Most sites are monitored by their customers. Five layers of monitoring — uptime, certificate expiry, form delivery, performance and error rate — cover the failures that actually cost money.
The most common way a business finds out its website is broken is a customer email. These five layers cost very little and remove most of that risk.
1. Uptime, from more than one location
A check every minute from at least two regions, hitting a real page rather than the home page alone. Single-location checks generate false alarms from network conditions at the checking location, which is how teams learn to ignore alerts.
2. Certificate and domain expiry
TLS certificates and domain registrations both expire, both are usually on auto-renew, and both occasionally fail to renew — normally because the card on file changed. Alert at 30, 14 and 7 days. This is the highest-consequence, lowest-effort monitor on the list.
3. Form and transaction delivery
Uptime checks confirm the page loads. They do not confirm that the contact form still sends an email, or that the checkout still completes. A synthetic transaction that submits a real test entry daily and alerts on failure catches the outage nobody would otherwise notice — a broken form is silent by definition, and can run for weeks.
4. Performance in the field
Real-user monitoring of Core Web Vitals, so a regression appears when it ships rather than when the rankings move a month later. If you run nothing else, the web-vitals library reporting into your analytics is a few lines of code.
5. Error rate and log alerting
Server errors and JavaScript exceptions, alerting on rate rather than on individual events. A 500 error on one request is noise; a hundred in ten minutes is an incident. Error tracking is also the fastest way to discover that a third-party script started failing in one browser.
The discipline that makes it work
Every alert needs a named owner, a channel someone actually reads, and a defined action. An alert that fires into an unmonitored inbox is worse than no alert at all, because it creates the belief that someone is watching.
Review the alert log monthly and delete anything that has fired without ever requiring action. Alert fatigue is how monitoring stops working, and it happens quietly.
Looking for more? Browse all resources.