Draft 2

QUALITY FUNCTION DEPLOYMENT (QFD)

Listening to the Voice of the Customer
Don Clausing
MIT ESD.33
It is imperative for success to relate all systems-engineering decisions to the voice of the customer and to the long-term improvement of QCD. However, in the day-to-day decision making,
experience has proven that it is all too easy to become overwhelmed by more local, parochial
goals.1 The original voice of the customer (VoC) has to be deployed horizontally through the
phases of development: product planning, design, production-process planning, productionoperations planning, and back to the customer in the form of the new system. The VoC also has
to be deployed vertically down through the levels of the system. In these many steps of deployment it is easy to lose sight of the customers imperatives unless we have a disciplined method to
guide us. Quality Function Deployment (QFD) is that disciplined method.
QFD is the subject of many books and seminars; you can learn the many details leading to mastery of this tool by reading a good book2, by attending seminars, and most important of all, by
using QFD. Our purpose here is to discuss the main points of the process, and to provide an extended view of QFD that best meets the needs of the Systems Engineer.
One lens on QFD is to relate it to the RCI process. The part of Figure 1 that is enclosed in red is
the core of QFD. V is the voice of the customer; more generally all stakeholders. E is the expectations of the enterprise for the new system. V and E cover the same information, but V is in the
voice of the customer, while E is in the language of the enterprise. T means that we translate
from the customers relatively qualitative language into the more quantitative engineering language of the enterprise. Experience has shown that this translation can introduce serious distortions unless it is done very carefully.
Before proceeding we briefly introduce the remainder of Figure 1. F is engineering functions, as
typically done in functional analysis; e.g., in value engineering. A means that the expectations E
can also be viewed as amplifications of the functions F. More about this later in this note.
TSP is technology-source planning and RM is reusability matrix. This planning for new technology and reusability is usually considered part of systems architecture, and will not be discussed in this course.

See the QCD note for Session 1 of this course.

We recommend Quality Function Deployment: How to make QFD work for you, by Lou Cohen, Prentice Hall/
Addison-Wesley, 1995.
1

Draft 2

F
A
E
V

I
TSP
RM
Figure 1. Context for QFD.

H refers to requirements passed down from higher levels of the system. More about that later.
The entire QFD process is conducted by the QFD team, also called the development team. This
team consists of all individuals who are responsible for the development, delivery and servicing
of the system. The team is often called a multifunctional team, because its members represent all
the key functional groups within an organization Marketing, Architecture and Design, Manufacturing, and Service. For large systems many of the main functions mentioned here are themselves multifunctional, so the membership of this critical QFD team must be decided on with
great care.
HOUSE OF QUALITY
We now turn to a more detailed description of the elements of QFD. For this well refer to Figure
2. Here we see a diagram in which each rectangle (and triangle) represents a task or step in the
process of reconciling customer needs with technical expectations. The triangle at the top of the
diagram is actually a matrix, cut along its diagonal and rotated 45 degrees. The complete planning chart in Figure 2 evolved over several years, and when complete it was seen to look like a
house; thus the name the House of Quality. Now that the diagram is seen to be a house, the elements from which it is composed become rooms. Well now take a tour through the House of
Quality.

Draft 2

6
Expectations

Voice
of
Customer

1 (V)

Expectations
Correlations

2 (E)
3 (T)

Priorities
Technical
Benchmarks

5
7
8

Market
Data

Relationships
Technical
Targets

Figure 2. House of Quality

Room 1 is labeled V, the Voice of the Customer. The VoC is represented by a hierarchically arranged list of customer needs, each expressing a desire or need about the system in the customers own language. Examples of such needs: Fly with one engine, A quiet ride, Doesnt
wear out.
The needs expressed in Room 1 are qualitative only.3 It is not possible to determine the relative
importance of these needs based on interviews only. The work done in Room 4, Market Data,
will enable the team to rank-order these needs.
These needs are typically collected as a result of in-depth, probing customer interviews. The interviews would normally be recorded and transcribed. The transcripts are studied by the team
members. The team members identify and extract the customers phrases that represent needs.
3

Sometimes customers do give numbers, but usually the system-engineering team should concentrate on the customers qualitative statements of needs. These will subsequently be quantified by competitive benchmarking and
other market information.
3

