The Real Cost of Factory Downtime (and How to Measure It)
Downtime costs more than the stopped machine. Measure it per hour as: lost output × contribution margin, plus idle labour, plus overhead that keeps burning. A mid-size Indian line losing 120 parts/hour at ₹85 margin, with 6 idle operators and running overhead, bleeds roughly ₹12,000 every downtime hour — and most of that never reaches a report.
Below is the exact method, a worked ₹ example, how it ties to OEE, and why so much of the loss stays hidden.
Why the stopped machine is the smallest part of the bill
When a line stops, most owners picture the repair cost. That is the cheapest part. The real bill has three parts that run at the same time:
- Lost output you can't sell — every good part not made is margin you'll never recover this shift.
- Idle labour — operators are paid whether the line runs or not.
- Overhead that doesn't pause — rent, depreciation, supervision, baseline power and interest keep accruing while nothing ships.
Globally this adds up fast. Siemens' True Cost of Downtime 2024 estimates the world's 500 largest companies lose about US$1.4 trillion a year to unplanned downtime — about 11% of combined revenue (Siemens / Senseye, 2024). You don't run a Fortune 500 plant, but the structure of the loss is identical on a 300-worker floor in Pune or Coimbatore.
The formula: downtime cost per hour
Use this as your standard unit. Everything else scales from it.
Downtime cost per hour = (Lost good units/hour × contribution margin per unit) + idle labour cost/hour + allocated overhead/hour
Two definitions that matter:
- Contribution margin (not gross price): selling price minus the variable cost to make one unit (material, consumables, per-unit power). This is what you actually lose per unfinished part — you don't spend the material on a part you never made, so don't double-count it.
- Idle labour: the fully-loaded cost of operators on that line who are paid but producing nothing during the stop (wage + PF/ESI + allowances).
A worked ₹ example (auto-components CNC line)
Take a mid-size auto-components plant running a CNC machining cell. Figures below are indicative July 2026 estimates for illustration — plug in your own.
| Input | Value | Basis |
|---|---|---|
| Ideal good output | 120 parts/hour | line design rate |
| Contribution margin/part | ₹85 | sale price − variable cost |
| Operators on the cell | 6 | one shift crew |
| Fully-loaded labour/operator | ₹120/hour | ~₹563/day ASI factory wage + statutory + overhead loading¹ |
| Allocated overhead to the cell | ₹1,500/hour | rent, depreciation, supervision, baseline power |
One hour of downtime:
- Lost margin: 120 × ₹85 = ₹10,200
- Idle labour: 6 × ₹120 = ₹720
- Overhead still burning: ₹1,500
- Total ≈ ₹12,420 per downtime hour
Now annualise. Suppose the cell runs two shifts, 26 days a month (~416 planned hours/month), and loses 10% to downtime — a mix of breakdowns, changeover overruns, material waits and micro-stops. That's ~42 lost hours/month:
- 42 × ₹12,420 ≈ ₹5.2 lakh/month ≈ ₹62 lakh/year — on one cell.
Even if your real numbers are half of this, the point holds: downtime is a seven-figure annual line item that rarely appears as one.
¹ India's Annual Survey of Industries put the average factory daily wage at about ₹563 in 2021–22 (CEDA, Ashoka University, on ASI data); we load it up for statutory contributions and overhead. Use your own payroll figure where you have it.
How this ties to OEE
OEE (Overall Equipment Effectiveness) is the standard way to express these losses as one percentage:
OEE = Availability × Performance × Quality (OEE.com)
- Availability — losses from the line being stopped (breakdowns, changeovers, no material).
- Performance — losses from running slow or micro-stops (the line "runs" but below rate).
- Quality — good parts vs. total (scrap and rework).
