as you honestly can
Published working theories as it went. On Thursday it said it had "not identified any recent change correlated with the start of impact." On Friday it said a network change had surfaced the issueTom's Guide, ABC NewsReport what you are doing and seeing. Keep cause theories out of public updates until confirmed
Sent the same text to X and the status page, with incident numbers (MO1550821, EX1550543) and a second link in case the admin centre would not loadTom's Guide, Windows CentralOne message, every channel, one reference customers can quote
Ran a status page that a Tom's Guide writer could not load for a stretch, and an admin centre that was itself affectedTom's GuideHost your status page away from your own product
Extended an email timeout window to 48 hours to protect delayed messages, and said soNBC NewsTell customers what is safe. "Your data is queued" removes a lot of tickets
Apologised in plain words on Friday ("We are deeply sorry") and promised a post-incident reportPCWorld, NBC NewsApologise once, clearly, and commit to the write-up

The Thursday line about recent changes is the one to learn from. A day later it read as wrong, and it is the line people quote back.

The brands that joined the conversation

While Microsoft posted status updates, other brands treated the outage as the biggest conversation of the week. Marketing-Interactive collected 13 of them. A sample, with the response it reports:

BrandWhat it postedReported response
YahooA text post on Instagram saying that with work email down, it was time for "your 346,538 unread personal emails"Over 102,000 likes
IKEA SingaporeIts jerboa soft toy with the line "Thank god there's no outage on this micro-soft."Over 21,900 likes
Chili'sA reply on X to the official Microsoft 365 Status account, with a giveaway of 500 free mealsOver 16,000 likes
KitKatA video on its global Instagram account of a man finding Outlook down and going for a break, plus a Malaysian post with a KitKat in the Outlook logoNot given
McDonald's"have they tried putting our emails in Sprite" on XNot given
Burger KingA post on X asking whoever was fixing the outage to take their timeNot given

Three things stand out. Yahoo is a direct competitor, and it got a six figure response without naming Outlook. Chili's posted its giveaway as a reply to Microsoft's own status account, so it sat under the update customers were looking for. And none of the others sells email at all.

Gulf Business covered posts from Emirates and KitKat, and KitKat USA posted on Facebook with a nod to Microsoft. We could not open either page, so we quote nothing from them.

What Microsoft could have done on social

This part is our opinion, not reporting. We could not open Microsoft's posts on X, so we do not know what replies it sent to individual users. From the coverage we read, its public posts were status updates.

Nobody can answer 19,650 reports a minute, and that was never the job. The job is to change what everyone watching sees. Five things would have helped.

  1. Join the joke early. A light reply from the official account to the jokes, followed by the current status. It puts your own update at the top of the thread people are sharing.
  2. Sort the noise into three piles. Jokes get a light touch. Questions get the status link and one line. Real harm, such as a missed deadline or a service that cannot reach its users, gets a named person and a direct message.
  3. Answer the most seen posts every hour. You cannot reply to everyone. Everyone can see who got a reply.
  4. Give people words to pass on. Many of the people posting were administrators being asked questions by their own staff. A line they can forward does your communication for you. There is one under "Messages you can copy" below.
  5. Close the loop. Afterwards, go back to the people who complained and tell them it is fixed and what changed.

On 14 November an IT services firm published After the Microsoft 365 outage: should your business switch to Google Workspace?. Its answer is that the outage is not, by itself, a reason to move every mailbox. We found no figures showing customers leaving. The cost is that the question was being asked at all.

Why the comment section matters

Comments, replies, reels and reviews are where a brand is judged, and the judgment is public. Someone deciding whether to buy reads the comments, and tends to believe them, because other customers wrote them.

Many brands publish and move on. A complaint with no reply under it tells every reader how they will be treated. Answering does two more things. A conversation under a post is engagement, which is what feeds reward with reach. And people come back to a place where they were heard.

Your first hour

These timings are a suggested starting point. They are not a benchmark. Adjust them to your product.

Minutes 0 to 10

  1. Three matching reports from different customers are enough. Raise an incident. Support is allowed to do this without waiting for a dashboard.
  2. Name one person who writes every customer message. Nobody else describes the problem.
  3. Check whether your own support email, chat or phone depends on what is down. If it does, switch to the backup channel now.

Minutes 10 to 20

  1. Post the first update on the status page and your main social account. Symptoms, what still works, time of the next update. No cause.
  2. Give every agent the same short reply. Pin the status link on your social profiles.

Minutes 20 to 60

  1. Search for your name on X, Reddit, Downdetector and your app store pages. Reply to questions with the status link. Leave the jokes alone.
  2. Tag every related message, on every channel, so you can write to those people when it ends.
  3. Send engineering what customers are seeing: counts, regions, odd cases. Customers often spot scope before monitoring does.
  4. Post the second update on time, even if nothing has changed.
  5. Tell customers what is safe: whether data is lost or queued, and whether they need to redo anything.

