A site improves output per labor hour by producing a larger batch that customers do not need. Another improves OEE by changing what it counts as planned production time.
Both measures can rise while the business becomes worse. Useful performance measurement requires a stable definition, relevant output and a management response aligned with customer value.
Define productivity precisely
Productivity compares output with an input. Labor productivity relates output to labor used; it does not describe every factor contributing to the result. The US Bureau of Labor Statistics defines the measure in those terms.
| Hypothetical team | Before | After |
|---|---|---|
| Comparable good units | 100 | 110 |
| Labor hours | 125 | 125 |
| Units per hour | 0.80 | 0.88 |
The ratio improves by 10%. Check comparability. Different product mix, outsourced work or deferred maintenance can change the ratio without the same underlying improvement.
Separate partial from broader measures
A labor measure can improve through additional equipment or purchased services. The result may still be valuable, but the other resources should remain visible.
Use a broader cost or input view where appropriate, with consistent valuation and timing. Avoid adding unlike physical inputs directly.
A measure should answer a defined question. It is not necessary for one ratio to explain the entire business.
Use OEE to expose equipment losses
Overall equipment effectiveness is commonly calculated as availability multiplied by performance multiplied by quality.
| Factor | Value |
|---|---|
| Availability | 90% |
| Performance | 80% |
| Quality | 98% |
| OEE (0.90 × 0.80 × 0.98) | 70.56% |
The factors require consistent definitions of planned production time, run time, ideal cycle time, total count and good count. Vorne’s calculation guide explains the relationships.
Choose the boundary deliberately and preserve it when comparing periods. A changed denominator can create an apparent improvement.
Interpret the losses before setting a target
Availability, speed and quality losses need different interventions. A single total conceals which mechanism deserves attention.
Validate the ideal cycle time and the treatment of changeovers and stops. An impossible performance result may indicate bad inputs or an unsuitable baseline.
Avoid a universal “world-class” threshold. Equipment roles, product mix and measurement boundaries differ. The useful target reflects the business requirement and a credible improvement path.
Connect local performance with the system
Higher OEE on a non-bottleneck machine may create more inventory without increasing delivered output. At a constraint, recovered capacity can be valuable if downstream processes and demand can use it.
Separate technical capacity released from financial benefit realized. A saved hour becomes an economic benefit through a specific action, not automatically through multiplication by an allocated hourly rate.
Design the review around decisions
Review data quality, the main loss, its customer consequence and the action required. Track whether improvement lasts and whether another loss grows.
Choose one important ratio and document its numerator, denominator, scope and exceptions. Ask whether someone could improve the figure while making the business worse.
The strongest measurement system exposes useful operating choices and makes manipulation harder. Its purpose is better performance, not a more impressive dashboard.