Uptime Monitoring: What It Checks and What It Doesn't
In short
Uptime monitoring requests your website from the outside at fixed intervals and alerts you when it doesn't answer or answers wrongly. Intervals between one and five minutes are common. A simple HTTP check only tells you the server responds. Whether forms, the shop checkout or scheduled tasks work is invisible to it. That takes keyword checks, heartbeat monitoring and separate tests for forms.
Most website outages aren’t noticed by the owner but by a customer. They call, send an email or quietly go to a competitor. Uptime monitoring turns this around: a service checks your site at fixed intervals and gets in touch when something is wrong. Below is which checks exist, how to set them up and where the limits are.
What uptime monitoring is
Uptime monitoring requests your website from the outside, just like a visitor. A server somewhere on the internet sends a request every few minutes and evaluates the answer. No answer, a slow answer or an error code means the site counts as down. Then a notification goes out.
“From the outside” is the key part. A WordPress plugin can’t report its own outage: if the server is down, the plugin isn’t running either. Monitoring therefore has to run on someone else’s infrastructure, separate from your hosting.
A short definition is also in our glossary under uptime monitoring.
Which checks exist
“Monitoring” is an umbrella term. Behind it are different kinds of checks that find different errors.
HTTP check. The basic form. The service requests a URL and checks the status code. 200 means reachable, 500 or no answer means down. This finds crashed servers, DNS problems and broken configurations.
Keyword check. The service loads the page and looks for a specific word in the content. If it’s missing, it raises an alarm. This matters more than it sounds. Many errors still return status 200, for example a blank page, a maintenance notice that got stuck, or a page suddenly showing foreign content. Choose a word that only appears when the page works, such as a piece of text from the footer.
SSL certificate. The service checks when the certificate expires and warns days in advance. An expired certificate is practically an outage for visitors: the browser shows a warning page that hardly anyone clicks through. This is becoming more important. Under the rules of the CA/Browser Forum, public TLS certificates have been limited to 200 days since March 2026, will be limited to 100 days from March 2027 and to 47 days from March 2029. More renewals means more chances for an automatic renewal to fail silently.
Domain expiry. Some services also check when the domain registration runs out. That happens rarely, but thoroughly when it does: without the domain, website and email are gone at the same time.
Heartbeat. Here the direction is reversed. The service doesn’t request your site, your site reports to the service at regular intervals. If the report doesn’t arrive, there’s an alarm. This suits everything that is supposed to run in the background: cron jobs, backups, imports. For WordPress, it’s the way to notice that scheduled tasks have stopped running. How to replace WP-Cron with a real cron job is described in our article Disable WP-Cron.
Response time. Most services measure how long the answer takes along the way. A gradual increase over weeks is often an earlier warning sign than the outage itself, for example with a growing database.
Why “200” doesn’t mean everything works
A green status in the monitoring says: the server answered. It doesn’t say that your website does its job.
Typical cases an HTTP check doesn’t see:
- The contact form stops sending emails. The page with the form loads perfectly. Only when someone submits it does it turn out that sending has been failing for weeks, for example because the mail server credentials changed or the messages land in the recipient’s spam folder.
- The shop checkout hangs. The homepage and product pages are reachable, but the payment interface doesn’t respond.
- Scheduled tasks don’t run. Backups, newsletters and scheduled posts don’t happen, while the site itself is reachable.
- The site serves malicious code. A hack doesn’t change the status code.
If you take forms seriously, you therefore need a separate check that actually submits the form and verifies that the message arrives. That is more work than an uptime check, but it covers the error that costs the most enquiries and goes unnoticed the longest.
Check interval and locations
Interval. Intervals between one and five minutes are common. Five minutes is usually enough for a business website. For shops and applications where every minute counts, one minute makes sense. Going below one minute rarely helps, because someone still has to respond.
Locations. Good services check from several locations and only raise an alarm when several locations confirm the outage. That prevents false alarms caused by a network problem between a single check server and your hosting. For a website with an audience in Europe, at least one location should be in Europe.
Confirmation before the alarm. A single lost request isn’t an outage. Set the service to check again before alerting. Otherwise you get used to false alarms and miss the real one.
Pause during maintenance. When you install updates or move servers, the site is briefly gone. Most services allow maintenance windows during which no alerts are sent. Use them, and switch them off again afterwards. A forgotten maintenance window is monitoring that isn’t running.