Draft 2

The QFD team must sort out true customer needs from possible solutions, or designs intended to
meet those needs. A good need set will be independent of the solutions for meeting those needs.
For example, if in an interview a customer asks for seats with 2 inches of padding, the astute interviewer will recognize that this is not a need at all; instead its the customers attempt at solving
an unstated need. The intrepid interviewer will then probe: If you had 2 inches of padding, what
would be the benefit?. The customers explanation of why the padding is desired will lead to the
true need, in this case for No muscle fatigue after a long ride. Note that this is an example of
not being mislead by the customers initial quantitative statement.
Room 2, labeled E, holds the development teams technical expectations in its columns. Good
expectations are expressed in measurable terms. Ideally the measurement is along some continuum. There will be a place on the continuum which represents the ideal performance of that expectation. The best performance may be at infinity (Larger is better), at zero (Smaller is better) or at some fixed value (Nominal is best.) While the team may never be able or even willing to meet the ideal value, these categories provide a direction of goodness that will be crucial
for setting meaningful system requirements.
The link between V and E is the translation matrix, labeled 3 (T). Each cell in the translation matrix corresponds to one row (need) and one column (expectation.) The QFD team must decide to
what extent performing well on that expectation helps the system to meet that need. This impact
is expressed as a number. One of the main tasks in QFD is to complete this process for all the
cells of the translation matrix, thereby evaluating all possible relationships between expectations
and needs. After doing that (and after completing Room 4) it becomes possible to prioritize the
expectations, thus enabling meaningful tradeoffs during the detailed system design phase.
Room 4 provides the quantitative analysis needed to rank-order the needs in Room 1. Room 4
contains several columns, each containing the results of different types of quantitative analysis of
the needs. Typically these columns display the results of surveys of many customers, measuring
the relative importance of the needs to the customers, and their current levels of satisfaction with
the companys current systems and with the competitions systems. Other columns contain strategic judgments made by the team. The results of all the columns are combined into a single
number, called the raw weight, representing the relative importance of each need.
The remaining rooms are repositories for other key types of information that flesh out the requirements process:
Room 5 displays the priorities of the expectations, based on the raw weights from Room 4, and
the impacts from Room 3, to produce weighted sums for each column (for each expectation.)
Room 6 analyses the interactions of expectations with each other.
Room 7 displays the results of technical competitive benchmarking.
Room 8 displays target values for each expectation, based on all the information in all the other
rooms.
A great deal of work is required to implement QFD as described thus far. The benefits of doing
this work are considerable.
First, all the work is done by the team of key developers. As the team grapples with its inevitable
disagreements and conflicting interpretations of a QFD matrix, they converge on a common un4

Draft 2

derstanding of the customers needs and on the system requirements. This provide great value. It
results in more consistent decision making, and more efficient work at every subsequent step in
system development. A critical element for success is the conviction by the system-engineering
team that their system will be welcomed by its potential customers and they know why it will
be welcomed. Working through the House of Quality gives them this confidence.
Second, the QFD structure promotes systematic analysis of a myriad of inputs. Finally, the QFD
process provides a roadmap to guide the team through the most far-reaching phase of system development: the requirements phase.
Next we go into more operational detail.
CORE OF THE HOUSE
The core of the House of Quality (HoQ) in Quality Function Deployment (QFD) is the combination of Customer Needs, Engineering Characteristics, and the Relationship between them. It is
important to recognize that both the Customer Needs and the Engineering Characteristics are requirements or specifications that the product system must satisfy. The distinction is that the Customer Needs are in the language of the customer, while the Engineering Characteristics are in the
corporate language of the system providers. A classic mistake is to use the columns (Engineering
Characteristics) to show design responses. This is premature. First we have to define the needs in
corporate technical language, before we can start making a design response.
The core of the HoQ is displayed in Figure 3.

ENGINEERING
CHARACTERISTICS

CUSTOMER

RELATIONSHIPS

NEEDS

Figure 3. Core of House of Quality (HoQ).

Many different variations of words are used for Customer Needs and Engineering Characteristics. The essential distinction is that the former is in customer language and the latter is in the
language of the provider. Customer Needs are measured by customer response. Engineering
Characteristics are measured by technical tests. The Relationship matrix helps to match the right
technical test with the related Customer Need or needs.
For a very simple product, Figure 3 is sufficient definition for the core of the HoQ. Do the market
research, enter the 5 to 15 important needs in as the rows, find the related engineering character5

Draft 2

istics that can provide technical guidance, and show the relationships between them. In its simplest form the Relationship matrix is diagonal, one Engineering Characteristic matching one
Customer Need. When the product is that simple, that is all that there is to the core of the HoQ.
However, many products are far too complex for this simple approach to suffice. There can be
many tens or even hundreds of Customer Needs. Also, the needs that are captured from customers can be a confusing array of apples and oranges, including obvious parents and children (some
are subsets of others), crying for further clarification. Therefore, more needs to be done to provide clear guidance in the planning of the new product.
CUSTOMER NEEDS (ROOM 1)
The customer needs are developed by market research. It is usually advisable for some of the
technical provider-enterprise people to be involved in this activity. Here we assume that this has
been well done, and we start with the needs that are relevant to the customers.
The needs present three challenges: (1) an overwhelming number of needs, (2) children and parents are included, and (3) different types of needs. There are two methods that have long been
used and are current best practice for meeting these challenges. They are the KJ Method, which
addresses challenges 1 and 2, and the Kano characterization, which addresses the third challenge.
KJ Method
The KJ Method was developed by Kawakita, Jiro (KJ), a Japanese anthropologist, starting in the
1950s. Later it was imported into the business world, and had been used successfully there for
many years.
The basic concept is to write each need on a notepaper or card, and then affix it to a large sheet
of paper on the wall. Then the team sorts these into groups by moving the slips around.4 When it
is feasible, it is good to have customers involved.
For example, there might be 50 needs. These might be organized into 15 groups at the second
level. These in turn could be organized into four groups at the third level. With numbers such as
these the team would usually then decide to work primarily at the second level. Fifteen needs are
feasible to work with, but 50 are typically too many. That takes care of the first challenge, too
many needs.
Also, the groups should take care of parents and children that are initially mixed together. They
will now be on different levels of the KJ (affinity) diagram.
Many teams do little more than has been described shuffle slips around to form groups. This
does form affinity groups, and it is acceptable to call this the affinity method. However, it does
not achieve the results that can be obtained by a more developed method.

