What does an hour of downtime really cost?

Three components, not one

The cost of an outage breaks down into three families: the time of people prevented from working, the commercial contribution lost during the outage, and the direct costs spent getting back to normal.

The sum gives an order of magnitude, not an invoice. It is there to compare two architecture scenarios, not to establish a claim.

The most useful rule: do not count twice

The most common mistake is adding the cost of the sales team’s time to the margin lost on the deals they could not close. In many cases the second already includes the first.

Another classic double count: “lost” orders that are in fact delayed. A customer who orders two hours later has cost nothing, unless they went to a competitor. Separate deferral from outright loss.

The cost depends on the moment

An hour of downtime on a Tuesday at 10 am and an hour on a Sunday at 4 am do not cost the same. For a retail site, the gap can exceed a factor of ten. Calculate over the periods you actually care about: the peaks, not the average.

That reasoning changes operational priorities: it sometimes justifies a night maintenance window rather than expensive redundancy, or the opposite for a service used around the clock.

From the figure to the decision

Once the order of magnitude is set, compare it to the annual cost of the options considered: more frequent backups, a standby environment, contracted on-call, monitoring. The calculation is only interesting next to a quote.

1. Estimate the hourly cost of downtime over the most exposed period.
2. Estimate a realistic outage duration, based on past incidents.
3. Compare it to the annual cost of whatever would shorten that duration.
4. Decide — and write the decision down, including the decision to do nothing.

An imperfect but shared figure beats an intuition: it turns a worry into a decision.