Cost of Downtime Calculator

About this tool

Enter the hourly revenue that depends on a system, how long a typical outage lasts, how many outages you see in a year, and optionally the hourly cost of staff who cannot work while it is down. The calculator returns the cost of each hour down, the cost of a single outage, and the annual exposure. It runs entirely in your browser — none of the figures you type are sent to a server, which matters when the inputs are commercially sensitive.

The problem it solves is a budgeting one. Redundancy, standby infrastructure and out-of-hours cover all have a visible price, while outages have a cost nobody has written down. Putting the annual exposure next to the annual cost of fixing the weakness turns an argument about risk into an arithmetic comparison. The output is an estimate for planning and internal business cases, not a quote, a forecast or a guaranteed loss figure. For the engineering side of the same question, see disaster recovery.

How to use it

Model one system at a time rather than the whole business:

• Set revenue per hour from the revenue that stops when this specific system stops.
• Use your actual mean time to restore from past incidents, not the target in the runbook.
• Take outages per year from your incident log for the last twelve to twenty-four months.
• For idle staff cost, count only the people who genuinely cannot work, at fully loaded rates.
• Run it twice — today, and the profile you expect after the fix — and compare the difference against the cost of the work.

Common questions

How is the figure calculated?

Revenue per hour plus idle staff cost per hour gives the hourly cost of being down. That is multiplied by the average outage length to get the cost per outage, and by the number of outages per year to get the annual figure. It is a straight linear model, so the output is an estimate for planning rather than a measured or guaranteed loss.

What should I enter for revenue per hour?

Use the revenue that actually depends on the affected system, divided by the hours it needs to be available. For a service that only matters during trading hours, divide by trading hours rather than by 8,760. Where revenue is deferred rather than lost — orders that arrive later anyway — enter only the share you genuinely never recover.

What does this model leave out?

Recovery labour and overtime, SLA credits owed to customers, data loss and the rework to correct it, regulatory or contractual penalties, customer churn and reputational damage. Cost also tends not to be linear: a four-hour outage usually costs more per hour than a twenty-minute one, so treat the result as a conservative floor.