Today it would be feasible to do this electronically. However, the social dynamics would be drastically changed.
6

Draft 2

A good summary of the KJ Method is in the book by Shiba, Graham, and Walden [1993]. They
show six major steps:
1. Agree on a topic
2. Write and understand the data
3. Group similar data
4. Title groups
5. Lay out groups and show relationships among groups
6. Vote on the most important low-level issues and draw conclusions
In one version of this there are a total of 19 detailed steps. This has been taught by Professor
Shiba in his classes at MIT and in his work with the Center for Quality of Management. Americans tend to have the reaction that a process with 19 steps is too much detailed rigor. However,
actual use of the 19 steps has brought results that are far beyond what could have been achieved
by simply shuffling cards around in an undisciplined way.
Kano Diagram
Kano pointed out that not all customer needs are of the same type. Rather there are three very
different types. This distinction is critical to effective use of the HoQ.

R
IE

TI
SF

SA

LI

EA

CUSTOMER
SATISFACTION

DELIGHTER

CORPORATE
ACHIEVEMENT

ST
MU VE
HA

Figure 2. Kano diagram three types of needs.

The type of need that is most frequently considered in the HoQ is the linear satisfier. Double the
achievement doubles the satisfaction; gas mileage is an example.
An example of a must have is brakes. A car must have brakes. However, in the evolution of
automotive technology there has been very significant advances in brakes. So even for a function
that the system must have, there is still opportunity for competitive advantage.
A delighter is a function that the customer was not expecting and is delighted to find. Sometimes
the entire product is a delighter, such as has happened recently with the Apple iPod.

Draft 2

Clearly the teams work to quantify the needs and the technical responses to them must take into
account these different types of needs. If a must have is missing, it can outweigh all of the linear
satisfiers.
Sometimes a delighter can be accomplished by combining together outstanding achievement on
all of the linear satisfiers. One could assert that this is the case with the Apple iPod. Its individual
features, such as a very small disk drive, are high up on a linear-satisfier scale. The combination
is a delighter.
Summary of Customer Needs
The customer needs must be carefully structured before there is any attempt to use quantification
to focus attention on a subset of needs that will be most productive in the enhancement of customer satisfaction. The structuring methods that are most developed are the KJ Method and the
Kano diagram.
Quantification inevitably means trying to convert human feelings and language into numbers.
This will always have much room for further research.
ENGINEERING CHARACTERISTICS (ROOM 2)
These are the columns in the HoQ, the technical equivalents of the customer needs. These are an
alternative statement of the requirements for the system, not yet saying anything about the design
of the system. This form of the requirements is technical and measurable. It has two sources: (1)
responses to the customer needs, and (2) amplification of the functions of the system.
Technical Responses
This approach is straightforward. Take each customer need (at the appropriate level) and brainstorm technical means of measuring the system performance that will assure customer satisfaction on the selected need. Then work this down to the one or two that will best guide the design.
As an example consider the paper-handling subsystem for a copier/printer. Typically a customer
need will be that it is reliable. One technical response is paper-handling shutdown rate. Such
metrics evolve over time to best match the need that is most important to the customers. Originally this metric was shutdowns per million copies. Eventually it was recognized that the customers were really affected by shutdowns per week. This became the technical response.
The HoQ is not an end unto itself. It is a critical step to guide the selection of the system architecture, including technology selection. For a paper feeder there will be two or at most three
technologies available for integration into the new copier/printer system. There will be the lowcost, low-reliability feeder, and the better feeder technology. If the shutdown rate per week for
the low-manufacturing-cost feeder is less than the specified amount, about one per week, then it
is selected. It the copier/printer will be used more often, so that the shutdowns per week for the
cheap feeder would be excessive, then the more reliable feeder is worth the extra initial cost.
This approach of finding one or two direct responses to measure the system performance for each
customer needs is always a good starting point. However, for some types of needs it is insuffi-

