A machine's nameplate speed does not equal the output it actually delivers. Overall equipment effectiveness (OEE) separates three loss categories:

$$OEE=A\times P\times Q,$$

where

• $A$ = availability: fraction of planned production time in which the equipment is running;
• $P$ = performance: actual running rate relative to the reference rate;
• $Q$ = quality: fraction of produced units that count as good output.

Suppose

$$A=0.90,\qquad P=0.85,\qquad Q=0.98.$$

Then

$$OEE=0.90\times0.85\times0.98\approx0.750,$$

or about 75%.

The decomposition matters. Two machines can have the same OEE for completely different reasons: one may suffer frequent downtime, another may run continuously but slowly.

OEE is therefore useful as a diagnostic framework when its components are measured consistently. It should not become a target that encourages hiding planned downtime, slowing reference rates or redefining scrap.

The engineering purpose is to identify where capacity is being lost:

planned time
  ↓ availability loss
running time
  ↓ speed loss
actual output
  ↓ quality loss
good output

Improvement should attack the underlying loss mechanism rather than the product of the three percentages alone.