Uptime monitoring for a small website: what to watch and when to alert
Monitoring tells you the site is broken; alerting wakes someone to fix it. The checks worth having for a small site, and how to keep alerts worth answering.
If your website goes down tonight, will you find out from an alert or from a customer? Uptime monitoring is how the answer becomes "an alert". This guide covers what monitoring actually watches, which checks matter for a small site, and how to avoid alert fatigue - grounded in Google's Site Reliability Engineering book and your own provider's status tooling.
What monitoring is (and is not)
Google's SRE book defines monitoring as collecting and presenting data about a system so you can tell when it is broken - or about to break. Alerting is the part that notifies a human to investigate. The distinction matters: dashboards and uptime percentages describe the past, while an alert exists to demand action now. Our SLA guide covers what uptime percentages actually mean; this guide is about the watching itself.
The checks worth having
- An external uptime check. A service outside your hosting requests your homepage (and any critical page) on a schedule you configure - often every few minutes as a starting point - and alerts when checks fail. This can catch full outages on the checked pages regardless of what your server reports about its own health.
- Certificate expiry. HTTPS certificates expire on a schedule, and an expired certificate makes a site unreachable much like a crashed server does. Monitoring services commonly offer expiry warnings ahead of time.
- Key user journeys. If your business depends on a checkout, booking or login, a synthetic check that walks that flow catches failures a homepage ping misses.
Symptoms versus causes
The SRE book draws a line that small teams can use too: alert on symptoms users feel - the site is down, the checkout fails - rather than on every internal signal. A slow database might matter, or might be fine; a checkout that fails matters to every customer who hits it. Alerting on causes creates noise, and noise teaches people to ignore alerts, which is how genuine alerts get missed.
Who gets woken up, and for what
Decide the severity ladder before the incident: which failures page someone immediately (site down, payment path broken), which become a ticket or email for the morning (a slow page, a non-critical error). Google's guidance is blunt: an alert should mean a human needs to act. If nobody would act on it, it belongs in a dashboard, not in someone's phone at 3am.
Getting started cheaply
You do not need an SRE team. Free and low-cost monitoring services can ping your site and email or message you on failure; your hosting provider or CDN may expose status and health data as well - our hosting and CDN guides cover what those platforms typically include. Start with one external check and a certificate expiry warning, see how often they fire, and expand only when a gap actually bites you.
FAQ
How often should my site be checked?
There is no single standard: the interval is a configuration choice, and defaults, limits and granularity vary by tool. As one example, a setup might check every few minutes and alert only after two or three consecutive failures, since shorter intervals can detect outages sooner but also produce more noise from brief network blips.
Is 100% uptime realistic?
No - downtime is possible even with an uptime guarantee in place, and a guarantee does not mean outages cannot happen. What is covered, and what remedy applies, varies by provider and plan, so it is worth reading the actual terms, as our SLA guide explains. Monitoring does not prevent outages; when the checks, the alerts and their delivery all work, it shortens the time until someone knows and responds.
Should I monitor from inside my own server?
Internal checks tell you how the server itself is doing, and an internal synthetic test can also call your public endpoints. What one internal vantage cannot show is whether visitors on other networks can reach you - DNS, certificates and routes can fail between them and your server. An external monitor tests a chosen location, endpoint or journey from outside, which complements internal telemetry rather than replacing it.
What about status pages?
A public status page is the communication complement to monitoring: when an incident happens, it gives customers a place to check instead of emailing you. Some monitoring tools include one, and it earns its keep the first time a real outage hits.