Draft 2

cient. Take the famous Toyota rust-prevention case from the 1970s. The customer needs (at the
third level) were:
1. Rust prevented while driving for any purpose
2. Rust prevented in all driving modes
3. Rust prevented in all maintenance conditions
4. Rust prevented in all natural environments
5. Rust prevented in all driving environments
6. Rust prevented in all road conditions
Exactly what these meant was spelled out in the two more-detailed levels.
The technical responses were in the following columns:
1. Rust prevented at edges
2. Rust prevented at joints
3. Avoid rust that is caused by defective part
4. Provide rust-prevention paint film
5. Resist corrosion holes
6. Resist spot rusting
An excellent job was then done in applying QFD to deploy these into the design and then into
production.
Note that although there are six customer needs and six engineering characteristics, there is definitely not a one-to-one matching. Reflection suggests that the detailed relationships between the
rows and the columns in this case are not important. What is important is coverage. Do the engineering characteristics cover all of the needs?
Functional Analysis
The engineering characteristics can also be derived from the functions (F in Figure 1.) that the
system must perform. This is not an alternative to deriving the technical responses from the customer needs, but a supplemental way that provides additional insight.
Functions are usually stated as a verb and a noun, make copy, for example. Clausing [1989,
1993] and his students [Pandey 1991, Sontow 1992], and Pahl and Beitz [1984, 1996] have
pointed out that there is a generic amplification or expansion that is associated with each function. This comes in two parts.
The first part is the primary purpose of the function. All functions transform and/or transport energy, material, geometry, and/or signals.

Draft 2

The second part is the expanded requirements that are associated with any function. When we
specify that the function is make copy, we mean that a copy should be made for a certain cost,
with safety, satisfying all social regulations, etc. Clausing and his students generated the following starter list for the amplification (A in Figure 1.) from a function:
1.! Explicit customer requirements
1.1.!Cost
1.2.!Size
1.3.!Mass
1.4.!Appearance
1.5.!Feel
1.6.!Taste
1.7.!Smell
1.8.!Sound
1.9.!Life expectancy
1.10.!
Maintenance cost
1.11.!
Operational range
2.! Implicit Customer Requirements
2.1.!Manufacturing, distribution, Servicing
2.1.1.! Dimensions
2.1.2.! Achievable tolerances
2.1.3.! Packaging
2.2.!Safety, ergonomics, latent needs
2.2.1.! Operational safety
2.2.2.! Environmental safety
2.3.!Government regulations
2.3.1.! E.P.A. regulations
2.3.2.! F.D.A regulations
We see that these are the type of characteristics that go into the columns of the HoQ. Therefore,
the team can review these to help generate the columns in the HoQ. However, most of these will
be judged to be insufficiently important for the HoQ. They will not help in making the subsequent system architecture decisions. Instead they will be addressed by standard tests and
knowledge-based engineering.
This is displayed in Figure 3.

10

Draft 2

Lens 1
Functional Tree

Lens 2
QFD matrix

FEED
SHEET
RETARD
SUBSEQUENT
SHEETS

TRANSPORT
SHEET

BREAK
SLUG
PROVIDE
WRAP
ANGLE
!

SEPARATE
SHEETS

Function amplified into requirements

AVOID
OVERLAP

PROVIDE
RETARD
FRICTION

PROVIDE
BELT
TENSION

PROVIDE
RETARD
LENGTH

rp

Lens 3
Function amplified into:
Standards
Knowledge-based
Engineering

User needs amplified into requirements

Figure 5. Expansion of functions into engineering characteristics.

Figure 5 displays that the columns in the HoQ can be thought of as coming from both the rows,
and from the amplification of the system functions.
RELATIONSHIP MATRIX (ROOM 3)
The relationship matrix is a formal analysis of the relationship between the customer needs
(rows) and engineering characteristics (columns) in the HoQ. However, as has been described,
the relationships have already been much considered during the development of the rows and
columns, especially the columns. In many cases there will be little more that needs to be done.
We have already seen in the rust-prevention case that the relative importance of the six engineering characteristics was not significant. All six were important, and all had to addressed successfully. The importance decision was to work on rust prevention. This was largely driven by
competitive benchmarking and customer reaction.
The HoQ is not an end unto itself. It is an important step in the development of the market-attack
plan (MAP). The objective is to plan a new system that will have competitive advantage. The
HoQ is a major step in the identification of characteristics that will provide competitive advantage. For example, the rust-prevention project gave Toyota a tremendous improvement over their
previous vehicles, and transformed rust prevention from a competitive disadvantage into a competitive advantage for Toyota.
In the relationship matrix we want to avoid excessive analysis that cannot help our team to
achieve competitive advantage. When we are guided by the role of the HoQ in the development
of a MAP, then we can include only those characteristics that are essential to success.
11