Before any of this

  • Keep one support channel that does not share a provider with your main one.
  • Save the contact list for your largest customers somewhere outside your email system.
  • Write the templates below now.

For roles, service credits and the closing message to affected customers, see our guide to handling a ticket spike during a SaaS outage.

Messages you can copy

Fill in the brackets.

First public update

We're investigating a problem with [sign-in]. Some customers can't [log in]. [Reports and exports] are working. We don't know the cause yet. Next update by [14:30 UTC].

When the fault is upstream

[Provider] is having an outage and it is affecting [email notifications] in our product. Your data is safe and messages are queued. We're watching it and will post here by [15:00 UTC]. You don't need to contact us to be kept informed.

Update with no fix time

Still working on [sign-in]. We have restarted the affected servers and some customers are back in. We don't have a fix time we can stand behind, so we won't give one. Next update by [15:30 UTC].

Reply to a customer who needs to tell their own customers

Hi Farah, here is wording you can pass on: "[Product] is having a sign-in problem that started at [13:10 UTC]. No data has been lost. Updates at [status link]." I'll message you directly when it is fixed.

Hard case: the customer wants the cause now

Hi Tomás, I don't have a confirmed cause and I won't guess. We have seen no sign of an attack. We'll publish a full account by [date] and I'll send it to you myself.

Mistakes this outage shows

  • Waiting to post until someone senior approves. The public count was past 100,000 within an hour.
  • Publishing a cause theory. It will be quoted after it changes.
  • Giving a fix time to end the pressure. Microsoft declined for more than a day and its updates stayed credible.
  • Running the status page or support inbox on the same stack as the product.
  • Reading tracker report counts as affected users.

What to measure

  • Time from the first customer report to your first public update. Here the reported gap between Microsoft's stated start and its first post was over an hour.
  • Longest gap between updates.
  • Share of public posts about the incident that got a reply or a link.
  • Share of affected customers who received a closing message.
  • Whether your backup support channel worked at its last test.
  • Statements you had to correct afterwards.

What runs in Relay today

Relay pulls customer messages from connected channels (email, chat, WhatsApp, social posts, reviews and app store reviews) into one inbox, and drafts replies that a person approves and sends. It scores each message for sentiment and groups messages about the same problem into a theme, which is the sorting described above: jokes, questions and real harm.

The limits. Each channel works only after you connect your own account or key. Drafts come from templates today. Nothing is sent automatically. The companion listener raises basic alerts on new negative mentions. An automatic trigger on a spike in volume is on the roadmap, and crisis mode, which checks sources more often, is switched on by hand. If that fits how you work, talk to us. See also how to monitor brand mentions and how to do online reputation management.

Common questions

How long did the Microsoft 365 outage last?

More than a day. NBC News reports Microsoft's stated start as 11:17 a.m. US Eastern on Thursday 12 November 2026, and at about 3:05 p.m. Eastern on Friday the company still had no time for full recovery. The Associated Press reports that Microsoft said on Saturday 14 November that service was restored to the remaining customers and regions.

What caused it?

Microsoft said it was a technical issue introduced during an earlier network capacity upgrade, which surfaced after a recent network change, and that it was not related to a cyberattack or to AI agents. A full post-incident report was promised and we could not find one published as of 14 November 2026. Anything more specific is unconfirmed.

Which brands reacted to the Microsoft outage?

Marketing-Interactive lists 13, including Yahoo, KitKat, IKEA Singapore, Chili's, McDonald's, Burger King and Wendy's. Yahoo's Instagram post is reported at over 102,000 likes, and Chili's replied directly to the official Microsoft 365 Status account on X.

Should we tell customers when the fault is our provider's?

Yes, in one factual line, and then keep owning the fix for your customer. Say what is affected in your product, whether data is safe, and when you will next update.

What if our support inbox is down too?

Have a second channel on a different provider and tell customers where it is, on your status page and social profiles. NBC News reports Air India rerouting its email during this outage so customer queries could still arrive. Test the fallback before you need it.

Sources

See what Relay can do for your customer service

Get a personalised 30 minute demo on your own channels and policies. We show you Relay handling your real conversations, then set it up with you.

  • Built and connected for you
  • Can run in your own cloud
  • A person on every refund

Brands we work with

  • Bolt
  • Dave & Buster's
  • ETV Win
  • Malachite Cars
  • Space Angel
  • Infinity Quest
Made by Thinkiyo© 2026 Thinkiyo. Relay is a Thinkiyo product.Examples use Lumen Parcel, a fictional brand.