What uptime measures
Uptime is the share of a period during which your site was reachable and returning pages normally. It is measured by a monitoring service that requests a page from outside your network at a set interval and records whether a response arrived, how long it took, and what status code came back. Downtime is simply the remainder.
Two details decide whether the figure means anything at all. The check interval sets the smallest outage you can detect, so a brief failure falling between two checks never appears. And the nature of the check matters: a server that answers with an error page is technically responding, so monitoring that only asks whether anything came back will happily report a broken site as available.
Why uptime matters
Every visitor arriving during an outage is lost, and the ones who cost money are lost twice. A paid click landing on an unreachable page is charged in full and returns nothing. Campaigns keep spending throughout, because the ad platform has no way of knowing what happened after the click.
There is a search cost as well. When a crawler repeatedly finds a site unreachable it slows down or stops crawling, so new pages and fixes take longer to appear in results, and prolonged outages can affect how pages are treated. There is a trust cost too: someone who meets an error page rarely comes back to try again.
Where uptime goes wrong
The usual mistake is trusting the provider’s own dashboard. Availability guarantees are typically written to exclude scheduled maintenance and any problem the provider judges to be your fault, and a host is not a neutral witness to its own outages. Measure from outside.
The second is monitoring only the homepage. Checkouts, booking forms, login pages and the landing pages behind your advertising can each fail on their own while the homepage stays perfectly healthy. Nothing is more expensive than a working homepage and a broken enquiry form.
The third is measuring nothing until a customer telephones. Short outages overnight are common and go unrecorded, and by the time anyone notices a pattern, the evidence has gone.
What to do about it
Put an external monitor on the pages that carry money — the homepage, the main service or product pages, the checkout, and any landing page behind a paid campaign — and have it confirm that the response contains expected content rather than merely that something came back. Route alerts to a person who will act, not to a shared inbox nobody reads.
Then treat the record as evidence. Compare outages against your host’s claims when renewal comes round, and check whether failures cluster at particular hours, which often points to shared-server load or a scheduled job rather than random misfortune. Consistent trouble is a hosting decision, and uptime monitoring gives you the grounds to make it. Keeping the site patched, backed up and watched is ordinary website maintenance, and it prevents most avoidable downtime.
