Uptime

Example

A 30-day month holds about 43,200 minutes. At 99.9 per cent the allowance is 43 minutes of downtime; at 99.5 per cent it is 216 minutes; at 99 per cent it is 432 minutes, over seven hours. Across a year, 99.9 per cent allows roughly 9 hours with no working site and 99 per cent roughly 88.

The number becomes legible once it is multiplied by trading. In a shop turning over €5,000 a month an hour is worth about €6.90 on average, so seven hours down is close to €48 of sales missed — small, until the outage lands on a campaign or a peak day, when an hour is worth several times that. The amounts are illustrative.

This is why any uptime promise also says how it is measured: from how many locations, at what interval, what counts as down (a 500 error, a timeout, a response slower than an agreed threshold), and which interruptions are excluded.

Why it matters for a business

A site that is down costs more than the orders missed while it is down. Advertising that keeps running spends budget on a page nobody reaches. A visitor who meets an error rarely returns the same day. And a crawler that finds the site unreachable several times in a row slows down how often it comes back.

The percentage is only half the story. The other half is how quickly somebody noticed: a site can be down all night and still look like 99.9 per cent on paper if nobody is watching. Checks from outside, and an alert when the site stops answering, are part of website maintenance rather than an extra alongside it.

What to ask

• Who measures uptime: the host, the maintenance supplier, or an independent tool?
• How often are the checks, and from how many locations?
• What counts as downtime — errors only, or responses slower than a threshold?
• Who receives the alert when the site stops, and during which hours?
• Is planned maintenance counted in the percentage, and how is it announced?
• Is there a monthly report with the history of interruptions?