Title: Logistics%20Decision%20Analysis%20Methods
1Logistics Decision Analysis Methods
- Quality Function Deployment Part III
- Presented by Tsan-hwan Lin
- E-mail percy_at_ccms.nkfust.edu.tw
2Construction of the HOQ
- The first section of the HOQ to be constructed
will almost always be the Customer Needs/Benefits
section. - Sections are also referred to as rooms.
- The Planning Matrix (also, Preplanning Matrix) is
often the second section to e constructed. - The third section of the HOQ to complete is the
Technical Response (also, Corporate Expectations)
section. - The fourth step is to complete the Relationship
section of the HOQ. - The fifth and sixth steps in completing the HOQ
are Competitive Benchmarking and Target Setting. - The seventh and usually final step in completing
the HOQ is to fill in the Technical Correlations
Matrix. - This part is also referred to as roof.
3Q A
4Technical Response - 1
- Just as the Voice of the Customer had a
qualitative and quantitative component (entered
into the Customer Needs/Benefits section and the
Planning Matrix section), so does the translation
of the VOC into the Voice of the Developer. - The translation (SQCs) will be placed in
qualitative form on the top and in quantitative
form at the bottom. - In QFD parlance, we use the term Quality
Characteristics to denote the customers needs
(the VOC). - The translation into technical term is called
Substitute Quality Characteristics because it
represents a translation from the customers
language into the organizations technical
language. - Most commonly, developers call this language the
Product Requirements (????) or the Design
Requirements(????).
5Technical Response - 2
- The nature of Product Requirements varies widely.
- Many organizations describe products and services
in more than one language, or more than one set
of terms (such as, Customer Requirements, Market
Requirements, Top-Level Specifications, Detailed
Specifications, or Technical Specifications, to
name but a few). - There is often much confusion about the
boundaries between various product description
languages, and there is little or no
standardization of the vocabulary of each
language. - The ideal relationship between various product or
service description languages is one in which the
languages are defined and ordered according to
how abstract, or solution-independent, each
language is. - If one product description allows for many
possible implementation, it is more abstract than
another which clearly describes or implies one
and only one implementation.
6Technical Response - 3
- One way that QFD practitioners describe various
levels of abstraction is with Whats/Hows
metaphor. - The language that appears on the left side of a
QFD matrix represents What is desired. The
language at the top of the matrix describes How
the developers will respond to the What. - To get the most out of QFD, the language of the
Whats should be distinctly more abstract than
the language of the Hows. - Given a term that describes some aspect of a
product, it is not always easy to decide what
language it belongs to. - Examples VOC tinted glass in an automobile gt
Needs privacy - (tinted glass gt What? How?)
- A practical method for dealing with the problem
of placing product requirements in the right
place in QFD is to use the Voice of the Customer
Table (VOCT).
7Technical Response - 4
How
How
How
deploying
(more detailed)
What
What
What
- Some generic formulation of SQC exist, notably
one by Stuart Pugh. - Such formulations can be used as starter kits
to get a set of SQCs established rapidly, and to
aid teams in arriving at a complete set. - The generic formulations must of course be
modified to meet the needs of the teams specific
project.
8Technical Response - 5
- Main categories of VOCT Part 2 (Review)
- Customer Needs (??????)
- Substitute Quality Characteristics (???????????)
- Functions (??)
- Reliability Requirements (?????)
- Target Values(???)
- Languages for Substitute Quality Characteristics
- (Top-Level) Performance Measurement(??????)
- Product Function (????)
- Product Subsystem (?????)
- Process Steps (????)
9Relationships
- The Relationships section provides a mapping
between Substitute Quality Characteristics on the
one hand, and Customer Wants and Needs on the
other. - Each Relationship cell represents a judgment,
made by the development team, of the strength of
the linkage between one SQC and one Customer
Need. - The strength of linkage is called the impact of
the SQC on the Customer Need. - There are four types of impacts.
- Judging impact of performance measures
- Judging impact of other SQCs
- Meaning and choice of impact values
10Determination of Priorities of SQCs
- Once the development team has determined all the
impacts or linkages, some simple arithmetic
provides one the key results of QFD the relative
contributions of the SQCs to overall customer
satisfaction (the priorities of the SQCs). - Relationship of technical response X to customer
satisfaction performance on need A Impact
(value) of technical response X to need A
Normalized raw weight for need A - Contribution of the SQC to overall customer
satisfaction Sum of Relationships over all
needs - The larger the contribution, the more influence
the SQC has on customer satisfaction performance,
and therefore the more important it is for the
product or service to do well in the
implementation of that SQC. - If the SQCs and their contributions are to be
transferred to the left side of another matrix,
it is useful to convert the contributions to
normalized contributions.
