These are deviations from the specified performance target causing "loss" (or scrap). Variability can be attributed to both controllable and uncontrollable sources. Identification and exact quantification of these sources are essential if such deviations are to be minimised, thereby increasing "profit" margins.

We are all too familiar with "fire fighting" attempts to reduce variability. These range from individual machine operator's views, to experimenting with the process in order to enhance qualitatively and quantitatively understanding of it. Experimentation, however, requires "time and resources" , not so readily available. Effective experiments are therefore those that posses "strategy" and "structure" in the way that they are designed, performed and analysed. True process knowledge is gained through efficient experimentation. The manner in which experiment are designed, performed and analysed determine the degree of success of the effort.