Every maintenance budget argument eventually loses to the same question: "what does an hour of downtime actually cost us?" The honest answer comes from your own margin and staffing, not from a consultant's headline. Build it from the pieces below and you'll have a number that survives a finance meeting — and a per-hour rate that turns "we should fix that" into a fundable project.
The core is two recurring rates plus one event charge:
Event cost = hours × (units/hr × margin per unit + idled workers × loaded hourly cost) + restart & scrap costs
Lost margin counts what the line would have earned: units per hour times contribution margin, which is price minus variable cost, not price. Idle labor counts the crew stuck waiting: headcount times loaded hourly cost, where loaded means wages plus benefits and payroll taxes, typically 1.3-1.5 times base wage. Restart costs land once per event: scrap, expedited freight, overtime catch-up, and consumed spares.
A failed drive stops the line for 3 hours. The line makes 1,200 units an hour at a $2.10 contribution margin — $2,520 an hour of lost margin. Fourteen staff are idled at a $28 loaded rate, adding $392 an hour. Blended rate: $2,912 an hour, about $48.53 a minute. The restart scrapes $1,800 of product. Total: 3 × $2,912 + $1,800 = $10,536 for one event.
Now the budget-scale view: at 25 events like it a year, that line bleeds $263,400. A condition-monitoring retrofit or a critical-spares program that prevents a third of them is worth $87,800 a year — which is how maintenance requests get written in a language capital committees already speak.
| Source | Finding | Figure |
|---|---|---|
| ITIC 2024 Hourly Cost of Downtime survey | Mid-size and large enterprises losing >$300,000/hour | >90% of firms |
| ITIC 2024 | Enterprises at $1M-$5M/hour | 41% |
| Siemens/Senseye, True Cost of Downtime 2024 | Automotive unplanned downtime (highest sector) | ~$2.3M/hour |
| Siemens/Senseye 2024 | Downtime losses, Fortune Global 500 industrials | ~11% of revenue (~$1.4T) |
Useful for context, useless for budgeting. The ITIC sample skews toward IT and services; the Siemens automotive figure reflects highly automated, interdependent lines at massive volume. Your number is the one that defends a purchase order.
Three lenses on the same events. MTBF tells you how often failures arrive; OEE tells you what share of planned time survived; downtime cost tells you what the losses are worth in dollars. A plant that tracks all three can rank its bad-actor assets by annual dollars rather than gut feel, which is exactly the ranking a limited maintenance budget should follow. The math chains directly: cutting failure frequency raises MTBF, which lifts availability, which raises OEE, which shrinks the annualized downtime bill.
Units, margin, crew size, and restart costs in — per-hour rate, event cost, and annualized exposure out.
Downtime Cost Calculator →An hour of downtime costs its margin, its idle labor, and its restart mess — nothing more, nothing less. Compute the rate once with honest numbers, annualize it against your event count, and you've converted maintenance from a grudge line item into a measurable return. The companion pieces are our guides on MTBF and availability and OEE benchmarks, which cover the reliability side of the same ledger.
Because the headlines are measured on someone else's operation. ITIC's $300,000-plus figure comes largely from IT-dependent enterprises, and Siemens' $2.3 million per hour is automotive at the extreme. A mid-size plant with modest margins and a few dozen staff per line will land far lower, and that's fine — the comparison that matters is your annualized downtime cost against your maintenance budget, not against Toyota's stamping line.
Planned stops are schedulable: changeovers and PM windows can shift to off-peak hours, staff can be redeployed, and there's no restart scramble. Unplanned stops arrive at random, consume overtime-rate labor, often damage tooling, and create scrap on restart. Both cost money, but the unplanned version carries premiums — and that premium is exactly what preventive and predictive programs sell against.
Three sources cover it: the CMMS or maintenance log for event frequency and duration, production reporting for units per hour and scrap on restart, and payroll for loaded labor rates. If the data is thin, run the calculation twice with optimistic and pessimistic assumptions — a range still beats a guess, and it tells you which assumption to nail down first.