Monitoring: how we know your site is down before you do
Monitoring is the difference between an outage and an incident. What we watch on every site, who gets alerted, and why a status page is not monitoring.
The short answer
Monitoring is a set of automatic checks, run from outside your site every minute or so, that confirm the things your business depends on are working, and that alert a person when they are not. It is the difference between an outage that a customer reports on Tuesday and an incident that was fixed before anyone noticed on Sunday night.
It is also broader than “is the site up”. Most of the failures we catch happen while every page loads fine.
What we watch, on every site
| Check | What it catches | How often |
|---|---|---|
| Availability from several locations | The site down, or down for one region | Every minute |
| Content check | The page loads but shows an error, a blank, or a hacked message | Every minute |
| Certificate expiry | A renewal that silently failed | Daily, alert weeks ahead |
| DNS records | A record changed or deleted, name servers moved | Daily |
| Form test | A submission that no longer arrives | Hourly or daily |
| Error rate | Functions or pages failing for some visitors | Continuous |
| Core Web Vitals | The site getting slower over time | Weekly from field data |
| Domain expiry | A renewal about to lapse | Monthly, alert months ahead |
| Security headers | A header that disappeared after a change | After every deploy |
Why uptime alone is not enough
- Forms fail silently. The site is up. The form service changed its API, or the mailbox filled up, or the spam filter got stricter. Every enquiry for three weeks went nowhere. Only a form test notices.
- Certificates expire quietly. Renewal failed a month ago; the certificate is valid until Thursday. Uptime is green until Thursday morning.
- Content changes without a crash. A hacked page with a spam message returns a perfectly healthy response. Only a content check sees that the page no longer contains what it should.
- Slowness is gradual. No single moment is an outage; the site just gets a little heavier every month. Only tracking Core Web Vitals over time shows it.
Alerts that reach a person
Monitoring produces information; the value is in the response. Every check needs an owner, a channel that gets attention, and a routine:
- Critical (site down, certificate expired, content changed): phone notification or call, to a named person, day and night.
- Important (form failing, error rate up, DNS changed): chat message to the team channel, acted on within the working day.
- Advisory (certificate due, vitals drifting, domain renewal approaching): weekly digest, into the maintenance queue.
What this means for you
Ask whoever runs your site two questions: which checks run against it, and who would be woken up at 3 a.m. if the site went down. A good answer lists the table above and names a person. If the answer is “the hosting company has a status page”, you do not have monitoring, and the first outage you hear about will come from a customer.
Frequently asked questions
Our site has never gone down. Do we need monitoring?
You know it has never gone down when someone was looking. Without monitoring an outage is discovered by a customer, hours later, or never. Monitoring is how 'never gone down' becomes a fact rather than an assumption, and it catches the quieter failures, such as a form that stopped sending.
What does a broken form have to do with uptime?
Nothing, which is the point. The site is up, every page loads, and the contact form has been silently failing for three weeks. Uptime monitoring shows green throughout. A synthetic form test, submitting a test message on a schedule and checking it arrives, catches it within the hour.
Who receives the alerts?
A named person with the access to act, through a channel that gets attention: a phone notification or a call for critical alerts, a chat message for the rest. Alerts sent to a shared inbox are read on Monday. That is a log, not monitoring.
Is monitoring expensive?
The tooling is inexpensive. The cost is the routine: someone configuring the checks, receiving the alerts, and acting on them. That is why it belongs inside a maintenance or partnership scope rather than as a separate tool nobody owns.