11Negative Impacts - 1
- It happens occasionally that a SQC is found to
have a negative impact on customer satisfaction
performance for a certain attribute. - Such negative relationships can happen when an
SQC has been selected for its positive impact on
one or more attributes, but the attribute with
which it is negatively linked has not yet been
considered. - Example SQC (wider range of services) vs. Needs
(transactions processed from a single source
ltpositivegt simple of use ltnegativegt) - Negative impacts complicate QFD computations.
- One way to solve this is to take negative value
in computation. - Compute two sums algebraic (signed) sum of
relationships and sum of the absolute values of
the relationships - If the difference between the two is small,
disregard the negative impact. If the difference
is large, the team is being challenged to define
one or more Technical Responses that provide
positive impacts only to replace the one that
contains negative impacts.
12Negative Impacts - 2
- An alternative way is,
- Express all impacts on customer satisfaction
performance as positive. - Study the negative impacts as they are reflected
in the Technical Relationships (the roof) section
of the HOQ.
13Q A
14Performance Measurements
- Probably the most valuable language for SQC is
the language of Performance Measurements. - These are measurements that the development team
derives directly from customer needs. - They should be general enough to be applied to a
product or service regardless of the specific
implementation. - They may provide ideal measurements (1) for
benchmarking competitive products or services and
(2) for providing a solution-independent starting
point for developing new concepts. - The standard method for developing Performance
Measurements is to begin with the customer
attributes (i.e., customer needs). For each
customer attribute, - Define measures
- Define measurements
15Define Measures - 1
- Defining measures is the process by which the
development team establishes the relevance and
the relationship between its measurements and
customers perception. - In a nutshell, the team translates (or deploys)
each customer need into a technical performance
measure. - For each customer attribute, define one or a few
technical performance measurements. Write these
along the top of the HOQ. For each measurement,
be sure that, - It can be measured while the product or service
is being developed, and before it is shipped or
deployed (i.e. it can be used as a predictor
rather than a lagging indicator of customer
satisfaction performance). - It can be controlled by the development team
(i.e. the team should be able to make decisions
that would effectively adjust the measurement up
or down to meet expected customer satisfaction).
16Define Measures - 2
- To be properly defined, each performance
measurement should be characterized in a few
ways. - First, the units of the measurement(????)should
be defined. - Examples voltage (in volts) time (in minutes)
process complexity (number of steps) accuracy
(defects per thousands of transactions) - Second, the direction of goodness(??????)should
be defined.
17Define Measures - 3
Desired Quality (Customer Need) Performance Measurement Direction of Goodness
Capacity Cubic feet of storage Larger the better
Price Dollars Smaller the better
Reliability Mean time between failure Larger the better
Speed Transactions per second Larger the better
BACK
18Direction of Goodness - 1
- The More the Better
- The implied target is infinity.
- Examples reliability (mean time between failure
MTBF) fuel efficiency (miles traveled per
gallon) bonding strength (pounds supported per
square inch of adhesive area) - Development teams need both the aggressive
objective (infinity) and an acceptable value for
the measure. - If we view targets as infinity (instead of an
acceptable value or tolerance(??)for the metric),
we may see possibilities for increasing
performance that we have not seen before. - Examples secondary storage capacity (1.4Mb
floppy gt 125Mb USB gt 40Gb USB HD) - On the other hand, arbitrarily high numerical
objectives are usually impossible or very
impractical to achieve.
19Direction of Goodness - 2
- The Less the Better
- The implied target is nominally zero.
- Examples service quality (number of defects per
thousands of transactions) process simplicity
(number of steps) startup speed (time to launch
a software application) - Sometimes in a case where the target is minus
infinity, the measurement can be shifted so that
zero become the lowest possible value. - Example refrigeration (absolute zero Celsius
273o)
20Direction of Goodness - 3
- Target Is Best
- The target is as close as possible to a nominal
value with no variation around that value. - Examples exactness of fit (diameter of a steel
rod) constancy of ideal temperature of a food
freezer container is 4oF (-16oC). - The majority of top-level performance
measurements in service applications will be one
of the More the Better and Less the Better
types.
21Define Measurements
- In this step, describe how each measurement will
be performed. - Also, document all assumptions and comments about
each measurement. - The omission of this step leads to much time lost
during planning and later on during development. - This step operationalizes the definition of the
measurement. - Measures without measurement cause confusion,
because one person will inevitably have in mind a
different measurement process than another. - Examples how to measure the startup time of a
software (existing application, operating system,
and RAM configuration assumptions may be
different) - The detailed description of the measurement may
guide developers about what to optimize. It also
makes all discussion involving the measurement
clearer and more efficient.
BACK
22Product Functions - 1
- It can be appropriate to use (product or process)
functions (instead of performance measures) under
the following circumstances - The product or service concept has already been
established. - Especially, breakthrough ideas, at least at the
strategic level, are not needed or are already
incorporated into the concept. - In such a case there may be a list of possible
extensions already expressed as features(??)
that need to be prioritized.
