Pin Me

Overview of the Taguchi Methodology

written by: Bruce Tyson • edited by: Ginny Edwards • updated: 9/26/2010

Genichi Taguchi pioneered methods that would statistically improve manufacturing processes with the end result being an improvement in product quality. Since its inception, Taguchi methodology has been adopted for use in engineering, biotech, advertising, and a variety of other business roles.

  • slide 1 of 4


    Orthogonal array Taguchi methodology addresses problems based on two broad categories: static and dynamic. A static problem is one where outputs are affected by controlling factors only, while a dynamic problem has a so-called "signal" input that will directly affect the outcome of the process.

    While evaluating problems, orthogonal arrays are used by Taguchi to format data in rows and columns to help with experimental design. The columns include factors that affect performance while the rows contain the states of each factor. The factors examined by Taguchi are those that have an observable effect on the processes that is under consideration.

    Generally speaking, Taguchi methodology can be applied anywhere where outcomes can be accurately observed, making it useful in business planning and management.

    Image Credit: Wikimedia Commons/Jayen466

  • slide 2 of 4


    Taguchi methodology can be applied to quality management and project management at numerous levels. Such as evaluating competitive strategy decisions, reducing errors in order processing, and speeding products to customers. With Taguchi, quality is measured differently depending on context and can briefly be summarized as follows:

    • A deviation of a finished product from its intended outcome: The more the result of a process deviates from plans, a loss occurs, which Taguchi seeks to minimize. The deviation could result in something like safety hazards, or a diminished reputation of the business among its potential customers.
    • A pattern of improvement in quality and cost ratios that supports competitiveness. Typically, this improvement consists of small increments over a long period of time.
    • Consistency, where the outcomes deviate by ever decreasing margins.
    • Environmental "noise" is the collection of non-linear factors that influence the outcome of a process. By dealing with non-linear influences, outcome quality can be enhanced without increasing the cost of producing that outcome.
    • Experiments (trial and error) can help identify parameters that best insulate outcomes from the effects of non-linear factors (noise).
    • The financial cost of quality can be calculated using formulas developed by Taguchi.

  • slide 3 of 4

    A Systematic Approach to Quality

    Taguchi methodology presents an organized approach to the generation of high quality assurance plan outcomes. This methodology comes in an experimental process of four stages:

    1. Planning
    2. Conducting
    3. Analyzing
    4. Implementing

    Planning includes recognizing the existence of the problem and building a team tasked to deal with it. Criteria for measuring the deviation of outcomes from design must be selected. Without the ability to quantify the problem, it is difficult to know when it is successfully addressed. Planning also requires the identification of factors that impact quality and the definition of those factors as control, signal, or noise.

    Also, as planning progresses, the number of levels of factors in the process needs to be decided upon (typically, Taguchi methodology requires at least two levels, three if non-linear inputs are at play) and the methods used to study the ways those factors affect the process. Finally, the orthogonal arrays should be constructed. Some fairly complicated mathematical equations can be used here, although for project management purposes exact calculations may not be necessary.

    Conducting the experiment is the operation that seeks to determine what solutions may work. After the problem solving process is planned, it should be set into motion (conducted). A place to work is necessary, and adequate resources to influence the various factors in the process and record the results.

    Analyzing the data is the step in the process that determines whether an effective solution to the problem was discovered. A well designed experiment under the Taguchi method should have valid data that can indicate the parameters that affect the mean, optimum, and variable performance of the process and reveal whether improvement based on those factors is possible or feasible.

    When performed to its mathematical and statistical specifications, Taguchi methodology produces graphical representations such as Pareto plots that help with data analysis.

    Implementing changes determined effective during the experimental process needs to be put into practice. If the results of the progress are not brought into line with expectations, sources of error should be identified followed by another iteration of the four stage process.

  • slide 4 of 4

    Wrap up

    It was not until after 1980 that Taguchi methodology gained traction in the United States, first taking hold at corporations such as AT&T and Ford. Since then, it has also taken root in the United Kingdom in various manufacturing and service industries. This overview has summarized Taguchi's statistical approach to quality. Centering on the specialized definition of problems, contextual quality, and a four part multi-stage process for improvement, Taguchi methodology has contributed a lot to our understanding of problems and their resolutions.