A process that does not change with changing noise is described as being robust. For instance, you may have a process making a product (for example, plastic door handles), but despite your best efforts the product quality varies wildly. You would typically concentrate on those variables you are able to control (raw material quality, machine settings such as temperature and time in mould) to try to make a better product. However, the product quality varies due to the variables you are not able to control, such as ambient temperature, humidity, and differences due to changes in operators from one shift to another. These variables are known as noise.

Rather than tightening up on the process variables, it is often better to try to adjust the level of these variables to reduce the effect of noise. This way you end up with a product or process that is not only high in quality, but gives you consistently high quality.

For example, in the case of our plastic door handles, there may be a combination of the formulation, machine setting, and speed of moulding that gives a process that gives consistently good product regardless of the ambient temperature and humidity. To do this you need to identify which factors have an effect on the variability of the process. This is done by looking at signal/noise in a robust design.

This is an important part of the experimental design and Taguchi, and is often investigated using a parameter design.