Website Uptime Monitoring: What to Check and Why

Website uptime monitoring checks whether a website is available from outside your own network. A monitoring service requests a page at regular intervals, evaluates the response, and alerts someone when its failure conditions are met.

For a small team or an agency managing client sites, the useful question is not just “Is the server on?” It is “Can people reach the pages they need, and will someone respond if they cannot?” Good monitoring starts with those pages, clear checks, and a named owner.

Website uptime is not the same as server uptime

A server can answer a ping while its website returns an error. A web page can return HTTP 200—the usual success status—while showing an empty template or a maintenance message. Server uptime monitoring and website monitoring answer related but different questions.

An HTTP check tests a web response. Adding an expected-text check can help catch a page that answers but lacks important content. Neither necessarily runs JavaScript, renders the page like a browser, or completes a customer journey. A working sign-in page does not prove that signing in works.

Likewise, an open database or mail-server port does not prove that a customer can finish a task. If you need to verify a multi-step checkout or booking, use a separately validated browser or transaction test. Record that coverage gap rather than assuming an availability check covers it.

Choose checks around the failures you care about

Start with a short list of important pages and their dependencies. For each, write down what passing means:

  • HTTP or HTTPS availability: Can the homepage and another important page respond successfully within your chosen timeout? Review redirects so you know which destination is being checked.
  • Expected text: Does the returned response contain a stable phrase that belongs on the correct page? Avoid timestamps or other changing content. Text generated only after JavaScript runs may not be present in an HTTP response.
  • DNS: Does the domain resolve to the records you expect? This checks a dependency of reaching the site, not whether the application works.
  • TLS certificate and domain expiry: Add expiry checks to help identify renewal problems before they interrupt access. They complement availability checks rather than replace them.
  • Server or port checks: Watch supporting services where useful, but interpret “port answers” narrowly.
  • Email round-trip, if relevant: When email is important to operations, a send-and-receive check tests more than an open mail port. It still does not prove that every customer's message reaches every inbox.

For example, an agency might monitor a client's homepage response and a stable heading on its contact page. That could catch an unavailable site or missing page content. It would not prove the contact form submits successfully; that needs a separate, safe test with a designated test destination.

Pick an interval and understand the delay

The check interval determines how often the service looks. With five-minute checks, a failure immediately after a successful check may wait almost five minutes for the next scheduled attempt. A brief outage between checks can go unseen.

Shorter intervals can reduce that sampling delay, but they do not guarantee an alert within the interval. Request timeouts, failure confirmation, notification delivery, and the time it takes a person to respond all matter.

Use checks outside your infrastructure, and consider the locations your visitors depend on. A second location can help distinguish a local network issue from a broader failure; it cannot eliminate every false alarm or detect every regional problem. BinaryCanary's monitoring process checks a failure from a separate node on a different network before sending an alert.

What do uptime percentages actually mean?

An uptime percentage needs a defined period and measurement method. For an elapsed-time calculation with complete coverage:

Uptime % = (total period seconds − downtime seconds) ÷ total period seconds × 100.

Define the period's start, end, and timezone. For this example, the denominator includes all elapsed time, including planned maintenance; any unavailable time counts as downtime. If your agreement excludes maintenance, disclose that separately rather than comparing unlike percentages.

Monitoring samples do not reveal every exact failure or recovery instant. Explain how check timestamps become estimated downtime. If monitoring was paused or records are missing, mark that interval as unknown—not automatically up. Without complete coverage, report the gap alongside any observed-coverage percentage; do not present it as a fully measured whole-period result.

Here is the downtime equivalent of four availability targets:

Availability target Downtime in 30 days Downtime in 365 days
99%
99.9%
99.95%
99.99%

These calculations use exactly 30 or 365 days of , with no exclusions. Values are rounded to the nearest second; the 99.99% equivalents are exactly and . They are planning figures, not a service-level guarantee or a report of BinaryCanary's measured uptime. Real calendar months vary in length.

A high percentage can also hide an inconvenient outage. Review individual incidents and affected pages, not just the monthly total.

Make the alert lead to a response

Give each important monitor a primary owner and a backup. Decide which notification routes they will use, when the backup takes over, and where incident notes live. Agencies should make client ownership clear rather than letting every alert land in one undifferentiated inbox.

Verify the setup with a test notification if your tool provides one, or a controlled test endpoint that cannot affect customers. Do not break a production site to see whether monitoring works. Confirm receipt, acknowledgment, and escalation—not just that an alert was generated.

When an alert arrives, verify the problem from another network if possible, record what is affected, and follow your runbook. After recovery, check the original failure condition and safely verify the important user journey. The outage-readiness checklist helps turn these decisions into a written response plan.

Start with a few meaningful checks

Begin with your most important pages, checks that match their failure modes, and people who know what to do next. Review coverage after changes to your site or team.

BinaryCanary supports HTTP/HTTPS and text checks, DNS and expiry checks, server monitoring, and optional email round-trip monitoring. Its published plans include checks as often as every minute; notification timing also depends on confirmation and delivery. Escalation profiles and client sub-accounts help organize responsibility.

Explore the monitoring features, review plans and pricing, or start a 14-day free trial.