If every batch coming off a production line could flawlessly reproduce the original design, there would be little need for statistical input in quality control. Unfortunately, in the real world of manufacturing, many factors combine to make each unit unique, so that some form of statistical control is necessary for any manufacturing process. Building control chart is the most efficient way to detect non random causes of variation that are easy to remove. Then the process is in a state of statistical control, capabilities studies are helpful to define if the process is meeting the specifications. When a process is stable over time and is found highly capable, the opportunity of applying optimization methods to really improve the products characteristics must not be lost. Multifactorial analyses are used to highlight the fundamental structures of the data above all when numerous variables are collected for a great number of items. Obviously, excellency can not be achieved just by using only one of these methods or by running a single shot approach, quality improvement is a long and exciting way by which the knowledge about the process is built up step by step, from day to day.
