What a minute of downtime costs you

An outage that stops a $5 million online store for one minute costs $14.00 at the settings this page opens with: $9.51 in sales that do not happen plus $4.48 in wages paid to twenty five people who cannot do their jobs. An hour of the same outage runs $839.76, which is closer to the number a manager has to explain the next morning than the per minute figure is.

The question behind this page usually comes from one of three places. An incident review needs a dollar figure for what just happened, so the postmortem says more than "it was down for a while." A vendor conversation needs a service level agreement credit that means something to this specific business, not a boilerplate percentage of a monthly bill. A budget request for redundancy, a second region, or an on call rotation needs a cost to weigh against the spend, since a fix that costs more than the outages it prevents is not worth building. A blog post quoting a single average cost of downtime answers none of these, because that average belongs to whichever company got surveyed rather than to the one asking.

The math behind the number

Two costs run at the same time while a system is down, and this page adds them. The first is lost revenue: take the yearly revenue that runs through the system that failed and spread it across the hours the business earns in, converted to minutes. The second is idle wages: take the headcount that cannot work, multiply by their average salary and the cost basis, and spread that across the same hours, also in minutes. Add the two and the result belongs to this business and this outage.

Worked through at the default: around the clock means 8,760 hours, or 525,600 minutes. Five million dollars over 525,600 minutes is $9.51 a minute of sales. Twenty five people at $65,000 loaded to 1.45 is $2,356,250 of annual cost, and the same 525,600 minutes turns that into $4.48. Together, $14.00 a minute.

One divisor, used twice

Both legs of that sum are divided by the same number of hours, and that is deliberate. Spreading a year of revenue over 8,760 hours while spreading a year of payroll over 2,080 would price a three in the morning outage as though a full day shift were sitting there watching it, which is exactly the failure this page had before September 2026. Choosing around the clock now says the same thing about both sides of the business: revenue arrives at all hours, and the wage bill is an average across all hours rather than a day rate.

Choosing business hours says the opposite about both. A company that sells only during the working day earns its whole year inside roughly a quarter of the calendar's hours, so each of those minutes carries about four times the revenue, and its idle payroll is concentrated in those same minutes and carries about four times the wages. Either choice is coherent. Mixing them is not, which is why there is one control rather than two.

Two outages, two very different bills

A software company where $1.2 million of yearly revenue runs through the system that just failed, selling on business hours, with eight people idled at an average $95,000 salary loaded to 1.45, faces a 45 minute outage costing $830.05: about $433 in lost sales against about $397 in wasted wages, a labor line nearly as large as the revenue line because the team is small and well paid relative to what the business brings in.

A manufacturer doing $180 million a year through the line that stopped, selling only during business hours, with 150 staff idled at an $80,000 average salary loaded to 1.45, faces a very different split on a three hour outage. That outage costs $284,711.54, of which about $259,615 is lost orders and about $25,096 is idle wages. A company that size earns most of its revenue in a narrow set of hours, so it loses far more to stalled sales than to the wages of the people waiting for the system to come back.

What the figure does not include

This total is a floor. It leaves out two things on purpose: customers who quietly leave after a bad outage, and any refunds or credits issued to keep the ones who stayed. Neither has a steady per minute rate, since both depend on how visible the outage was and how the company responded, not on how long a clock ran. It also leaves out recovery work after the system comes back, which is real labor and does not stop when the status page turns green. Add separate estimates for those when a decision needs one. This page answers what happened to sales and payroll while the system was down, not what happens to trust afterward.

Common mistakes

1. Reusing a vendor's average cost of downtime instead of running the numbers for this business, which has its own revenue and its own headcount.
2. Entering the whole company's revenue when only one product line runs through the system that failed.
3. Dropping idle wages from the total because nobody was fired, when the company still paid for hours that produced nothing.
4. Picking business hours for a company that actually earns around the clock, which understates every minute of downtime by roughly a factor of four, or picking around the clock for a daytime business, which understates it by the same factor in the other direction.

For the meeting scheduled to review the outage, the meeting meter on the home page prices exactly that room while it runs, a much smaller number worth setting beside the outage total above.