QFD is a comprehensive quality assurance system as well as a disciplined approach for the implementation of total quality management (TQM). Quality assurance, represented by Ishikawa as a type of promise or contact with consumers regarding quality [
21], is the central theme of company-wide quality control (CWQC), the Japanese-style total quality control, and QFD is the “operational definition” of CWQC, described by Sullivan [
22]. QFD was developed at the time the major industries of Japan were in need of a system to assure quality throughout the flow from design to production to increase their competitiveness [
23]. Akao, the founder of QFD, explained the fundamental aim of QFD is to clarify and to solve all major issues of quality assurance in product development, and establishing control points prior to production start-up is the central idea of QFD for assuring product quality [
24]. As the deployment technique for assuring product quality is equally applicable to managing business process, QFD uses the same approach to facilitate the operation of
hoshin kanri, a participative kind of management method developed in Japan in the early 1970s [
25]. Organizations could use QFD as a methodology for setting policy at the management level and deploying the policy to the operation level for execution. QFD serves as a vehicle for TQM implementation by means of providing a quality-driven approach for making business planning as well as a customer-oriented approach for doing product and service development, with both of them effectively involve the participation of organization members.
Two kinds of deployment
There are two kinds of deployment in QFD: (1) interpretation and (2) conversion. As early as in the early to mid-1960s, Japanese manufacturers had already made use of quality charts, which were first used in Kobe Shipyard, to help convert the qualities demanded by buyers into corresponding parameters to assure the make and the production quality of their products [
26]. Some early QFD cases of the automotive industry revealed the use of quality charts to conduct the conversion kind of deployment for managing quality. For example, Toyoda Gosei applied QFD charts in formulating the assurance network for managing the quality of rubber and plastic parts [
27] and Hino Motors used the charts to link design quality to production methods and final assembly for cost control [
28]. The conversion kind of deployment greatly helps translate items of one aspect, for instance, the given requirements, into items of another aspect, such as the features, the parts, and the processes, which would collectively contribute to meeting the given requirements. However, it does not help find out the items at the origin, that is, the needs of the customers with the product. The formulation of the interpretation kind of deployment has overcome this limitation. This added deployment is a process starting from collecting the voice of the customer (VOC) from various sources, extracting need items from the collected VOC, organizing the need items into needs, and finally identifying the important needs—the control points for assuring the design quality. Upon completing the interpretation kind of deployment, the team could use the conversion kind of deployment to continue to assure the make and the production quality. The two kinds of deployment of QFD enable manufacturers to identify control points at every stage of manufacture to implement quality assurance.
Industrial applications
QFD has been widely applied in two major areas of business management: (1)
hoshin kanri, or policy management, and (2) product and service development. For the first area of
hoshin kanri, including both policy planning and policy deployment, QFD provides a visible link starting from the capture of VOC in the very early
hoshin generation process down to the development of breakthroughs for achieving the
hoshin objectives and effectively facilitates communication between departments [
29,
30]. For the second area of product and service development, many applications just made use of certain subsystems and/or components, instead of the full system, of QFD. Because of this, applications are of great variety. Some used QFD simply to understand the customers’ needs. For example, a laboratory of NASA used QFD to gather the requirements and analyse the diverse needs of the internal customers to design a subsystem programme [
31]. Some used QFD to identify quality characteristics, or design parameters, and/or to define technical requirements, such as the case shared by a hotel in Ibadan used QFD and Pareto analysis to identify the “vital few” items to optimize resources for improving its services [
32]. Some focused on defining the product and service features with QFD, such as the studies of mountain bike [
33] and pruning shears [
34]. There are some studies, of course, aimed at using QFD to design actual products. An example is the study on designing a device for holding an endotracheal tube of the patient in a more secure manner during anaesthesia [
35].
QFD is not new to the passenger transportation industry. Here are some applications. An airline used a modified QFD to understand the customers’ needs with respect to flight attendants, in-flight products, and cabin environment in the design of a new service for its intercontinental business class [
36]. In a study rectifying the problems found in the launching phase of a smartcard for paying riding fare of public transportation, the team used QFD to translate the VOC into the technical specifications to improve the electronic system [
37]. In Sapporo, there is a case on using the house of quality to make suggestion for improving the road maintenance service in winter. The team first conducted an evaluation by identifying the road users’ dissatisfaction items and then deployed the dissatisfaction items into the technical aspects of the service to formulate advice [
38]. Recently, the passenger transportation industry has applied QFD to handling the voice of multiple customers. For example, in a study on suggesting an airport in Japan on how to ensure the generality and sustainability of quality implementation, the team formulated a multilayered QFD model to deal with the conflicting requirements of the airline companies and the passengers [
39]. A study conducted for highway bus services took similar approach. The team used fuzzy QFD to integrate the views of the service provider and the customers to devise quality improvement strategies [
40]. Although the areas of application of the studies were diversified, they carried the same aim of improving quality and they all started with customers.
Dealing with the fuzzy front end
VOC is a raw form of data that would inform the supplier about the needs of the customers. From the QFD point of view, VOC is the basis for partnership, both within the supplier and between the supplier and the customers [
41]. To be a responsible and professional supplier, it is necessary to take active role in capturing and deploying the VOC all the way down to staff actions [
42]. However, VOC is elusive in nature and seldom explicit. In order to collect VOC as complete as possible, multiple collection methods are required [
43]. The set of collection methods used depends on the nature of the subject that is understudied. In an electric wheelchair project, the design team collected VOC by arranging focus groups to meet the current users as well as conducting activity analysis with the situations in which electric wheelchair was used [
44]. However, in a development project of a phone service company, the team used a variety of methods to collect the VOC, including diary method, group interview, critical incident, and problem detection, to collect the “Whats” for QFD as a huge customer group and diversified needs were involved [
45].
