November 16, 2026
The chlorine warning never got through, because the mail was down
During the Microsoft cloud outage, a warning sat undelivered in a hotel and 15 guests were injured. Here is how you name the three messages in your business that must never travel by mail alone, and how you set up the second route for them.

At a hotel at Phantasialand, 15 guests were injured because a warning never arrived. It was about chlorine in the pool area, the information was meant to go out by mail, and the mail sat there during the Microsoft cloud outage. No hacker attack, no data loss, no encrypted server. One channel was down, and the message had no second one.
That is the part of the story that concerns you. Not the question of whose data centre failed, but the question of which message in your business has exactly one route.
What businesses actually did during the outage
In Japan, companies fell back on fax machines during the same outage, and orders, appointment confirmations and delivery papers kept moving. The fax is not the point here. The point is that there was a second line that did not hang off the same login, and that people knew how to use it. Whoever usually mocks the fax had a working company that day.
Shortly after the outage, the next wave rolled in. There was a phishing campaign that exploited the failure. The mails pose as a notice about the incident and ask you to confirm your login. It works because on a day like that your people are expecting exactly this mail.
Name your three messages
Sit down for half an hour and write down which messages in your business must never be left lying around. Not twelve, three. Nobody remembers more than that on a Sunday.
Typical candidates, depending on the business:
- Danger to people. Chemicals, gas, electricity, a closed-off area, a machine with an open cover.
- Anything that breaks a deadline. The delivery standing at the loading bay at 6 in the morning, the shift that falls through, the transport that has to be rerouted.
- The outage itself. The message that a central service is down right now and that from this moment on the second route applies.
The third one is forgotten most often and matters most. If nobody says out loud that the mail is down, your people will write into a hole for four hours and wonder about the silence.
The second route has to hang somewhere else
A second route only counts as one if it does not hang off the same login, the same provider and the same network as the first. Chat and mail at the same provider are one route, not two. A phone number on paper, a group SMS through a different service, an answering machine with a recorded message, a WhatsApp group, a fax for all I care. Which one it is matters less than that it hangs somewhere else and that it has been practised.
Practised means something concrete. Twice a year, 20 minutes, one of the three messages goes out over the second route, and afterwards you note how many of the people you reached answered within 30 minutes. If the phone chain has six names on it and the third one carries an old mobile number, you want to find that out on a Tuesday morning and not on a Sunday at six.
Who triggers it on a Sunday
This is where a person decides, and a person named by name. Technology can notice that the inbox has accepted nothing for 40 minutes, and report it. It should not go and wake 80 people by SMS on its own, because half of all fault reports resolve themselves within ten minutes.
So write down two names, a stand-in and a decision rule. For example like this. If one of the three messages has to go out and the usual route is not confirmed within 15 minutes, the shift supervisor triggers the second route, no questions asked. Those three lines fit on one sheet of paper pinned up in the office, and they are the whole difference between an unpleasant morning and 15 injured guests.
That is how we build it in software too. The machine detects and proposes, and a person who has been named for it does the triggering.
How your people spot the mail about the outage
Say it in advance and it sticks. Three sentences are enough.
- During an outage, nobody from your own company asks for a login by mail. No exceptions, not even from management.
- Anyone who receives a mail about the outage clicks nothing and calls the number on the sheet in the office.
- The information that an outage is over comes over the same second route as the information that it started.
That costs you one notice on the wall and five minutes in the next meeting.
How much independence you want is your call
Between “everything stays as it is” and “we run all of it ourselves” there are several steps, and you pick one.
The smallest costs almost nothing. You leave your mail where it is and simply put the second route next to it. The middle one is a second service at a different provider, deliberately separated from the first, rented per mailbox. The third one is open software on your own server, maintained by someone who does that for a living, so third-party software on your own machine with outside maintenance. That is the category we work in, and it is the least known of the three. The fourth is running everything yourself, nights included.
Every step has its price, and the highest is not automatically the right one. What matters is that you keep the choice, because the data belongs to you and every provider stays replaceable. And that you do not find out on the day of the outage which message in your business had exactly one route.
Was this helpful?