Output Per Machine Hour

Output per machine hour is the number of good units a machine produces per hour of available machine time. It is a machine-centric productivity metric, good units divided by machine hours, that tells a capital-heavy plant how hard each expensive asset is actually working.

In a plant where the machines cost more than the payroll, the scarce resource is machine time, not labor time. Output per machine hour puts the number where the money is: it measures productivity against the asset, so a slow, well-staffed line and a fast, lean one get compared on the thing that actually limits them. This guide covers the formula, how it differs from units per labor hour, when each one is the right lens, and how it rolls into OEE.

What is output per machine hour?

It is a rate: good units out, per hour the machine was available to run. If a press makes 4,800 good parts across a 10-hour shift, its output per machine hour is 480. The word "good" matters, scrap and rework do not count, because a machine that runs fast and produces defects is not productive, it is just busy.

The metric answers a specific question that capital-heavy operations live and die by: for every hour we own this asset, how much sellable product does it make? That framing is different from asking how busy the operators are or how many units left the building. It isolates the machine as the unit of account, which is exactly what you want when the machine is the expensive, constraining thing.

How do you calculate output per machine hour?

Divide good units produced by the machine hours that produced them. The formula is simple; the discipline is in the two inputs.

The first input, good units must exclude scrap and rework so the metric rewards sellable output, not motion. The second, machine hours needs a defined basis: available hours (the machine was staffed and could run), scheduled hours, or calendar hours. Pick one and label it every time. Output per available machine hour and output per calendar hour are both valid, but they answer different questions and must never be mixed in the same chart.

The most common mistake is a moving denominator. A supervisor quotes output per hour off run time one week and off scheduled time the next, and the number swings for reasons that have nothing to do with the machine. A close second is counting units at the wrong point, tallying everything that came off the tool instead of only what passed inspection, which quietly credits the asset for scrap. Lock the basis and the count point before you trend the metric, or the trend will lie. A machine-centric number is only as honest as the two inputs feeding it, and both are easy to get wrong by hand.

How is output per machine hour different from units per labor hour?

They measure the same output against different denominators, and the denominator is the whole point. Units per labor hour divides good units by labor hours, it rises when a crew does more with fewer people. Output per machine hour divides the same good units by machine hours, it rises when an asset makes more per hour it is available. In a capital-heavy plant the two can move in opposite directions, and knowing which one to trust is the difference between a good decision and an expensive one.

A crew that runs three automated cells sees units per labor hour soar while output per machine hour on each cell may be mediocre, the labor is efficient, the assets are underworked. The reverse happens on a hand-pack line, where output per machine hour is meaningless and labor productivity is everything. The two metrics are complements, not rivals: track both, and let the plant's cost structure decide which one drives investment. Where labor is the constraint, the workforce equivalent of OEE, overall labor effectiveness is the deeper lens.

How do you improve output per machine hour?

Every gain comes from one of three places: more available hours, faster good-unit production, or fewer defects. Work them in order of cost.

1. Recover available hours. Cut unplanned downtime and shorten changeovers so the machine is available more of the shift. This is usually the cheapest output gain because it needs no capital.
2. Close the speed gap. Compare the actual rate to the machine's design rate. Chronic minor stops and slow-running bleed output per hour without ever showing as a full stop.
3. Stop making scrap. Every defective unit is machine time spent producing nothing sellable. Raising first-pass yield lifts good output per hour with zero extra runtime.
4. Match the product to the asset. Running a job the machine is poorly suited for drags its hourly output. Routing work to the best-fit asset raises the fleet average.
5. Verify the ideal rate is honest. If your target rate is set too slow, the machine looks fully productive while real capacity sits idle. Base it on the fastest rate sustained on good product.
6. Re-measure and re-rank. Once one lever is pulled, the binding constraint moves. Re-check which of the three, hours, speed, or quality, now caps the number.

When does output per machine hour matter most?

When the machine is the expensive, scarce resource, capital-heavy operations. In plastics, metal forming, food and beverage processing, and any plant built around presses, ovens, molders, or reactors, the asset dwarfs the payroll and machine time is the thing you cannot buy more of cheaply. There, an hour of lost output on a key asset is the most expensive thing that happens all day, and output per machine hour is the metric that surfaces it.

It matters less on labor-paced lines, where output tracks headcount and the machine is incidental. The tell is simple: if adding a person raises output more than speeding the machine would, you are labor-constrained and should lead with labor productivity; if the machine caps you regardless of crew size, output per machine hour is your north star. Most real plants are a mix, which is why the honest answer is to measure both and let each department use the one that reflects its constraint. Choosing well is part of choosing the right plant KPIs.

How does output per machine hour compare across the fleet?

The macro numbers say most assets have room to run harder. The U.S. Federal Reserve's G.17 industrial production release put manufacturing capacity utilization at about 75.8% in April 2026 roughly 2.4 percentage points below its 1972–2025 average, a plain signal that factories run well below their theoretical asset ceiling. Capacity utilization is broader than any single machine's output per hour, but it frames the gap: idle and slow-running assets are the norm, not the exception.

Labor productivity tells the complementary story. The U.S. Bureau of Labor Statistics reported that manufacturing-sector labor productivity rose 1.9% over 2025 output per hour worked climbing as plants squeezed more from each labor hour. Read together, the two series make the case for tracking both denominators: the asset side has slack to close, and the labor side is where recent gains have come from. Neither replaces measuring your own machines, but both show why one metric alone never tells the whole productivity story.

These figures are hypothetical but they show the metric's value: three presses on identical 10-hour windows produce very different output per machine hour. Line C's gap is not a speed problem, it is lost availability to a long changeover, which output per machine hour surfaces immediately and a plant-wide units total would hide.

How does output per machine hour connect to OEE?

Output per machine hour is the plain-language version of what OEE decomposes. OEE multiplies availability, performance, and quality into a single percentage; output per machine hour bakes the same three effects into one intuitive number, good units per hour. Lose availability and the hours you can produce in shrink; lose performance and the rate per hour drops; lose quality and good units fall. The metric moves for exactly the reasons OEE does, which is why they belong on the same dashboard.

The advantage of output per machine hour is that operators feel it without a lesson in ratios, and the advantage of the OEE calculation is that it tells you which of the three losses caused the drop. Use output per machine hour to spot the asset that slipped and OEE to explain why. That handoff only works when good units and machine hours are captured from the equipment itself rather than tallied by hand, which is what makes the number trustworthy shift after shift. Harmony logs run time and counts from machine signals with operator-confirmed reason codes (see the platform or the CLS results). From there, cross-check against your plant KPIs compare with cycle time on the constraint, and model scenarios in the OEE calculator.