Taguchi

Genichi Taguchi, a Japanese engineer, published his first book on experimental design in 1958. His methods of fractional factorial design have certainly earned him the most well known "brand" in the field of Design of Experiments. However, historically, it has been primarily used in manufacturing and product design and development rather than in website optimization. The aim of the Taguchi design was to make a product or process more stable in the face of variations over which we have little or no control (for example, making sure that a car engine can perform reliably in the face of different ambient temperatures).

Taguchi arrays are "pre-planned" fractional factorial designs. Usually, these designs allow a single effect to be estimated but confound interactions among two or more factors. This limitation reduces the scope of any experiment considerably.

Most statisticians hold Taguchi's contribution to the philosophy of experimental design in high regard. However, given modern computers and methods of multivariate testing, the pre-planned designs themselves are being replaced with more efficient and less restrictive computer-generated experiments for website testing. This is especially true when certain combinations included in a Taguchi array cannot be run because they are known, for example, to produce very poor results, or cannot be offered to customers due to company policy or logical inconsistency (i.e. "constraints").

For the marketer who is trying to achieve website optimization, this restriction becomes very important. For these reasons, Optimost does not typically rely on Taguchi arrays to conduct its experiments. Rather it normally employs the Optimal Design method of multivariate testing, which builds upon Taguchi's theories and other experimental design techniques, and combines these with modern computing power and a unique content generation engine to test these relationships among many variables on a site.