The widely cited world-class benchmark is ~85%, a target rooted in the TPM tradition (Lean Production). Illustratively, that 85% is usually broken out as roughly 90% availability × 95% performance × ~99% quality — the exact split varies by source and industry. In practice, plants that actually measure typically sit well below world-class; a 60–75% band is commonly cited as indicative for discrete manufacturing (treat it as a rule of thumb, not a hard statistic — measure your own). The gap between your OEE and 85% is your downtime cost, expressed as a ratio. A jump from 65% to 75% OEE on the cell above is worth roughly a third of that ₹62 lakh back.
Why most downtime stays hidden
Here's the trap. Manual downtime logs capture the big, obvious stops — a breakdown someone had to call maintenance for. They miss the losses that quietly eat the most:
| Downtime type | Typically logged? | Where it hits OEE | Why it hides |
|---|---|---|---|
| Major breakdown | Yes | Availability | Loud, someone raises a ticket |
| Changeover overrun | Sometimes | Availability | Treated as "normal", not timed |
| Waiting for material/operator | Rarely | Availability | No single machine "failed" |
| Micro-stops (< a few min) | Almost never | Performance | Too short to log by hand |
| Slow running / reduced speed | Almost never | Performance | Line looks "up" the whole time |
As a rule of thumb, micro-stops and speed losses tend to be the largest and least-recorded bucket — short enough that operators reset and move on without logging them — which is why a plant can believe it has "a few breakdowns a month" and still run well below world-class OEE. (This is an indicative pattern from practitioner experience, not a single audited statistic — the whole point is that these losses are the ones nobody counts.) You cannot cost what you never counted.
Surfacing the hidden hours with camera-based monitoring
This is where a camera already looking at the floor earns its keep. A person can't stand at every cell with a stopwatch, but video can be read continuously — an unmanned station, a cell idle while an operator hunts for a trolley, a changeover that ran 40 minutes instead of 15, a machine cycling below rate. Those are exactly the Availability and Performance losses that manual logs drop, and each one carries the ₹/hour tag from the formula above.
This is the wedge Mama is built around: you record a short phone walkthrough of the floor, and it returns a camera placement plan (how many, where, ceiling vs wall) that covers the work zones where downtime actually accrues — then reads those feeds into a plain-language efficiency summary, so hidden idle time becomes a number you can act on, not a blind spot. You get the measurement method and the coverage to feed it.
Do this on Monday
- Pick your worst cell or line. Get its ideal output/hour and contribution margin/unit.
- Compute its cost per downtime hour with the formula above.
- For two weeks, capture all stops — including micro-stops and slow running, not just breakdowns.
- Multiply. The annualised figure is your business case for fixing it.
The measurement is cheap. Not measuring is the expensive part.
FAQ
How do you calculate the cost of downtime in a factory? Per hour: lost good units × contribution margin per unit, plus idle labour cost, plus overhead that keeps running. Use contribution margin (price minus variable cost), not full price, so you don't overstate the loss. Multiply the hourly figure by your true downtime hours to annualise.
What's the difference between downtime cost and OEE? OEE expresses losses as a percentage (Availability × Performance × Quality); downtime cost expresses the same losses in rupees. OEE tells you how much capability you're losing; the ₹/hour formula tells you what it's worth. Use both — the gap between your OEE and ~85% is your money on the table.
What is a good OEE score for an Indian factory? The world-class benchmark is around 85%, and a 60–75% band is commonly cited as indicative for plants that have started measuring (treat it as a rule of thumb, not a hard number). For a mid-size Indian factory just starting to track it, getting reliable measurement in place matters more than the exact number — you can't improve what you don't yet count.
Why is so much downtime "hidden"? Manual logs capture loud breakdowns but miss micro-stops, slow running, changeover overruns and material waits — often the biggest bucket. No machine visibly "fails," so nothing gets recorded, and the loss never reaches a report even though it's costing money every shift.
How do cameras help measure downtime? Video can be read continuously across every cell, catching idle stations, slow cycles and over-long changeovers that no one has time to log by hand. Those are Availability and Performance losses; tagging each with its ₹/hour cost turns a blind spot into a quantified, fixable line item.