Draft 2

Current Practice
The best current practice is to use 9, 3, 1, and 0 in the cells of the relationship matrix to quantify
the relationship between the engineering characteristic and the customer need. (Cells that are left
blank are 0, or more precisely, insignificant relationship.)
The non-linear (geometric) sequence of 9, 3, and 1 tends to emphasize the relative importance of
the most important characteristics.5 This is the primary reason for using this relationship ranking.
Engineers are good at finding many relationships. A common problem is to fill the relationship
matrix with symbols, such as 9, 3, and 1. A rough rule of thumb is that at most only about 1/3 of
the cells should be filled. More than that suggests that there was too low a hurdle in the identification of a relationship.
Using 9, 3, and 1 while taking care to keep the matrix sparse is usually sufficient. However,
many other approaches have been used. An obvious one is to use 5, 3, 1, and implicitly 0.
Many teams use graphical symbols in the cells, such as thin hollow circles, thick hollow circles
and solid circles. In making calculations these have numbers, such as 9, 3, and 1, assigned to
them.
The common practice is to use only positive relationships between the engineering characteristics and the customer needs. Conflicts between engineering characteristics are identified in the
attic of the HoQ, where both positive (synergistic) and negative (anti synergistic) pairs of engineering characteristics are identified.
These weights are determined by the multifunctional system development team, using their best
judgment. Market researchers sometimes use a quantitative method, conjoint analysis [Urban and
Hauser 1993]. Multiple versions of a system are defined and described, either on paper or on a
computer, which can also be displayed online. By analyzing the customers perceptions in response to the different system-feature sets quantitative information is derived about the relationship matrix.
The relationship matrix is used to calculate the relative importance of the engineering characteristics. The importance of each need is multiplied by the relationship weight, and all cells in a
column are then added to give the importance rating of that engineering characteristic. However,
there are other factors that go into deciding which engineering characteristics to emphasize.
QUANTITATIVE MARKET DATA (ROOM 4)
Each customer need is characterized by more than its importance. This is displayed in Figure 6.

In some literature this is called a non-compensatory scale; i.e., 1+3 cannot compensate for a 9.
12

Draft 2

Importance to customer
Customer satisfaction performance
Competitive satisfaction performance
Goal
Improvement ratio
Sales point
Raw weight
Normalized raw weight
CUSTOMER
NEEDS

Figure 6. Room 4 of the House of Quality. After Cohen [1995].

In addition to the customers perceptions of importance, the competitive situation is a major
driver in deciding on the new product. The relationship matrix is used to convert this into the
competitive rankings on the engineering characteristics based on customer perceptions. This is
compared with the actual technical evaluations that are displayed in one of the lower rooms of
the HoQ. Finally all of the information is used to calculate first the raw weights and then the
normalized weights for each customer need. The relationship matrix is then used to translate this
into guidance on the engineering characteristics.
The primary objective while working in Room 4 is to identify sales points that will make the new
system attractive to the potential customers. This competitive advantage is critical to the revenue
and profit of the enterprise.
PRIORITIES (ROOM 5)
In Room 5 the relative importance of each engineering characteristic (column) is calculated. This
is determined from the information in Room 4 and the relationship matrix.
Systems development teams must not allow the importance scores in Room 5 to lead into mindnumbing numerology. A priority score of 100 vs a priority score of 95 is not a significant difference. What is important is to find some cluster of customer needs, translated into the supporting
engineering characteristics, that will provide sales points and give competitive advantage. The
priority scores are used to make a qualitative decision: this set of system features will give us a
winner in the marketplace.

13

Draft 2

CORRELATIONS (ROOM 6)
This is the attic of the HoQ. Each cell in this triangular matrix is at the intersection of two columns (engineering characteristics). Here the team evaluates the conflict or synergy between pairs
of engineering characteristics. Of particular interest are negative correlations (conflicts). In conflicts success in the achievement of one characteristic will have the tendency to cause difficulty
with the other characteristic. Special attention will be needed to achieve customer satisfaction on
both.6
TECHNICAL BENCHMARKS (ROOM 7)
Here are displayed scores from technical tests of our existing product and the leading competitive products. An example is horsepower.
It is important to note here whether superiority on the technical tests is matched in Room 4 by
customer perception that the product is superior. If not it means that the tests are measuring what
the engineers are interested in rather than what the customers respond to. Then further development is needed to make our technical tests and market evaluation consistent.
TECHNICAL TARGETS (ROOM 8)
Here is the final result. What values for each of the engineering characteristics will give us competitive advantage? 300 horsepower 10 years without rust image quality above 93 These
are the type of decisions that are recorded here.
The team now has to be committed to the evaluation that achievement of these targets will meet
the enterprise goals for revenue and profit. The steps of the HoQ are simply aids to the team in
reaching this conclusion.
CONTEXT OF QFD
There are several types of new-product developments:
1. Breakthrough product
2. Product family
3. Generic improvement
The role of QFD varies greatly for these different types of development.
Breakthrough Product
QFD has little role for a breakthrough product. Breakthrough products have a huge advantage
over previous technologies in a few characteristics. An example is the xerographic copier. When
Chester Carlson invented it in the 1930s he created tremendous advantages in three characteristics: (1) dry process, (2) automatic, and (3) good copy quality. These advantages greatly expanded the market for copiers to the benefit of the Xerox Corporation in the 1960s. QFD could
not have helped. However, as Xerox progressed on to its second and third generation copiers in
the 1960s and 1970s QFD could have been a big help. In practice Xerox started to reap the benefits of QFD in the late 1980s.
6

Sometimes we can find a way around the conflict. In other cases the best-balanced trade off will be successful.
14

