Optimizing Website Content via the Taguchi Method

Classical experimentation – the practice of holding the other variables constant, while analyzing the effect of one variable under consideration on the outcome – was widely accepted in the sciences, until modern experimentation was put forth in R.A. Fisher’s The Design of Experiments (1935). Design of experiments (DOE) is a methodology that consists of simultaneously manipulating levels or amounts of selected independent variables (i.e., factors or parameters) to examine their influences on the dependent variable(s) (i.e., effects, outcomes, or target variables). Specifically, DOE (aka factorial design) is defining and analyzing all possible conditions in an experiment to optimize the outcome.

Taguchi’s contribution in the late 1940s to the science of DOE consists of the following: Taguchi simplified and standardized the (fractional or partial) factorial designs in such a manner that two statisticians conducting the same experiment apart from each other will always use a similar Taguchi design, and tend to obtain similar results. Whereas with DOE, the factorial and fractional DOE would suffer from: 1) Experiments with a large number of variables become unwieldy in cost and time. 2)Two designs from the same experiment may yield different results and interpretation. Effectively, the Taguchi method produces a standardized design methodology that can easily be applied; designs for the same experiment by two different statisticians will yield similar data and will lead to similar conclusions.

I present a clear understanding of what Dr. Taguchi has done to make the DOE technique more effective in application. The Taguchi Method ™/DOE (yes, the term is trademarked) is a recent experimental strategy in which Taguchi uses a modified and standardized form of DOE. In other words, the Taguchi method is a form a DOE with special application principles. Moreover, I introduce the Taguchi/DOE/GenIQ© Method as an advanced approach to Taguchi Method/DOE for A/B and multivariable testing that looks to optimize the process from the early stages of the test.

The purpose of this article is to review the Taguchi Method in a case study as the approach of choice for optimizing website content. The site factors that Taguchi/DOE can test and optimize are such as: Site content; email marketing; internal promotions and special offers; mutli-step processes (e.g., checkout, registration; site navigation, form for explicit customer input); and of course, the outcome (a click, a request for further info; request to be contacted by salesperson, or a purchase).