Gemba visiting is a method commonly used to collect VOC in QFD. “Gemba” is a Japanese word meaning “real place”, a place where real actions are to be taken. Imai contended that
gemba is the site of all improvements and the source of all information for adding customer-satisfying value [
46]. Since many customers’ needs are often unvoiced, “going to the gemba” therefore could help a supplier understand how and under what circumstances its product or service was being used [
47]. In general,
gemba visits provide two major kinds of information for improving product and service design. First,
gemba analysis could help identify the potential failure modes and root causes in designing products and services, which often missed in conventional problem analysis [
48]. Second,
gemba visits could assist suppliers on discovering customers’ latent and unspoken demands so that the product or service to be developed could surpass their basic requirements [
49,
50]. Many applications have reported using
gemba visiting to collect information about the customers’ needs. A cement manufacturer in New Mexico improved their technical support services by paying visits to their key customers to document their cement use and concrete production processes [
51]. A home utensil company adopted the
gemba technique and used it to encounter consumers in different situations to help the development of more innovative concepts [
52]. In the project on designing a dinosaur robot for a theme park, the engineers visited a petting zoo to observe how children interact with live animals [
53]. A phone manufacturer conducted
gemba visits in the metropolitan areas of Tokyo to capture the basic and latent needs of cellular phone users [
54]. To improve customer satisfaction with bagel sales in the airports of the USA, the team discovered from
gemba visits that customers wanted toasted bagels but had never offered before [
55]. For the project on understanding how consumers identify with the brand of a beer, the team members went to supermarkets to observe how people buy beer and went to pubs to see how people drink beer [
56,
57].
Interacting with customers is another method frequently used to collect VOC in QFD, which is usually conducted in the form of focus group or individual meeting. The common approach of this method is to ask customers to give comments for the existing product or the service and/or their expectations with the product or the service. An example is an improvement project of a distributor of convenience food to retail outlets. The team invited the key players and customers to come to a meeting to discuss and to identify the main problems within the existing supply process [
58]. Another example is a project on making a service plan of a regional acute-care hospital in Singapore. The team asked the patients to give their expectations with health providers [
59]. Although in most cases the responses are audio-recorded, some teams would like to interact with customers in a more natural way. They simply dialogue with customers and listen to their stories. In a project on boosting morale, the team invited the employees to come to a lunch meeting to let them voice out their difficulties and feelings [
60]. Another example is a project on using QFD in designing maternity party dresses. In an afternoon tea meeting, pregnant women and mothers-to-be shared their stories on attending banquets with their “big bellies” [
61].
The processing of the collected VOC starts from extraction, that is, to bring out the embedded meanings from the voice and to turn the embedded meanings into items informing the needs. For example, in a surgery satisfaction survey, the requirements of “good appearance” and “small wound” were extracted from the voice of “want unnoticeable scar” and “want to wear bikini”, respectively [
62]. Similarly, in the development of a finger vein authentication device, the requirements of “pleasant to touch” and “appropriate size for a hand” were extracted from the voice of “pleasant to use” [
63]. Interpretation usually operates with a two-time extraction. Besides, it is easier to decompose a statement and reword a phrase; the in-between pause helps enrich the contents of the deployment. Interpretation not only to be made to verbal expressions, but also to images, such as photos and videos, as well. Referring back to the maternity party dress project, the team searched photos from the web and extracted possible embarrassing situations pregnant women would like to avoid at party time [
64]. Similarly, in the project on finding out the passengers’ needs of the seat of high-speed rail in China, the team members extracted need items from the photos they took on the train [
18].
The next step is to group the need items into needs. Affinity diagramming, a bottom-up clustering technique, is often used for this process. Affinity diagram has its origin in the KJ Method® developed by Jiro Kawakita, a Japanese anthropologist, for establishing an orderly system from chaotic information. This technique provides an approach for grouping items that are naturally related and helps identify one concept generic enough to tie the group [
65]. It is highly effective for bringing the picture of a matter clearly into view and is a creative process that could break through preconceived notions about the situation. An example of using affinity diagramming in QFD application is the development of an online travel agency website. The team used this technique to put the needs collected from interviewing the participants of web design courses into customer requirements [
66].
The identification of important needs is to put the needs into a survey for customers to indicate how important the needs to them. Traditionally, the indication is by rating or by ranking. However, the ordinal numbers collected in such ways could hardly produce valid mathematical meanings. Among the methods suggested for solving this problem, analytic hierarchy process (AHP), a decision-making model formulated by Saaty [
67,
68], is one that has been widely used in QFD. There are two reasons. First, AHP asks respondents to give judgement by making pairwise comparison so that the received responses on the importance of the given needs would be actual and more exact. Second, AHP uses ratio numbers to present priorities. As ratio numbers are mathematically operative, subsequent deployment of the priorities therefore is feasible as the priorities could be transferred from one matrix to another matrix with high accuracy [
69‐
71]. An example of using AHP in QFD is a project of an insurance company on finding out the priorities of need attributes of various customer segments and the management of [
72].