Email is not a workflow engine: lessons from the Microsoft outage

When Microsoft’s email went down last week, a lot of UK public services went down with it. The lesson isn’t that email failed. It’s that too much work could only move by email.

On the evening of Thursday 12 November, Microsoft’s Exchange Online started to fail. By Friday morning Outlook, Teams and other Microsoft 365 services were disrupted for users in more than 70 countries. Microsoft said there was no evidence of a cyber attack. Email queues only began to return to normal late on Friday.

What it meant for UK public services

The outage caused widespread disruption across the public sector:

  • Health: Surgeries, pharmacies, and hospitals faced communication breakdowns and reverted to phone or paper.
  • Policing & Justice: Online crime reports were delayed, and court case papers were unable to arrive on time.
  • Government: Central departments and local councils experienced major email service interruptions.

The 999 and NHS 111 lines kept working. They were designed not to depend on a mailbox, and that is the point.

Why email makes a poor workflow engine

Email rarely fails loudly. Messages are queued, not lost, so for hours a stalled workflow looks like a quiet one. A case that hasn’t been assigned looks exactly like one that was never sent. An approval stuck in a queue looks like a manager who hasn’t got round to it.

Meanwhile the statutory clocks keep running: 20 working days for an FOI request, one month for a subject access request.

The failure was also correlated. Mail, chat and file sharing all sat with one supplier, so the channel you would normally use to say “email is down” was down too. And when a shared mailbox is the intake, the assignment tool and the audit trail, the inbox has quietly become the database.

Three rules for resilient casework

We don’t argue for getting rid of email. Citizens and colleagues will keep using it, and they should. The argument is against relying on email alone. In practice, that comes down to three rules:

  1. The case system is the source of truth. The state of a case, task or approval lives in the application. An email can say “something needs your attention”, but nothing should be true only because an email was sent.
  2. Every email-borne action has a non-email twin. If someone can be asked to do something by email, the same request should sit in their work queue in the product.
  3. Intake and release never depend on email alone. Requests should be able to arrive through a web form or API too, and sensitive material should be shared through the system, not as attachments.

These rules are the thinking behind eCase. Requests can come in through eCase Capture web forms, which log, triage and assign them automatically, or through APIs and webhooks, as well as by email. Work sits in eCase workbaskets, not personal inboxes. Responses can be released through secure sending. One Home Office team said the biggest benefit of eCase was no longer having to communicate constantly by email.

The next outage is a matter of when

This wasn’t the first Microsoft 365 outage of 2026, and it won’t be the last. No supplier can prevent the next one. What we can all choose is whether casework stalls when it happens. Email should be one of several ways work moves, and never the only one.

Leave a Reply