Quality Function Deployment


(developed by Dr A.J. Lowe in collaboration with Prof. K. Ridgway of the University of Sheffield, England)                     


Introduction to QFD

                         Yoji Akao is widely regarded as the father of QFD and his work led to its first implementation at the Mitsubishi Heavy Industries Kobe Shipyard in 1972. The interest in QFD in the West was stimulated by reports of the achievements made by Toyota through its application between 1977 and 1984. These included a reduction in product development costs by 61%, a decrease in the development cycle by one third and the virtual elimination of rust related warranty problems (Ref 1).  Yoji Akao defined QFD as "a method for developing a design quality aimed at satisfying the consumer and then translating the consumer's demands into design targets and major quality assurance points to be used throughout the production phase". (Ref 2)  The main features of QFD are its focus on meeting customer needs through the use of their actual statements (termed the "Voice of the Customer"), its facilitation of multidisciplinary team work and the use of a comprehensive matrix for documenting information, perceptions and decisions. This matrix is commonly referred to as the "House of Quality" and is often perceived to represent QFD in its entirety.  In addition to the "House of Quality" matrix, QFD utilises "Seven Management and Planning Tools"which are used in many of its procedures:                                 


1    Affinity diagrams.

2    Relations diagrams.

3    Hierarchy trees.

4    Matrices and tables.

5    Process Decision Program Diagrams (PDPC)

6    The Analytic Hierarchy Process (AHP)

7    Blueprinting


Affinity diagrams

This is a powerful method used by a team to organise and gain insight into a set of qualitative information, such as voiced customer requirements. Building an Affinity Diagram involves the recording of each statement onto separate cards which are then sorted into groups with a perceived association. A title card which summarises the data within each group is selected from its members or is created where necessary. A hierarchy of association can be achieved by then sorting these title cards into higher level groups.

Hierarchy trees

A Hierarchy tree or Tree Diagram also illustrates the structure of interrelationships between groups of statements, but is built from the top down in an analytical manner. It is usually applied to an existing set of structured information such as that produced by building an Affinity Diagram and is used to account for flaws or incompleteness in the source data. Working down from the top a team can amendments at each level and the completed hierarchy can be drawn as shown below.

Matrices and tables

The matrix is a tool which lies at the heart of many QFD methods. By comparing two lists of items using a rectangular grid of cells, it can be used to document a team's perceptions of the interrelationships that exist, in a manner which can be later interpreted by considering the entries in particular cells, rows or columns. In a prioritization matrix the relative importance of items in a list and the strength of interrelationships are given numerical weightings (shown as numbers or symbols). The overall priority of the items of one list according to their relationships with another list, can then be calculated as shown below.

Tables are also used in QFD to study the implications of gathered or generated items against a specified list of categories. Examples include production planning and analysing customer statements in the Voice of the Customer Table (VOCT) shown below.

Relations diagrams

Relations diagrams or Interrelationship Di-graphs can be used to discover priorities, root causes of problems and unstated customer requirements.

Process Decision Program Diagrams (PDPC)

PDPC are used to study potential failures of new processes and services.

The Analytic Hierarchy Process (AHP)

AHP uses pairwise comparisons on hierarchically organised elements to produce an accurate set of priorities.


Blueprinting is a tool used to illustrate and analyse all the processes involved in providing a service.

The House of Quality

The "House of Quality" matrix is the most recognised form of QFD (Ref 3). It is utilised by a multidisciplinary team to translate a set of customer requirements, drawing upon market research and benchmarking data, into an appropriate number of prioritised engineering targets to be met by a new product design. There are many slightly different forms of this matrix and this ability to be adapted to the requirements of a particular problem or group of users forms one of its major strengths. The general format of the "House of Quality" is made up of six major components which are completed in the course of a QFD project:

  1. Customer requirements (HOWs) - a structured list of requirements derived from customer statements.
  2. Technical requirements (WHATs) - a structured set of relevant and measurable product characteristics.
  3. Planning matrix - illustrates customer perceptions observed in market surveys. Includes relative importance of customer requirements, company and competitor performance in meeting these requirements.
  4. Interrelationship matrix - illustrates the QFD team's perceptions of interrelationships between technical and customer requirements. An appropriate scale is applied, illustrated using symbols or figures. Filling this portion of the matrix involves discussions and consensus building within the team and can be time consuming. Concentrating on key relationships and minimising the numbers of requirements are useful techniques to reduce the demands on resources.
  5. Technical correlation (Roof) matrix - used to identify where technical requirements support or impede each other in the product design. Can highlight innovation opportunities.
  6. Technical priorities, benchmarks and targets - used to record the priorities assigned to technical requirements by the matrix, measures of technical performance achieved by competitive products and the degree of difficulty involved in developing each requirement. The final output of the matrix is a set of target values for each technical requirement to be met by the new design, which are linked back to the demands of the customer.

