What happened
Microsoft 365 started failing in the early hours of Friday 13 November, Australian eastern daylight time. Outlook and Exchange Online got the worst of it. Teams, SharePoint, OneDrive and the Office apps came back sooner, but email was down for about a day, and Microsoft didn't call it fully resolved until just after 3 am on Saturday.
Microsoft's preliminary post-incident report, which tenant admins can read in the Microsoft 365 admin centre, puts it down to a defect in a July update to the gateway servers that carry Microsoft 365 traffic. That update had been running worldwide for more than three months without a problem. Microsoft had load tested it on older hardware with fewer cores, and the defect turned out to be worse on newer servers with more of them.
Then a few things went wrong at once. A Teams problem the day before had pushed Teams traffic onto the gateway Exchange Online also uses, and an unrelated fault had already knocked out about 14% of that gateway's capacity. Under the extra load, servers locked up and handed their traffic to the next server, which locked up as well. Within two hours more than 80% of the gateway servers had stopped responding. Microsoft could move the other services off the bad gateways. It couldn't move Exchange Online, which is why email stayed down longest.
The final report is due later this week. It may add detail, but it won't change the part that affects you: Exchange Online almost certainly missed its service level for November, and plenty of businesses are owed a credit.
Microsoft owes you something, but only if you ask
Microsoft's SLA for Exchange Online commits to 99.9% uptime each month, and backs that with service credits when it falls short. You don't get the credit unless you claim it. Microsoft awards it after approving a claim, and the SLA says its outage notices aren't confirmation that a service level was missed. So you need to lodge a claim, with evidence, before the deadline.
How the credit is worked out
For Exchange Online, downtime means any time users can't send or receive email in Outlook on the web. Microsoft counts it in user minutes: how long the outage lasted, multiplied by the number of users it hit. The test is Outlook on the web, not the desktop app, so that's where your evidence should come from.
November has 30 days, which gives each user 43,200 minutes in the month. The credit depends on where your uptime landed:
- Below 99.9%, which is more than about 43 minutes of downtime per user, the credit is 25%.
- Below 99%, or more than about 7.2 hours, it's 50%.
- Below 95%, or more than 36 hours, it's 100%.
If every one of your users was locked out of Outlook on the web for the whole 24 hours, you'd be at about 96.7% for November, in the 50% band. Most businesses had a patchier outage than that, and Microsoft checks its own data as well as yours, so don't plan on more.
The percentage only applies to what you paid that month for the service that missed its target, and it comes off a future bill rather than back into your account. On a suite like Business Premium or E3, Exchange Online is one part of the price, and the SLA doesn't say how big a part. For most small businesses the credit will be small. It's still worth twenty minutes to claim.
What Microsoft needs from you
The SLA lists five things a claim has to include, and says Microsoft will reject a claim that leaves any of them out:
- A detailed description of the incident. The incident ID for this outage is EX1550543. You'll find it in the admin centre under Health, then Service health, on the History tab.
- The time and duration of your downtime. Use your own times, not the figure in the news.
- The affected resources. Your tenant name, your primary domain, and the service, which is Exchange Online.
- The number and location of affected users. How many people, and which offices they work from.
- A description of the errors. What people saw in Outlook on the web, any bounce messages, and screenshots if anyone took them.
If nobody wrote anything down on the day, piece it together now from help desk tickets, phone logs and what staff remember. It gets harder every week you leave it.
Where to lodge it, and by when
If you buy Microsoft 365 directly from Microsoft, open a support request from the admin centre and say in the first line that it's an SLA credit claim, not a request for help with the outage.
If you buy through a CSP partner, Microsoft bills the partner rather than you, so the claim goes through them. Send them the five items above and ask them to lodge it. Open licensing and product-key customers are handled differently again: the SLA gives them extra days of service instead of money.
Microsoft has to receive the claim, with everything it needs, by the end of the month after the incident. For this outage that's 31 December 2026. Most businesses shut for part of that fortnight, so get it in before you close. Microsoft says it usually processes claims within 45 days.
What the credit doesn't cover
Under the SLA, service credits are your only remedy. There's nothing for lost revenue, operational costs or any other loss. A day without email cost most businesses far more than a slice of one month's licence fee, and that money isn't coming back.
So lodge the claim, and then spend half an hour on what you'd do differently next time.
- Know where to look. Service health in the admin centre is the record of what Microsoft has acknowledged. Make sure more than one person can get to it, so nobody spends the first hour of the next outage troubleshooting your own network.
- Have a way to reach people that doesn't run on Microsoft 365. When email and Teams go down together, an SMS list or phone tree for staff, and a holding message you can put on your website, will get you through the day.
- Write things down as they happen. When it started, who was affected and what they saw. A claim asks for exactly that, and you'll want it for your own review anyway.
If we supply your licences
We're a Microsoft CSP partner, so if your Microsoft 365 licences come through InterIntra, the claim goes through us. Send us the details above, or point us at the tickets you raised on the day, before the Christmas break and we'll lodge it with Microsoft. Our cloud licensing page explains how that arrangement works.
The short version
Exchange Online's SLA promises 99.9% uptime, and if your email was down for more than about 43 minutes in this outage, Microsoft almost certainly missed it. The credit isn't automatic and won't be big. To claim it you need EX1550543, your times, your user count and what people saw, and it has to reach Microsoft or your partner by 31 December.
If you'd like a hand with the claim, or with a plan for the next time Microsoft 365 goes down, have a look at our Microsoft 365 management service or get in touch.