Draft 2

In general it is true that breakthrough projects, as with xerography, succeed by making huge improvements on two or three characteristics. The team can stay focused on the two or three large
improvements, and does not need to depend on QFD to help deploy and balance 15 or 20 smaller
improvements.
Breakthrough products usually result from a major invention; e.g., xerography. Later in this
course we will consider TRIZ, a powerful aid for invention.
Product Family
The most important use of the House of Quality is to help in the planning of a product family.7
Here the great innovation in QFD has been the Master House of Quality [Clausing and Cohen
1994, Cohen 1995]. The several products in a product family will obviously have Houses of
Quality that are closely related. Cohen took this a step farther by developing the Master HoQ.
The key insight is that the customer needs and the responsive engineering characteristics are the
same for all market segments that are addressed by a product family. The difference among market segments is the relative importance given to each characteristic, and the performance that
will be competitive. The Master HoQ can be tailored to develop the specific HoQs for each product. This is illustrated in Figure 7. Quoting from Clausing and Cohen [1994]
the team created a handbook which explained to future product developers in the company
how to customize the master HOQ for their specific market segments. Each future product
team is expected to determine the relative importance of the four customer types and the secondary customer attributes for their market, plug those numbers into the master HOQ, and
thereby prioritize the performance measures for their market segment.
By experimenting with some hypothetical importance values, the team determined that
it would be possible for most future development teams to identify a small handful of
performance measures out of the master list of 150 that would be critical for success in
a particular market segment. Thus the master House of Quality will be the source for
many market-specific Houses of Quality that will be created by future market segment
teams.
This is with reference to a particular case study for which Cohen provided the QFD leadership.

A product family is several products based on a significant set of common parts. Each product addresses a different
market segment. Sometimes the products are said to be based on a common platform.
15

Draft 2

House of Quality for market segment A

House of Quality for market segment B

Figure 7. The Master House of Quality, tailored to two different market segments.
Clayton Christensen wrote a book, The Innovators Dilemma [1997], which became very popular. It is about the creation of new market segments. When technical capabilities are rapidly improving, as they were in the disk-drive industry that Christensen studied, new market segments
can make or kill a company. The Master HoQ is a good analysis to guide market segmentation.
The HoQ is a key element in the development of a Market-Attack Plan (MAP) [Holmes and
Campbell 2004]. To be perceived in context the HoQ, Master and individual, must be considered
in its larger role as an element of the MAP. To quote from Holmes and Campbell:
The strategic plan should constitute a complete map for achieving success in the targeted
market; accordingly, it can be referred to as a Market Attack Plan (MAP). The plan is resourced
for all elements, e.g., technology sets, products, services, and value chain enablers, that are
strategically aligned with market segments, key business goals, and corporate priorities. The
business goals, expected outcomes, and funding plans are documented in a corporate Plan of
Record.

16

Draft 2

With the end-to-end business perspective pervading the strategic front end, companies
achieve a more effective balance among new products and services, products currently
under development, and products and services already in the field. By funding the
product team and the total value chain as a whole, companies assure better alignment of
resources and priorities of the entire company with the portfolio decisions made in the
strategic front end.
This is the role of the HoQ for new products and product families. All evaluations, speculations, and proposals for improvement to the HoQ must past the test that they will improve the MAP. Focusing on the Relationship Matrix in isolation is a classical example of
minutely examining the twigs while being lost in the forest. Changes in the HoQ methodology will be significant only if they lead to different Market-Attack Plans.
The ultimate role of the HoQ is to be a step in the development of the Market-Attack Plan. Here
we start with the Market-Attack Plan, and work backwards to identify changes in the HoQ that
would make a difference.
The MAP addresses the following questions [Holmes and Campbell 2004]:
! What business are we in?
! What markets do we/ can we serve?
! In what segments can we effectively participate?
! How large is the business opportunity?
! What are our expectations for the competitive environment?
! What are our key value propositions?
! What will we establish as major vectors of differentiation?
! What are the platforms to address the targeted customer and markets?
! What architectures and technology sets will enable these platforms?
! What will we offer as product families and services?
! How will we organize the value chain strategies and enablers?
! What are the resources required to deliver against the product and strategic plans?
! What is our expected integrated business outcome?
The MAP is a complete, integrated plan for success in the targeted market. How can the HoQ
contribute to the development of the MAP?
Consider the first four questions.
! What business are we in?
! What markets do we/ can we serve?
! In what segments can we effectively participate?
! How large is the business opportunity?
These questions are addressed by a Master HoQ. A single-product HoQ in the absence of a Master HoQ will likely leave these questions not effectively addressed. To be an effective part of a
MAP the HoQ must be related to a Master HoQ.

