•
Madalin
Development enhanced by AI

When Microsoft 365 went down for 30 hours: the case for owning your email

Microsoft 365 was down for over 30 hours. The service credit is a few euros; the lost clients are yours. Why small businesses should own their email, with backups and a restore they can prove.

Madalin
Written by
Platform & infrastructure engineer
A modern building with a large orange sign and red triangular symbols at sunset.

On Thursday 12 November, thousands of businesses lost their email for more than a day. Microsoft’s own incident record for Exchange Online, EX1550543, puts the impact at 30 hours and 5 minutes. Mailboxes would not open in Outlook, on the web or on phones, and mail delivery was delayed.

It did not stop at email. Because so much of Microsoft 365 depends on Exchange Online, calendars and meeting scheduling broke, Copilot stopped answering, and compliance and security tools such as Purview and Defender for Office 365 were degraded. Admins trying to investigate found parts of their own admin tools affected too.

The cause was not an attack. Microsoft traced it to a fault introduced during an earlier network capacity upgrade, which surfaced after a later network change. In other words: routine maintenance, at a scale no customer can see or influence.

Since then, the advice in most inboxes has been about one thing: how to claim your service credit. That advice is fine. But it answers the wrong question.

The credit is not the point

Under Microsoft’s SLA, a business that lost this time may get a service credit. You have to request it yourself, before the end of December, through the channel you buy from.

Look at what that credit actually is. It is a percentage of one month’s fee for the affected service, and for suite licences only the share of the fee that belongs to that service. For a 20-person company, the Exchange Online credit for a full day and a half without email is likely to be around €60: half of roughly €6 per user, for one month. The SLA calls service credits the sole remedy for availability problems. The day your team could not invoice, quote, answer customers or reset a password is not compensated.

So claim it, by all means. But the real question for a small or medium business is a different one: what happens next time, and what if next time is worse?

Your customers don’t care whose fault it was

A large company can absorb a day and a half without email. It has account managers who phone key clients, a press office, and customers locked into contracts. A small business has none of that cushion.

When a small business goes silent, the damage lands on the business, not on Microsoft:

  • Existing clients send a request, get no answer, and start wondering whether you are still reliable.
  • New enquiries go unanswered. A potential customer who writes on Thursday and hears nothing by Friday has usually already written to your competitor.
  • Deadlines for tenders, quotes and orders do not move because a provider had a network fault.
  • Your reputation takes the hit. Nobody outside your company sees Microsoft’s incident number. They only see that you did not reply.

You cannot put a note on your website saying “closed for two days, our email provider is down”. Customers don’t see that as an explanation. To them it looks like a business that isn’t in control of its own tools.

And this is the part worth saying plainly: Microsoft’s commitment ends at the credit. The SLA makes the service credit the sole remedy, so a lost client, a missed tender or a damaged reputation is entirely your loss. The provider refunds a few euros of its fee. You may lose a real share of your business, and nobody will compensate it. Protecting your business against that is your job, and that is why it is worth owning the parts it cannot run without.

The real risk: one dependency behind everything

The striking thing about 12 November is not that a service failed. Every service fails sometimes, including the ones I run. It is how much failed at once, and how little anyone outside Microsoft could do about it.

One internal fault in one provider took out, together:

  • Email, the channel most customers and suppliers still use to reach you
  • Calendars and scheduling, so you could not even move the meetings the outage disrupted
  • The AI assistant many teams now lean on for daily work
  • Security and compliance tooling, at the exact moment you would want it most
  • Parts of the admin tools you would use to investigate

This is concentration risk. When mail, calendar, files, identity, security and AI all sit behind the same provider, you do not have five systems. You have one, with five faces.

For a large enterprise, a day and a half of this is painful but survivable. For a small business, it can mean a lost tender deadline, unanswered orders, or customers who assume you have gone quiet. And this was the mild version: a fault that Microsoft found and fixed. A longer outage, an account suspended by mistake, a billing dispute, a licence change or a regional service decision would leave you with the same problem and even less control.

What owning your email looks like

Self-hosted email does not mean a server under someone’s desk. It means mail runs on infrastructure you control, in the EU, built from open standards: Postfix for sending and receiving, Dovecot for mailboxes, IMAP and SMTP for any client your team already uses. Your domain, your mailboxes and your data stay yours, and you can move them to another host in hours, not months.