Models for Applying QFD Tools

The "House of Quality" can be used as a stand alone tool to generate answers to a particular development problem. Alternatively it can be applied within a more complex system in which a series of tools are used. The "Clausing Four-Phase Model" is the most widely known and utilised of these approaches (Ref 3). It translates customer requirements through several stages into production equipment settings; using three coupled QFD matrices and a table for planning production requirements (as shown below).

The less known and more comprehensive "Matrix of Matrices" model provides developers with thirty matrix tools and tables (Ref 4) which consider development steps not included in the "Four-Phase" approach (see below). This represents the full QFD tool kit and practitioners should select and adapt from this set as appropriate rather than attempt to implement it in its entirety.


  1. Such is the flexibility of the matrix tools utilised by QFD, its methods have now been applied in many fields other than product development. One of the main areas is in strategy formulation and implementation.

    As with other Japanese management techniques some problems have been encountered when applying QFD within the western business environment. These are mentioned on the practitioner's tips page and are also addressed from a US perspective by Bob Hales.

The Implications of Arrow's Impossibility Theorem on QFD

The Impossibility Theorem

Hazelrigg (Ref 12) has argued that design approaches such as QFD, that seek to optimize the value of a design to its customers can lead to highly erroneous results. He bases this argument on the impossibility theorem first presented by Kenneth Arrow (Ref 13).

Kenneth Arrow considered the problem of constructing a utility function to express the preferences of a group and showed that apart from in some very special cases, utilities cannot be used. To demonstrate his theorem the concepts of optimisation and utility must be understood.


Optimisation is performed only with reference to some specific measure of value. In the example below the optimisation of "y" can be achieved by the appropriate selection of "x".

y = f (x) (Value function for "y")

If the value function "f (x)" cannot be expressed, the optimisation cannot be accomplished. The purpose of optimisation methodology is to find the set of values of "x" that obtain the maximum value of "y", even when it is necessary to search across an infinity of possible values of "x".


Utility is an economic value of preference and has the unit of utiles. For example an individual's preference for three alternatives A, B and C; where A is preferred to B and B is preferred to C, can be expressed as :

A > B > C

In this case each option can be assigned a number of utiles, as a measure of the utility of each preference. The above could then be expressed as a utility function as shown :

uA >uB > uC

The impossibility theorem considers the preferences of a group of three rational individuals which are shown in the table below :



A vs. B

B vs. C

A vs. C


 A>B>C, A>C





 B>C>A, B>A





 C>A>B, C>B




Group preferences





While each individual has a rational set of preferences it is obvious that combining these to form a group utility function presents a problem. Optimisation for the group using this data is impossible. Hazelrigg (Ref 12) argues that this situation is not a rare case but is in fact the norm and as the preferences of individuals within a group are defined in greater detail, the higher the chance of encountering this type of problem.

The Impossibility Theorem

Hazelrigg (Ref 12) illustrates the potential consequences of the impossibility theorem with a simple example that shows how combined preference data can lead to the design of a product that satisfies none of the customers :

A simple product possesses three attributes: colour, size and shape. Each of these can be one of two options; colour can be Red or Green, size can be Large or Small, shape can be Bumpy or Flat. There are three rational customers whose preferences for each of the attribute options are described in the table below. In this case values of "Attribute Utility" are used (where 1.0 is the maximum utility and 0.0 represents the lowest utility where the customer will not want the product at any price), these are multiplied together in order to measure the overall utility a customer has for a particular product configuration. (e.g. customer I has a utility of 1.0 × 0.9 × 1.0 = 0.9 for a Red - Small - Flat product).

An optimisation approach based upon such preference findings would result in a design that was Red - Large - Bumpy; as on average for the group Red is preferred to Green (customers I and II preferring Red, while only customer III prefers Green), Large is preferred to Small and Bumpy is preferred to Flat. But this combination has a utility of 0.0 for all three customers (e.g. for customer I the overall utility for Red - Large - Bumpy = 1.0 × 1.0 × 0.0 = 0.0) and so is the worst possible design. Based on these examples Hazelrigg (Ref 12) argues that the use of averaged group preference data is inappropriate in the optimisation of product design.

Consequences for QFD Practitioners

The above example is not entirely representative of the manner in which customer requirement data is defined and combined in QFD. Despite this the issues that the theorem raises still need to be considered, as any form of averaging of individual responses will not necessarily optimally represent group requirements.

In the example a strategy for the manufacturer's success is to supply a range of products in each colour, size and shape combination to satisfy the three customer types simultaneously. This case though is simplistic and in genuine engineering design situations the number of attributes, customers and sets of preferences are very much greater. The occurences of such irrationalities in complex grouped preference data would be considerably less obvious.

Considering the many documented examples of QFD application, this problem does not appear to be occuring on the magnitude that would be predicted by Hazelrigg's (Ref 12) arguement. It could therefore be concluded that in real, complex products these problems only manifest themselves to a small extent, lowering the optimality of the overall design, but not rendering it entirely inappropriate as may be anticipated. Alternatively the focus of QFD designers upon specific market segments may minimise the occurence of incompatible preference groups. Accurate segmentation of the market through the use of tools such as the Voice of the Customer Table (VOCT) could be used to minimise the consequences of these problems.