17

Draft 2

The next three questions are:

! What are our expectations for the competitive environment?
! What are our key value propositions?
! What will we establish as major vectors of differentiation?
These are addressed in Room 4 of the HoQ, see Figure 4. Quantification must help to translate
these customer perceptions into technical objectives for design and production.
The next three questions address the system architecture:
! What are the platforms to address the targeted customer and markets?
! What architectures and technology sets will enable these platforms?
! What will we offer as product families and services?
This emphasizes the point that the HoQ can only guide decisions among available options. For
any function of the system there are usually only two or three options to choose from. These decisions are dominated by a few characteristics. The only changes in the HoQ that we need to
consider are those that would change a system decision.
Generic Improvement
The rust-prevention case study is a perfect example of this. The HoQ was not the most important
element of this case. It was used to help in the decision to make a major effort on rust prevention.
Most of the help from QFD in this case came from the subsequent QFD matrixes, design, production planning, and production operations.
The rust-prevention case provided Toyota with great competitive advantage, but it was not a
breakthrough. As noted earlier, QFD is superfluous for a breakthrough development, such as xerography.
COMPLETE QFD
Complete QFD includes horizontal deployment from the HoQ to production processes, process
planning, and production operations. It also includes vertical deployment downward from the
total system, to subsystems, and on down to piece-part features. The role of the HoQ in the total
development is revealed by the complete QFD, and will vary by project. For generic improvements, such as rust prevention, the additional QFD steps are often more important than the HoQ
itself.
Horizontal Deployment To The Factory Floor
The nature of the horizontal deployment to the factory floor is displayed by the examples in Figures 8 11. These are taken from the rust-prevention QFD that was done at Toyota Auto Body in
the 1970s. These go from the voice of the customer in Figure 8 to the design of the die in Figure
11 that is used to stamp the sheet steel. Careful study of these four figures will reveal the general
nature of horizontal deployment.

18

Draft 2

PREVENT RUST
AT EDGES
Will not rust:
!

In any type of application

In any method of operation

In any maintenance condition

In any natural environment

In any transport environment

In any road condition

In any delivery environment

Figure 8. Example from House of Quality.

Figure 8 shows one column in the House of Quality. In order to have good rust prevention for the
vehicle it is necessary to prevent rust at the edge of the sheet steel that is used to manufacture the
body.

SHAPE OF BURRS

QUALITY!

COUNTERMEASURE!

METRIC

PREVENT RUST GOOD COATING ADHESION

GOOD SEALING
AT EDGES
CORROSION-RESISTANT MATERIAL

30 *
CYCLES

* ACCELERATED CORROSION TEST

GEOMETRY
Figure 9. Example from Parts Deployment.
The prevention of rust at the edges is decomposed down to three requirements to overcome the
three common failure modes. It is further noted that more than 30 cycles in an appropriate accelerated corrosion test is believed to be sufficient to make the vehicle a winner for rust prevention.

19

Draft 2

It is further noted that the shape of the burr that is left when the sheet-metal part is stamped from
the raw sheet will have a major influence on the rust prevention.

60 degrees minimum

0.2 mm
max.
Figure 10. Burr geometry on the edge of the sheet steel, good enough to prevent rust.
Figure 10 shows the burr geometry that is good enough to avoid the burr being a site for rust to
start. Qualitatively the burr must be blunt. It should not be a feather sticking out, which is easily bent and/or broken.

DIE HARDNESS
! DIE MATERIAL
! ! CLEARANCE
! ! ! GUIDE POST
! ! ! ! STEP ON DIE
SHAPE OF BURRS
!
!
!
!
RC

! ! ! 2 MM
! ! FITTING
! 6 TO 8 % OF PLATE THICKNESS
TENSILE STRENGTH, 50 KG/MM2
5560

Figure 11. Example from Production Process deployment.

20

Draft 2

Finally Figure 11 gives the characteristics of the die that is used to stamp out the sheet-metal
parts. When the die has the values that are shown, then the burr will be at least as good as shown
in Figure 10.
Now working backwards, if care is taken with the die as shown in Figure 11, then the burr on the
edge of the sheet steel will not be worse than shown in Figure 10, with the result that the rust
avoidance will be at least as good as shown in Figure 9, with the result that all of the customer
needs in Figure 8 will be satisfied. The strategic advantage will be that customers will prefer our
cars because they have superior rust prevention. This QFD was a major enabler for the Toyota
thrust into the American market.
Horizontal and Vertical Deployment
The table on the next page shows both horizontal and vertical deployment.

21

Draft 2

Symbols in
top row relate
to Figure 5
These are examples
of item(s) described
above in each cell

QFD Product
Matrix

Product
Functions

Pugh Concept Production

Selection
Process
Functions

EF E F

C H (CP)

EF EP

QFD
Production
Operations
Planning
Matrix
EP EO

House of
Quality

Total
System
Functions

Total System
Architecture

VoC ETS
Always get a copy
Misfeed rate

F TS
Make copy

C TS
Copier

Total
System
Process
Functions
P TS

Final
Assembly
Planning
Matrix
ETS ETSA

Final
Assembly
Operations
Matrix
ETSA ETSO

TS/SS Design
Matrices

Subsystem
Functions

ETS ESS

O
I

C
Total System
E
TS

Assemble
modules

Misfeed rate
Subsystem alignment

Subsystem alignment
Alignment procedure

Subsystem
Concepts

Subsystem
Process
Functions

Subsystem
Assembly
Matrices

F SS
Feed sheet

C SS
Paper feeder

P SS
Assemble
feeder

ETSA/ESSESSA

Subsystem
Assembly
Operations
Matrices
ESSA ESSO

SS/SA Design
Matrices

Subassem.
Function

Subassembly
Concepts

Subassem.
Process
Functions

Subassembly
Assembly
Matrices

ESS ESA
Misfeed rate (SS)
Wrap angle

F SA
Provide wrap

C SA
Belt wrapped
around roll

P SA
Assemble belt
and roll

ESA ESAA

Piece-part
Concept
Functions

Piece-part
Concepts

SA/PPC
Design
Matrices

ESA EPPC

Wrap angle
Beam straightness

F PPC
Position roll and
belt

C PPC
Al die-casting
beams

Part
Formation
Process
Functions
P PPC

Part Feature
Design
Matrices
EPPC EPPF

Piece-part
Feature
Functions
F PPF

Piece-part
Feature
Concepts
C PPF

Subsystem
SS
O
F

Misfeed rate
Misfeed rate (SS)

