Most articles about the cost of manufacturing downtime quote a single dollar figure with no source and no year. The number is usually wrong for your plant, because the real spread between industries is roughly 60 to 1.
Here is what the best-documented recent dataset actually says — and the finding that most coverage skips entirely.
Cost of unplanned downtime per hour, by industry
| Segment | Cost of one lost hour | Direction since 2019 |
|---|---|---|
| Automotive | $2.3 million (more than $600 a second) | 2× higher |
| Heavy industry | Not published per hour; plant-level cost $59 million a year | 4× higher per hour |
| SME manufacturers (top end) | $150,000 | — |
| Fast moving consumer goods (FMCG) | $36,000 | Stable |
| Oil & gas | — | Down sharply in 2023 as the oil price fell |
At plant level, Siemens puts the cost of an idle automotive production line at a big plant at $695 million a year, 1.5 times higher than five years earlier, and a heavy industry plant at $59 million a year, 1.6 times higher than in 2019.
Aggregated across the world’s 500 biggest companies, unplanned downtime is estimated at $1.4 trillion a year — 11% of their revenues, up from 8% in 2019. Siemens compares that total to the annual GDP of a major industrial nation like Spain.
Source: Siemens / Senseye Predictive Maintenance, “The True Cost of Downtime 2024.” The figures come from 181 completed online interviews with maintenance, engineering and IT professionals at large industrial organizations across automotive, FMCG, heavy industry and oil & gas, covering April 2019 to March 2023. Note that Siemens sells predictive maintenance software; the survey data cited here is distinct from the company’s own product-outcome projections, which are not used on this page.
The finding most coverage misses: recovery is getting slower
The headline numbers get quoted constantly. This one rarely does, and it is the more actionable half of the report.
Manufacturers have genuinely reduced how often production stops:
- 25 downtime incidents a month per facility, down from 42 in 2019
- 27 hours lost per plant per month, down from 39 in 2019
- 326 hours a year per plant, down nearly a third on 2019
- In automotive and heavy industry, hours lost to unplanned downtime have halved over five years
And yet each individual incident now takes longer to resolve:
Five years ago, it took an average of 49 minutes to get production back up and running following downtime. Now, it takes 81 minutes.
That is a 65% increase in recovery time during a period when failure frequency fell by roughly 40%. Siemens attributes it to two causes: businesses losing skilled maintenance labour during the post-COVID “great resignation,” creating a skills and knowledge gap, and global supply-chain issues making emergency replacements harder to source.
The strategic reading is straightforward. Prevention has worked. Response has not kept up.
Why this changes where the next investment should go
If failures are getting rarer but each one costs more hours, the marginal return shifts from preventing the next failure toward shortening the next recovery. Those are different problems with different owners:
- Fewer failures is reliability engineering — condition monitoring, predictive maintenance, preventive maintenance programmes, better MTBF.
- Faster recovery is service execution — how quickly the right knowledge and the right part reach a stopped machine. That is MTTR.
Siemens’ own explanation points at the second one, and specifically at knowledge rather than equipment: the people who knew the machine left, and the expertise did not transfer. No amount of additional sensing fixes that. A sensor can tell you a bearing is failing; it cannot tell a junior technician how to replace it correctly at 2am.
How to calculate your own number
Published averages are useful for scale and useless for a business case — a 60-to-1 spread between FMCG and automotive means the average describes nobody. Build your own:
- Lost production revenue for the downtime period — and be honest about whether the output can be recovered later in the week, because if it can, this term is much smaller than it looks.
- Wages paid to staff who cannot work during the stoppage.
- Salaries of the people responding — maintenance, engineering, and whoever is managing the escalation.
- Knock-on costs that are real but easy to miss: expedited freight, scrapped in-process material, missed delivery commitments and the contractual consequences of them.
Divide the total by downtime hours. Then track it, because this figure is the one that makes service response times legible to a CFO.
If you sell equipment rather than run it, this number is your customer’s number — and it sizes what your response time is worth to them. That is the underlying economics of a service contract, and it belongs in the field service KPI set for exactly that reason.
Where remote visual assistance fits — and where it does not
Being precise here matters more than being enthusiastic.
It does not prevent failures. VSight performs no sensor-based prediction and no IoT condition monitoring. If your problem is failure frequency, this is not the tool.
It attacks recovery time, and specifically the part Siemens identifies as the cause of the regression: getting scarce expertise to a stopped machine. With AR remote assistance, a specialist joins the on-site technician’s live camera view within minutes and annotates the equipment itself, instead of the line waiting for a scheduling and travel window. Sessions run on phones, tablets and smart glasses. VSight Workflow captures the repair as standardized digital work instructions, so the knowledge that walked out of the door is at least written down for the next shift.
What it does not compress is parts lead time. If your recovery clock is dominated by waiting for a component, that is an inventory and supply problem, and it should be measured separately.
To put numbers on the travel and time component for your own operation, the field service ROI calculator works from your inputs rather than ours.
Frequently asked questions
How much does unplanned downtime cost per hour in manufacturing? It depends heavily on the sector. Siemens reports the cost of a lost hour at $36,000 in fast moving consumer goods at the low end and $2.3 million in automotive at the high end — more than $600 a second. SME manufacturers can reach $150,000 an hour at the top end. Any single industry-average figure hides a spread of roughly 60 to 1.
What is the total global cost of unplanned downtime? Siemens estimates unplanned downtime costs the world’s 500 biggest companies $1.4 trillion a year, equal to 11% of their revenues, up from 8% in 2019. For scale, Siemens compares that total to the annual GDP of a major industrial nation like Spain.
Is manufacturing downtime getting better or worse? Both, and they are different metrics. Manufacturers have cut the frequency of downtime — 25 incidents a month per facility, down from 42 in 2019, and 326 hours lost a year, down nearly a third. But each incident now takes longer to recover from: 81 minutes on average, up from 49 minutes five years earlier.
Why is it taking longer to recover from downtime? Siemens attributes the increase partly to businesses losing skilled maintenance labour during the post-COVID period, which created a skills and knowledge gap, and partly to global supply-chain issues that make emergency replacement parts harder to source. In short, the problem has shifted from how often machines stop to how quickly the right expertise and parts reach them.
How do you calculate your own cost of downtime per hour? Add lost production revenue for the period, wages paid to staff who cannot work, and the salaries of the people responding to the incident, then divide by the downtime hours. Published averages are useful for scale but not for a business case — the sector spread is too wide, and your figure depends on line utilization, margin and whether lost output can be recovered later.
Can AR remote assistance reduce downtime costs? It addresses the recovery half of the equation, not the prevention half. Remote visual assistance removes expert travel time from the repair clock by letting a specialist see and annotate the equipment immediately, which targets exactly the expertise gap Siemens identifies behind slower recovery. It does not predict or prevent failures — that requires condition-monitoring infrastructure.
Related reading
Field service KPIs: formulas and benchmarks · uptime and downtime in industrial operations · MTTF and MTTR · OEE · total productive maintenance
Sources
- Siemens / Senseye Predictive Maintenance — The True Cost of Downtime 2024 (181 completed interviews, April 2019 – March 2023)