Guide
Email alerts
A monitor that breaks should tell somebody. This is what Mend sends, when it decides a failure is worth an email, and how to point those emails at a team alias rather than whoever happens to own the account.
Last reviewed: 8 August 2026
What email is for
Email is the exception report. It writes to you when a monitor stops working and again when it starts working, and stays quiet the rest of the time. That is the opposite of Slack notifications, which post every run including the clean ones. Most people want both, doing their separate jobs; the two are configured independently and turning one off leaves the other alone.
A “failure” here means the scan could not complete — the page would not load, a sign-in step no longer matches, a stored credential has expired. It does not mean the page has accessibility issues. Issues are the normal output of a working monitor, and mailing about them would mean mailing every day forever.
Why the first failure is often silent
By default Mend waits for a second consecutive failure before emailing about certain errors — timeouts, renderer crashes, dropped connections, DNS hiccups. These are usually the network, the container, or a site having a bad minute. None of them is something you can act on at 3am, and nearly all of them are gone by the next run.
Every other error mails immediately, because everything else in the scan path stays broken until a person does something: an expired session, a missing vault credential, a selector that no longer matches, a redirect that lands somewhere unexpected.
If you would rather hear about the one-off timeout than not hear about it, Alert on the first failure turns the wait off for that monitor. It is a deliberate trade of more noise for less delay, which is why it is a setting rather than a number we picked for you.
You are told once per outage
While a monitor is failing, Mend remembers that it has already told you. A page that is down for a week produces one email, not seven. When the monitor succeeds again you get a single “working again” message, and the cycle resets — the next outage is a new outage and gets its own email.
Switching alerts off does not break that bookkeeping. If a monitor recovers while its alerts are muted, Mend still closes the episode internally, so re-enabling alerts months later does not leave you permanently silenced by a flag left over from an outage you have long since forgotten.
The three settings
These live on each monitor, under Email alerts on the Monitors page. They are per-monitor rather than per-account on purpose: the thing people want silenced is one flaky staging URL, not every alert they get.
- Email me about this monitor — the master switch. Off suppresses failure and recovery mail alike. On by default.
- Tell me when it recovers — whether the “working again” message sends. On by default. Some people want the all-clear; others consider the silence sufficient.
- Alert on the first failure — skips the wait-for-a-second-failure rule described above. Off by default.
A monitor nobody has configured behaves exactly as monitors did before these settings existed, so you do not need to visit this panel to keep what you had.
Sending alerts somewhere else
By default alerts go to the address you signed in with. You can add others and choose, per monitor, which one hears about it. This exists for the common case where the person who set up monitoring is not the person who should be woken by it: a team alias, a shared inbox, an on-call rota address — and for the equally common case where the marketing site and the payments service should not wake the same people.
Addresses are kept once per account and reused across monitors. Add one under Alert addresses on the Monitors page; once it is confirmed it becomes selectable as the recipient under Email alerts on any individual monitor. A monitor you never touch keeps sending to your account address.
A new address has to be confirmed before anything is sent to it. Mend emails that address a verification link, and alerts keep going to your account address until someone opens it. This is not a formality — without it, a typo would silently redirect every alert you have into a stranger's inbox, and you would find out the next time something broke and nobody heard about it.
The link is good for 24 hours and works once. You can send a fresh one if it expires, with a short cooldown between requests so the button cannot be used to repeatedly mail somebody who did not ask for it.
Opening the link does not require a Mend account or a signed-in session, which is deliberate. The person confirming accessibility@your-company.com is usually not the person who typed it in, and often has no login here at all — demanding one would make exactly the addresses this feature exists for impossible to verify. The link itself is the authorisation: it carries 256 bits of randomness, is stored only as a hash, expires, and burns on use. All it can do is approve the one address it was minted for.
Removing an address returns every monitor that was using it to your account address. The panel shows how many monitors each address is currently the recipient for, so you can see what a removal will affect before you do it.