Subassembly
SA

Piece-part
concept
PPC

QFD Process
Planning
Matrix

Misfeed rate (SS)

Align feedhead

Wrap angle
Adjust angle

Align feedhead
Alignment steps

Subassembly
Assembly
Operations
Matrices
ESAA ESAO
Adjust angle
Adjustment steps

Part Formation Part Formation

Matrices
Operations
Matrices

Make castings

EPPC EPF

Straight beams
Casting temperature

EPF EPFO
Casting temperature
Controller operation

Final
Processing
Functions
P PPF

Final
Processing
Matrices
EPPF EFP

Final
Operations
Matrices
EFP EFO

O
Piece-part
features
PPF

M
E
R

Beam straightness
Machining precision

Position roll

Low distortion

Machine land
on beam

Machining precision
Clamping fixture

Clamping fixture
Clamp to datum D

This table is taken from Clausing [1995], and is in turn a further development from an earlier paper by Clausing and Pugh [1991]. The example is for a copier/printer paper feeder.
In this table deployment is done horizontally through the stages defined by the column headings,
and vertically through the system levels defined by the row headings.

22

Draft 2

One feature that is revealed is that complete deployment requires that concepts be selected,
which is done by the Pugh Concept Selection (convergence) process.
Also, the functions F are shown here for completeness. As discussed earlier, these can be amplified into the expectations E. For example, the function make copy has as one amplification the
misfeed rate. In other words, when the function is identified to be make copy, there are a set of
implied expectations, including that the misfeed rate from the paper feeder will be less than one
per week.
It is not to be expected that any one project will use all of the QFD that is implied by the table.
For example, only a tiny fraction of all piece-part features would be deployed by QFD. Careful
consideration will reveal which cells will give competitive advantage for each specific project.
SUMMARY
Quality Function Deployment (QFD) provides the systems-engineering team with methods that
help to carefully deploy the voice of the customer throughout the system-development activity.
This gives the team meaningful goals for all of their myriad of activities. These clear goals help
the team to achieve a system that is attractive to the potential customers.
Of course, it is easy for QFD, or any other method, to become bureaucratic. If it is allowed to
deteriorate into a mind-numbing exercise to fill in the forms, the result will be worse than useless.
If the team works in a style of constantly asking how they can leverage the many aspects of QFD
to develop a better system, then they will be able to ascertain the specific applications of QFD
that will best improve their system to make it attractive to potential customers.

23

Draft 2

REFERENCES
Christensen, Clayton. The Innovators Dilemma. Harvard Business School Press. 1997.
Clausing, Don. Quality Function Deployment: Applied Systems Engineering. 1989 Quality and
Productivity Research Conference. University of Waterloo. 1989.
Clausing, Don, and Stuart Pugh. "Enhanced Quality Function Deployment." In Proceedings of
the Design Productivity International Conference, Honolulu (1991): 15-25.
Clausing, Don. Total Quality Development. ASME Press. 1993
Clausing, Don, and Lou Cohen. Recent Developments in QFD in the United States. Presented at
Gaining Competitive Advantage by Design, Institution of Mechanical Engineers Conference,
Coventry, March 23 and 24, 1994.
Clausing, Don, "EQFD and Taguchi, Effective Systems Engineering." First Pacific Rim
Symposium on Quality Development (Sydney, 1995).
Cohen, Lou. Quality Function Deployment. Addison Wesley. 1995.
Holmes, Maurice F., and Ronald B. Campbell, Jr. Product Development Process: Three Vectors
for Improvement. Research-Technology Management, 47(4) July-August 2004, pp. 47-55.
Pahl, G. and W. Beitz. Engineering Design, London: Springer, 1996. (First edition 1984.)
Pandey, Amitabh, and Don Clausing. "QFD Implementation Survey Report." Working Paper of
the Laboratory for Manufacturing and Productivity, MIT, November 1991.
Shiba, Shoji, Alan Graham, and David Walden. A New American TQM. Productivity/Center for
Quality Management, 1993.
Sontow, Karsten. Integration of Quality Function Deployment with Further Methods of Quality
Planning, Diplomarbeit, Institute for Production Technology, RWTH Aachen, 1993.
Suh, Nam P. The Principles of Design. Oxford University Press. 1990.
Urban, Glen L., and John R. Hauser. Design and Marketing of New Products. Prentice Hall. 1980
and 1993.

24