Reading availability percentages correctly
Monitoring services and hosts like to state availability as a percentage. The numbers all sound high, the difference lies in the downtime they allow. Calculated for a month with 30 days:
| Availability | Allowed downtime per month |
|---|---|
| 99% | 7.2 hours |
| 99.5% | 3.6 hours |
| 99.9% | 43.2 minutes |
| 99.99% | 4.3 minutes |
Two things matter here. First, your monitoring only measures as precisely as it checks: with a five-minute interval, a four-minute outage can slip through. Second, a host’s availability commitment is no guarantee that your site is running. It applies to their server, not to your WordPress, your plugins or your mail delivery.
Monitoring many websites
If you look after several websites, for example as an agency or as a company with several brands, you quickly run into questions of your own. A few rules that have proven useful:
- Name consistently. Client, website and check type in the name of every check, otherwise you’re searching for the right site at night.
- Group alerts. If several sites sit with the same host, they often go down together. A service that groups related alerts spares you a flood of identical messages.
- Record responsibility. For every site it must be clear who responds and whom they inform at the client.
- Clean up regularly. Websites that have been shut down belong out of the monitoring, otherwise the team becomes numb to permanent alarms.
Where the alarm goes
An alarm nobody reads isn’t one. Clarify three questions beforehand:
- Who gets it? Not only the person who built the website, but the one who can also respond or bring someone in.
- Through which channel? Email is the standard but gets buried. For critical sites a second channel makes sense, such as an app notification, SMS or a team chat channel.
- What happens next? Who looks at the error, who calls the host, who decides on a maintenance page? Without this agreement, the alarm only costs nerves.
Many services also offer a public status page. It’s worth it when customers or staff depend on your application. For a regular business website it’s usually unnecessary.
Which kinds of providers exist
There are roughly three kinds of monitoring:
- Specialised uptime services. They do nothing but reachability, certificates and heartbeats. Well-known names include UptimeRobot, Better Stack or Uptime Kuma, the latter as software you host yourself. For websites, this is usually the right class.
- Monitoring by the host. Some hosts monitor their servers. That helps, but only sees their own infrastructure, not how your site looks from the outside.
- Large observability platforms. They monitor servers, logs and applications in detail. For a business website that’s usually too much, for your own application it can fit.
When choosing, look at where the check servers are located, whether the service can check again before alerting, and where the service stores its data.
What monitoring doesn’t replace
Monitoring reports, it doesn’t repair. When the alarm comes, the damage is already done. What prevents or limits it runs independently:
- Updates close holes before someone exploits them.
- Backups with a tested restore decide how quickly the site is back after an outage. Our article WordPress backups done right describes what matters.
- Security checks find malicious code that no uptime check sees.
Seeing what the server answers yourself
When an alarm comes in, or you simply want to know what your server answers, our HTTP status checker helps. It shows status code, redirects, response time and headers for up to 20 URLs. It doesn’t replace monitoring, because it only checks when you run it.
How we handle it at Codeäffchen
Uptime monitoring is part of every package in our WordPress maintenance. Depending on the package, we check every five minutes or every minute, and the larger packages add heartbeat monitoring of scheduled tasks. Monitoring runs around the clock. We respond within our service hours, Monday to Friday from 8 am to 6 pm, within the response time of the respective package. Outside of them we can usually be reached, but we don’t promise it.
You want to know whether your website is quietly failing without you noticing? Find out more on the WordPress maintenance page, or drop us a line.
Frequently Asked Questions
What is uptime monitoring?
A service that requests your website from the outside at fixed intervals, usually every one to five minutes. If it doesn't answer, answers too slowly or returns an error code, it sends a notification by email, SMS, app or chat. That way you learn about an outage before customers call.
How often should a website be checked?
Five minutes is enough for most business websites. For shops and applications where every minute of downtime costs revenue, a one-minute interval makes sense. Shorter intervals bring more false alarms if the service doesn't check again before alerting.
Is an HTTP check enough for a WordPress shop?
No. An HTTP check only sees that the server responds. A page with a database error or a blank page can still return status 200. Shops need at least a keyword check for a word that only appears when the page works, plus monitoring of scheduled tasks.
Does monitoring replace maintenance?
No. Monitoring reports a problem, it doesn't fix one. Updates, backups with a tested restore and security checks run independently. Without someone who responds to the alert, monitoring is just bad news delivered earlier.
Related Services
Sr. Full-Stack Developer
Frontend & Backend. Builds scalable web applications with a focus on clean architecture and maintainability.