Three things make it reliable rather than risky.

1. Backups that leave the server. Mailboxes are backed up automatically, at least daily, to a separate location with a different provider. Backups are encrypted, versioned and kept for a defined period, so a deleted folder from last month, or a mailbox damaged by ransomware, can come back.

2. A restore you can prove. A backup nobody has restored is a hope, not a plan. On a schedule, a restore is actually performed: a full mailbox server rebuilt on fresh infrastructure from backup alone, timed, and the result checked. You end up with a written record: on this date we restored everything in 47 minutes. That record is the difference between “we have backups” and “we know we can recover”.

3. A small blast radius. Your mail server does not share a fault domain with your calendar, your files, your security tooling and the AI assistant. If one piece fails, the rest keeps working, and the failed piece is small enough to understand and rebuild quickly. The person fixing it can see the logs, and can tell you what happened and when it will be back.

That last point matters more than any uptime figure. During a big-tech outage, you wait and refresh a status page. With infrastructure you own, someone accountable to you is working on your problem.

The honest trade-offs

Self-hosting is not automatically more reliable. A single server run carelessly will go down more often than Microsoft does. What you gain is control and a smaller blast radius, and you pay for them in responsibility:

  • Deliverability takes work. SPF, DKIM, DMARC, a clean IP reputation and reverse DNS have to be right, and monitored, or your mail lands in spam.
  • Security is on you. Patching, spam and malware filtering, rate limits and intrusion protection are ongoing jobs, not a one-off setup.
  • It needs a person. Someone has to own it: watch the alerts, apply updates, run the restore drills.
  • Some features are Microsoft-shaped. If your business lives inside Teams, SharePoint and Copilot, moving mail alone will not remove the dependency, and moving everything is a bigger project.

Self-hosting is usually the wrong choice if nobody will own it, if you need features only one vendor provides, or if a short outage costs you less than the effort. It is a strong choice when email is how your customers reach you, when your data has to stay under your control, and when you want to know, not hope, that you can recover.

The expert behind it

The difference between a mail server that fails and one that recovers is the person looking after it. That person can be in-house, if you have a technical team. For most small and medium businesses it is someone contracted: a sysadmin who knows your setup, keeps it documented, and answers the phone when it matters.

Many businesses prefer not to depend on a single person either, and that is sensible. The same principle that argues against one big provider argues against one irreplaceable admin. A healthy setup is documented well enough that a second expert could take over: a written runbook, infrastructure defined as code, credentials held by the business, and restore records anyone can read. You can work with one specialist or several, and change them without losing your email.

That is the real goal. Not “no outages”, which nobody can promise, but no outage you cannot understand, and no failure you cannot recover from on your own terms.

A practical checklist

Whether you move or stay, these are worth doing this month.

If you are staying on Microsoft 365:

  1. Claim your service credit before 31 December, naming every affected service.
  2. Keep an independent backup of mailboxes outside Microsoft, and test restoring one.
  3. Write down how your team reaches customers when email is down: a phone list, a status page, a second channel.
  4. Make sure at least one admin account and its 2FA work without depending on the services that failed.

If you are considering your own mail:

  1. List what actually depends on email in your business, and what depends only on Microsoft.
  2. Start with mail alone. Calendar and files can follow later, or stay where they are.
  3. Agree on a backup schedule, a retention period and a restore drill, and get the first restore record in writing.
  4. Make sure the setup is documented well enough that someone else could run it.

Owning it is a choice you can make now

Big providers are very good at running services at scale. They are not accountable to you, and when they fail, they fail for everyone at once. For a small or medium business, the safest position is not blind trust in one giant, but a setup you understand, backed up somewhere else, with a restore you have already proven.

I build and run this kind of infrastructure for businesses across the EU, and I wrote Mail Admin for managing Dovecot and Postfix servers. If last week made you rethink where your email lives, send a few lines to hi@madalin.me and we can look at what makes sense for you, including whether staying put is the better option.


Sources

Madalin

Madalin

Platform & infrastructure engineer

I build custom web platforms and run the systems behind them, from Huelva, Spain.

Continue the conversation

If this article reflects the challenges your organisation is navigating, explore more practical guidance across Madalin.