The Voice of the Customer Table (VOCT) is a component of the "Comprehensive QFD" (Ref 6) system (itself a subset of the Matrix of Matrices) and its use is a valuable preliminary exercise before building a "House of Quality" matrix. It has two parts.

The purpose of the VOCT Part 1 is to :

  1. Identify customer usage of the product or service.
  2. Predict possible usage of the product or service.
  3. Assist in market studies through usage analysis.

The VOCT Part 1 is completed for each customer statement, for which a customer I.D., a customer demographic (sex, age, location etc...) and product use information are recorded. The product use questions are categorised into Who, What, When, Where, Why and How (e.g. : Who uses it? Who will use it?; What is it used for? What could be the use?...) which when analysed in conjunction with the demographic information serve to highlight different market segments into which customers may be divided. The two columns under each "Use" category differentiate entries that are gathered directly from customers from those that are generated internally within the company.

The VOCT Part 2 identifies demands that are spoken and unspoken based upon the Voice of the Customer and usage information. These are transformed into useful reworded statements for use in the QFD process (e.g. a customer requirement entry in a "House of Quality" matrix). For each statement the table requires :

  1. Demanded Quality (Customer Requirements) items are developed in terms of the organisation.
  2. Measurable Quality Characteristics (Technical Requirements) that ensure Demanded Quality items will be met, are identified.
  3. The Functions that are necessary for the product to be acceptable to the customer are identified.
  4. The Reliability of the product in satisfying its intended Function for a specified time period is considered.
  5. Comments can be added to raise other important factors

The outputs of the VOCT can then be entered into a "
House of Quality" matrix to begin the product planning process.
The application of the VOCT for identifying segments in a customer group is particularly important in avoiding the problems highlighted by the Impossibility Theorem.


1.   Sullivan, L.P., 1986, "Quality Function Deployment", Quality Progress, June, pp 39-50.

  1. Akao, Y., Ed. 1990, "Quality Function Deployment: Integrating Customer Requirements into Product Design", Translated by Glenn Mazur. Cambridge, MA: Productivity Press.
  2. Hauser, J.R. & Clausing, D., 1988, "The House of Quality", Harvard Business Review, May-June, pp 63-73.
  3. King, B., 1989, "Better Designs in Half the Time", Third Edition, GOAL/QPC, Methuen, Massachusetts.
  4. Cohen, L., 1995, "Quality Function Deployment: how to make QFD work for you", Addison-Wesley Publishing Company, Massachusetts.
  5. Nakui, S., 1991, "Comprehensive QFD System", Transactions from the Third International Symposium on QFD, Novi, Michigan, pp 137-152.
  6. Hales, R.F., 1995, "Adapting Quality Function Deployment to the U.S. culture", IIE Solutions, Oct. (27/10), pp 15.
  7. Scanlan, J.P., Winfield, A. & Smith G., 1994, "Modelling the Design Process within the Aerospace Industry", Factory 200 - Advaned Factory Automation, 3-5 October, Conference Publication No. 398, IEE, pp 645-650.
  8. Smith, J.A. & Angeli, I.I., 1995, "The use of Quality Function Deployment to Help Adopt a Total Quality Strategy", Total Quality Management, Vol. 6, No. 1, March, pp 35-44.
  9. Poolton, J. & Barclay, I., 1996, "Concurrent Engineering Assessment: a Proposed framework", Journal of Engineering Manufacture, Proceedings of the Institution of Mechanical Engineers, Vol. 210, No. B4, pp 321-328.
  10. Veness, P.J., Chidolue, G. & Medhat, S.S., 1996, "Concurrent engineering infrastructure; tools, technologies and methods in British industry", Engineering Management Journal, Vol. 6, No. 3, pp 141-147.
  11. Hazelrigg, G.A., 1996, "The implications of Arrow's impossibility theorem on approaches to optimal engineering design", Journal of Mechanical Design, Vol. 118, June, pp 161-164.
  12. Arrow, K.J., 1963, "Social choice and individual values", 2nd Ed., John Wiley & Sons, New York.
  13. Prahalad, C.K. & Hamel, G., 1990, "The core competence of the corporation", Harvard Business Review, Vol. 68, No. 3, May-June, pp 73-91.
  14. Gallon, M.R., Stillman, H.M. & Coates, D., 1995, "Putting core competency thinking into practice", Research and Technology Management, Vol. 38, Part 3, May-June, pp 20-28.
  15. Kirkwood, D.H., 1994, "Semisolid metal processing", International Materials Reviews, 39(5), pp 173-189.
  16. Zairi, M. & Youssef, M.A., 1995, "Quality function deployment - a main pillar for successful total quality management and product development", International Journal of Quality & Reliability Management, Vol. 12, No. 6, pp 9-23.