Using visual and auditory cues an experienced Manufacturing Manager can often sense when a factory is running well. Silence means no production. No workers present, no production. It’s an informal, and highly subjective, form of machine monitoring with “data” obtained through “Managing By Walking About” (MBWA.)

The practice has much to recommend it, but it’s not sufficient for maximizing output and minimizing costs: that takes real measurement. Most factories use some combination of efficiency, utilization and productivity for this. In recent years many have added OEE to that mix. But what do these metrics measure? Will they provide accurate feedback? Are these metrics the most appropriate measurements?


MRP systems plan production, efficiency and log output. Manufacturing Execution Systems (MES) organize and track work as it moves through the factory. Both yield data on what was produced and when but are of limited help in controlling costs and maximizing output from a fixed set of resources.

This is where efficiency, utilization and productivity measures come in.

Efficiency is a ratio of input to output. An efficient process converts most of the input to output, one that’s inefficient wastes a higher proportion. Some processes are inherently inefficient though. Machining an aerospace bracket involves cutting away most of the metal, in effect, wasting it. Is that efficient? Perhaps 3D printing would be more efficient, even though it might take longer and require more processing steps.

In short, efficiency as a gauge can be ambiguous and even misleading.

How about utilization? This is a measure of resource activation. A press that’s punching out metal parts is being utilized. One that’s waiting for repair is not. But is low utilization necessarily negative? A press that’s part of a manufacturing cell might only operate once a minute, because that’s what the Takt time dictates. Press utilization is low even though the cell is running efficiently.

Productivity is determined by how much is produced per unit of input. Most often it’s expressed in terms of labor productivity: how many man hours required to assemble a car, for example. The rationale is that fewer hours per vehicle means lower costs.

That might have been true before widespread automation, but the measure has less value today. In highly automated factories very few people are directly involved in making things and direct labor’s contribution to total costs is small.


Over the last 25 years or so many manufacturers, in search of a better way of measuring factory performance, have adopted Overall Equipment Effectiveness (OEE) OEE is a percentage, arrived at by multiplying availability with performance rate and quality rate. To understand the value of OEE it’s necessary to drill down into those terms.

  • Availability – this is the percentage of total time that a machine could be running. (There are 168 hours in a week, so use that as the denominator.) It involves set-up hours and time when it’s not available due to breakdown. Note: how this differs from utilization, which only measures whether the machine is running.
  • Performance rate – this is the true output being achieved, as a percentage of what the machine or process is capable of. It considers reduced speed running to increase tool life or reduce scrap as well as the micro-stoppages that plague some operations.
  • Quality rate – this is the percentage of saleable product coming out of the process compared to total output. It accounts for scrap due to process and material variation as well as start-up and shut-down losses.


OEE highlights opportunities for improvement. It helps managers identify what to focus on and what to overlook. If a machine or line takes a long time to set up that may be reflected in all three contributing factors. If the process isn’t capable of holding the required tolerances, that will show up in the quality rate. Likewise, micro-stoppages reduce the performance rate and poor reliability will appear in availability.

OEE has two significant deficiencies. It doesn’t measure productivity, and, in common with every other measure, it requires accurate data.

Addressing productivity first, the availability component can be boosted by adding labor. As with a race car pit stop, more people could reduce set-up time. In the same vein, it may be possible to reduce the impact of micro-stoppages by stationing workers at the trouble points. The quality rate could perhaps be addressed by inspecting incoming components or material before they enter the process. Such actions could improve OEE but at the expense of productivity. The bottom line is: OEE alone is not sufficient for managing factory operations.

Inaccurate data is a problem with all performance measures. In factories without automated machine monitoring data comes from manual entry, combined perhaps with some MES tracking of when jobs were started and finished.

In a busy factory, supervisors can easily forget to log every stoppage and incident, especially those with minor impact. (Though if they recur frequently they constitute a chronic loss.) Manual data entry is also subjective and at risk of conscious or unconscious bias. And as is well-understood, unreliable data translates to unreliable performance measures.


Accurate manufacturing data about what each machine in a factory is doing through the shift and day requires a machine monitoring solution. Historically, this was expensive and difficult. Today, low-cost sensors and an array of communication technologies, (MTConnect, OPC UA, Ethernet,) make it inexpensive and relatively straightforward.

Two terms associated with machine monitoring are the Industrial Internet of Things, (IIoT) and Industry 4.0. While not identical, they overlap and both relate to gathering highly granular data about factory and machine activity at any moment in time. As this discussion of factory performance methods has shown, by harvesting accurate data, manufacturers that implement machine monitoring technology will improve their ability to measure, and hence improve, factory operations.