DD DevDome Track. Protect. Grow.
Site Care

WordPress Site Down? How to Know Before Your Visitors Do

Most outages are discovered by a customer email. Uptime checks, real traffic as proof of life and an alert that reaches you fix that for free.

On this page
  1. Define down properly
  2. Check from outside, on a schedule
  3. Use real traffic as proof of life
  4. Watch the things that fail quietly
  5. Make the alert reach you
  6. When the alert fires
  7. False alarms and how to stop them
  8. The short version

A WordPress site down for three hours on a weekday afternoon is not rare, and the usual way the owner finds out is an email from a customer. The fix is boring and free: an outside check on a schedule, a definition of down that includes the quiet failures, a proof-of-life signal from real traffic, and an alert that reaches you. Here is how to set each piece up and what to avoid.

Define down properly

A server that answers is not a site that works. Count all of these as down:

  • 5xx errors from the server or PHP.
  • The database connection error, which returns a page with a 200 status and a message nobody wants to see.
  • The white screen, a PHP fatal with nothing rendered.
  • Timeouts, where the page eventually arrives but no visitor waited.
  • Certificate problems, where browsers show a warning and most people leave.
  • Hacked redirects, where your page loads and then sends the visitor elsewhere.
  • DNS failures, where the domain does not resolve at all.

A monitor that only checks for a 200 status misses half of this list. Look for one that checks content for an expected keyword, follows redirects and recognises the WordPress error pages.

Check from outside, on a schedule

An external monitor requests your homepage, and ideally one inner page, every one to five minutes from one or more locations. If a check fails, it should retry once or twice before declaring an outage, because a single blip from one network is normal.

The interval matters less than people think. One minute catches an outage a few minutes sooner than five; what matters more is that the check runs at all and that it tests the right things. If your host is small, five minutes is kind to it.

Use real traffic as proof of life

An external check is one request from one place. Real visitors are hundreds of requests from everywhere. If your analytics show human pageviews in the last fifteen minutes, the site is up for real people, whatever a single check says. If a site that normally has visitors every minute shows none for twenty minutes, something is wrong even if the check passes.

DevDome Analytics shows live visitors and a human versus bot split, which makes this proof-of-life readable at a glance: humans arriving means the site is serving people. DevDome Site Monitor builds on that with uptime, performance and error monitoring, Core Web Vitals from real measurements and email alerts, giving each site one health score. We make both, so weigh that; the principle of combining an outside check with real-traffic evidence applies to any setup.

Watch the things that fail quietly

TLS expiry. Certificates expire on a schedule and renewals fail silently when a DNS or host setting changes. Monitor the expiry date and alert two weeks out.

DNS. A domain that stops resolving is down everywhere and the server has no idea. Check that the name resolves to the expected address.

Response time trend. A site that took 400 milliseconds last month and 3 seconds this week is on its way down. The trend is the early warning; the outage is the late one.

Scheduled tasks. WordPress cron that stops running breaks backups, scans and scheduled posts without any visible error. A health check that confirms cron ran recently catches it.

Make the alert reach you

An alert in an inbox you check twice a day is a report, not an alert. Use email plus one channel you actually see: a phone notification, a chat message, an SMS for the sites that pay the bills. Require a confirmed outage, two failed checks a minute apart, before the alert fires, so you are not woken by a blip.

Then write down what to do when it fires. Who to call at the host, where the backups are, how to roll back the last update. Outages at 2 a.m. are not the time to look that up.

When the alert fires

The first minute decides whether an outage is short. Open the site yourself in a private window to confirm, then look at the error rather than guessing: a 5xx with a host status page incident is the host’s problem and your job is to wait and tell customers; a database connection error is usually the host too, unless you just changed credentials; a white screen after an overnight auto-update is a plugin, and the activity log names it; a certificate warning is a renewal that failed; a redirect to a spam site is a hack and starts the incident process. Having each of these mapped to a first action, written down in advance, is what turns a panicked hour into a ten-minute fix. Keep the host’s support contact and your backup location in the same note.

False alarms and how to stop them

A monitor that cries wolf gets muted, which is worse than no monitor. The usual causes are a single network blip from one checking location, a check interval so short the host’s rate limiting trips it, a keyword check that fails when a cache serves a slightly different page, and a bot protection rule that blocks the monitor itself. Fix them in order: require two consecutive failures a minute apart before alerting, check from at least two locations, choose a keyword from the footer that every page variant carries, and exempt the monitor’s addresses from any blocking you run. After that, every alert means something, and you will keep reading them.

The short version

  1. External check every one to five minutes, testing content, not just status.
  2. Treat database errors, white screens, certificate warnings and redirects as down.
  3. Confirm before alerting; alert on a channel you see.
  4. Use live human traffic as proof the site is serving people.
  5. Monitor certificate expiry, DNS, response time trend and cron.

Set up once, and the next outage is something you fix before anyone emails you about it.

Disclosure: DevDome publishes this blog and makes the products it mentions. We describe what they do rather than promise results.

Key takeaways

  • Down is more than a server error: a white screen, a database connection error, an expired certificate and a hacked redirect all count.
  • Check from outside, on a schedule, and confirm before alerting so a blip does not wake you up.
  • Real human traffic in the last few minutes is proof the site is up; its absence on a normally busy site is a signal.
  • Monitor the pieces that fail quietly: TLS expiry, DNS, and the response time trend.
  • An alert is only useful if it reaches you where you are. Email plus one more channel.

Sources

  1. WordPress.org: Common WordPress errors : the white screen, database connection and other failure modes
  2. MDN: HTTP response status codes : what 5xx codes mean
  3. Let's Encrypt: certificate lifetimes : why certificate expiry is a common silent outage

Links last checked October 17, 2026.

Frequently asked questions

What does WordPress site down actually mean?

Anything that stops a visitor seeing the page they asked for: a 5xx server error, the database connection error, a white screen from a PHP fatal, a timeout, an expired certificate warning, or a hack that redirects visitors elsewhere. A monitor should treat all of them as down.

How often should uptime be checked?

Every one to five minutes from outside the site. A check every minute catches an outage within a minute; every five minutes is enough for most sites and avoids load on tiny hosts. Confirm with a second check before alerting.

Why did my monitor say up when the site showed an error?

Because it only checked that the server answered with a 200. An error page can return 200. Better monitors check the content for a keyword, follow redirects and detect the database error and white screen patterns.

Do I need a paid uptime service?

No. Free monitors cover uptime checks at reasonable intervals. Paid tiers add shorter intervals, more locations and status pages. A WordPress-side plugin can add proof-of-life from real traffic and health signals the external check cannot see.

DevDome Team WordPress plugin builders

The DevDome team builds lightweight, performance-first WordPress plugins and free tools for site owners, founders and marketers. We write about the exact problems our plugins solve, in plain language for people who run real sites.