07480 257117 Talk to us
For business ownersBuyer’s guide

What to set up before your email provider goes down

Six things a business can put in place, most of them in an afternoon by our estimate, so that an outage at its email provider costs it a slow day and not a silent one. Each comes from published guidance or from how we build our own clients' sites.

6 set large on a pale grey background beside six stickers naming what to set up: contact list, website notice, form that saves, status page, emergency admin, a practice run, above the words: Things to set up before your email provider has a bad day
6things to set up, five of which need no new software
2emergency admin accounts Microsoft tells organisations to keep, at least
90 dayshow often Microsoft says to check those accounts still work, at most
4 secondswhat the zxn.uk quote form gives the save before moving on to the email

On Thursday 12 November a business with its email at Microsoft could not send a message, and for most of them that also meant it could not tell anyone it could not send a message. We wrote up what happened. This is the list we would hand a client the week after one.

Five of the six need no new software. They are a matter of deciding, on a quiet day, where small things live. The times below are our estimates.

The rule: nothing you need on the bad day lives in the thing that failed

Many businesses keep their mail, their calendar, their chat and their address book with one supplier. That is sensible and we would not tell anyone to split it up. The risk is keeping the fallback there too, which is easy to do: the customer phone numbers in Outlook contacts, the staff rota in a shared mailbox, the “how to reach us” page that says only “email us”.

What fails together, and what does not

Four things most businesses keep with their email provider, and four worth keeping outside it.

Inside your email providerone outage takes all four togetherOutside itstill there on the bad dayMailboxessending and receivingCalendar and chatmeetings, messages, callsThe address bookcustomers, suppliers, staffThe admin centrewhere you would look for newsA contact liston phones and on paperA notice on your websiteone line you can change yourselfA form that saves firstenquiries kept, then sent onThe public status pagebookmarked, no sign-in needed

The left column fails together, because it is one service. The right column is cheap, and none of it depends on the left.

The National Cyber Security Centre says the same thing in its guidance for organisations whose systems have been disrupted: set up “alternative mechanisms for communicating with staff, stakeholders and customers”. It wrote that for cyber attacks. It applies just as well when the supplier has broken itself.

The six things

1. A contact list outside your mailYou, this week

Staff mobile numbers and the customers you could not afford to go quiet on, kept on phones and on one printed sheet. If the only copy is in Outlook, it is not a fallback.

2. A notice you can put on your website yourselfWhoever runs your site

One line at the top of the home page: "Our email is down. Ring us on this number." Ask now how you would change it, and how long it takes. If the answer is "email whoever built it", it is worth agreeing a faster route now.

3. An enquiry form that saves before it sendsWhoever built your site

A contact form that only sends an email keeps no copy of its own. If the mailbox is down, the enquiry depends on a retry you cannot see. A form that writes each enquiry to its own record first keeps it whatever happens next.

4. The public status page, bookmarkedYou, two minutes

Microsoft says that if you cannot sign in to the admin centre, the status page at status.cloud.microsoft is where to look. Service health emails can go to two addresses. Our advice is to make one of them an address that is not at Microsoft.

5. Two emergency admin accountsYour IT person

For a different bad day: the one where the fault is your own sign-in setup and every administrator is locked out. Microsoft says to keep two or more, and to check them at least every 90 days.

6. A practice runEveryone, an hour

Pick a Tuesday, pretend the mail is off, and try to reach three customers and all your staff. The National Cyber Security Centre publishes free exercises for small organisations if you want a script.

2emergency admin accounts, at least
90 daysbetween checks on them, at most
2addresses service health emails can go to

What you are allowed to send customers

Owners hold back from texting or emailing customers from a second address because they worry about marketing rules. The Information Commissioner’s Office draws the line clearly. A message sent “for administrative or customer services purposes” is a service message, not marketing. By our reading, “Our email is down, your order is unaffected, ring this number” is one: it tells people how to reach you about a problem and sells nothing.

The line is crossed the moment you add a promotion. The guidance is blunt that a message with any marketing in it counts as marketing as a whole, even if that was not its main purpose. So no “while you are here, 10% off”. Say what has happened and how to reach you, and stop.

One caution of our own, which is not from that guidance: send only to customers who would expect to hear from you. An outage is not a reason to message everyone who ever filled in a form.

What you do not need to buy

You do not need a second email system on standby to catch incoming mail. The standard email runs on advises the sender’s server to keep trying, waiting at least 30 minutes between attempts, for at least four to five days, and Microsoft said service was back for the last customers on the Saturday, the third day. Most mail sent to you should arrive late, not never. What you cannot recover is the days in which nobody could reach you and you could not say why.

So spend the effort on the outgoing half: the notice, the list, the form. A notice only helps if customers can find the site it sits on. When we checked 5,599 UK business homepages against Google’s own advice, only 27% passed six basic checks together.

What we do about it for our own clients

Two of the six are things we build, so here is how, plainly.

The quote form on zxn.uk saves each request to our own record first and gives that save four seconds. Then it sends the email, then a short alert to us. The visitor sees “sent” if the email or the alert got through, and a failed save never stops either. The enquiry form on a client site we look after works the same way round: the enquiry is written to the site’s own record before any email is sent. So when a mailbox is down, the enquiry is still on record.

The sites themselves are served from Cloudflare, so an outage at Microsoft does not take a site or its form down. If the client’s mailbox is at the provider that failed, the notification email waits, and the saved enquiry is still there. The honest other half: if Cloudflare has a bad day, the site and the form are down together, and we have written about that risk too.

What we do not sell today is a standby service that takes over a company’s customer messages when its provider fails. This week made the case for one, and we are working out what it would have to do to be worth trusting. We will write that up when there is something real to show.

Where the numbers come from

  1. Microsoft Learn, How to check Microsoft 365 service health, read on 14 November 2026: the public status page for when you cannot sign in, and service health emails to up to two addresses.
  2. Microsoft Learn, Manage emergency access admin accounts in Microsoft Entra ID, read on 14 November 2026: two or more cloud-only accounts, checked at least every 90 days.
  3. National Cyber Security Centre, What to do when cyber attacks disrupt your organisation, immediate activities, read on 14 November 2026: setting up alternative ways to communicate with staff, stakeholders and customers.
  4. National Cyber Security Centre, Exercise in a Box, read on 14 November 2026: free practice exercises for small and medium organisations.
  5. Information Commissioner's Office, direct marketing guidance, Identify direct marketing, read on 14 November 2026: service messages and what turns one into marketing.
  6. RFC 5321, Simple Mail Transfer Protocol, section 4.5.4.1: sending servers should wait at least 30 minutes between attempts and generally keep trying for at least 4 to 5 days.
  7. Our own write-up of how the zxn.uk quote form saves a request before sending it, 13 November 2026: the order of save, email and alert, and the four seconds.
  8. Our post on the Microsoft outage of 12 to 14 November 2026 and the press reports listed there.

Edward Clarke, Director. A decade of website development, testing and customer success. Builds our client sites.