23Product Functions - 2
- The development team lacks either the time or the
interest to develop and prioritize performance
measures. - Since prioritization of performance measures does
not define a products or services features, the
QFD process must be used at least once more to
translate prioritized performance measures into
prioritized features. This extra step is
time-consuming and may not always be worth the
effort. - Some development teams, especially software
development teams, are unaccustomed to using
performance measures in their product definition
process. - While translating customer needs directly into
functions lowers the chances for breakthrough
ideas, teams that dont normally use performance
measures may be better doing just such a
translation.
24Product Functions - 3
- Many products and services have large numbers of
capabilities or functions. - Depending on the level of detail the developers
use to describe the functions, the OQ could be
correspondingly small or large. - The developers can use the Affinity Diagram
process to decide what level of functional
details they want to work at. - The affinity diagram hierarchy of product or
process function will have several levels. - Analyzing at the higher levels will present the
advantage of quicker analysis and disadvantage of
less depth in the analysis. - The HOQ at the strategic analysis level (i.e.,
higher level) will indicate which few critical
functional areas require more detailed planning.
These areas can then be singled out, and the
development team can analyze only those areas in
a subsequent HOQ.
25Product Functions Function Tree
- As is true of all Affinity Diagrams, functions
are organized from the bottom up. - Another approach, the Function Tree method (by
Don Clausing), uses the Tree Diagram method and
organizes the functions from the top down. - In this approach, the primary functions of the
product or service are identified first. Each
primary function is then broken down into
subfunctions. Each subfunction is elaborated
into finer detail until the development team has
reached the level of detail it needs. - This top-down approach creates a function tree
that helps the development team to focus on the
most important functions of their product or
service. - A Function Tree may be indistinguishable from an
Affinity Diagram of functions, once it has been
completed. However, the method creating it and
the associated points of view are different. - To get the best of both methods, the team may
consider first brainstorming functions and
affinitizing them, then completing the structure
with the Tree Diagram method by working from the
top down.
26Product Subsystems
- Most commonly the QFD process translates from
- the Voice of the Customer, to
- Performance Measures, to
- Functions, to
- Product design.
- Each successive pairs of topics (1 to 2, 2 to 3,
3 to 4) represents the left and top,
respectively, of a new matrix. - While the design elements of a product are not
normally chosen to be the Substitute Quality
Characteristics, there are occasions where the
choice is appropriate. - Development team have also created Voice of the
Customer to Product Design Matrices (1 to 4). - Such a matrix shows the development team how
various components of the product design
influence various customer satisfaction
attributes.
27Product Subsystems Tree Diagram
- The most common method for representing the
design is by use of the Tree Diagram. - First, identify the primary subsystems of the
product. - Each of these subsystems and components can then
be described at several levels of additional
details, thus providing the development team with
the multiple-level Tree Diagram it needs for QFD
analysis. - Examples camera gt imaging, film management,
viewfinding, and exposure time management
subsystems gt lens, film plane, and lightproof
compartment between lens and film (components)
(for imaging subsystem)
28Process Steps - 1
- For teams developing new processes or services,
the following choices for SQC are as applicable
as for products performance measures and process
steps. - Performance measures for services are much the
same as for physical product. - Their values are generally under the control of
the team that designs or lays out the processes
underlying the service. - Each process will have one or more clearly
defined inputs and outputs. Development teams
can define performance measures of these
processes, based on time, cost, or quality of
result (i.e., inputs or outputs). - In QFD, the service development team can evaluate
different process performance measures to
determine which ones drive customer satisfaction
performance most strongly.
29Process Steps - 2
- Services are delivered by processes. These
processes may be conceived of at various levels
of abstraction (just like a product). - While product subsystem analysis may not apply to
services, a closely related method of analyzing
services and processes (i.e., process steps) does
apply. - Example telephone customer support
- Anyone who has analyzed the telephone customer
support will know that beneath the simple
(high-level view) process lies an enormously
complex structure for handling customer requests. - This structure includes many decision points and
subprocesses at each of the three main steps. It
also includes tools and technology to support
these decision points and steps.
30Process Steps vs. Performance Measures
Performance measures Process steps
Advantages Advantages
Generally solution-free, providing stronger likelihood of creative solutions. Measurements can be used to manage the processes. Concrete, easily envisioned. Can be used in the HOQ at the level of detail appropriate to the teams needs.
Disadvantages Disadvantages
Difficult to understand or to create in organizations where measurement is not the norm. (i.e., service) Expensive to implement (cost-benefit issues aside). Incomplete definitions of process steps can lead to confusion during QFD. Focus on concrete process steps too early reduces the chances for breakthrough solutions.
31Telephone Customer Support - 1
High-level view of the Process
Route incoming calls to Customer Service Associate
Classify customer request
Respond to customer request
32Telephone Customer Support - 2
33Relationships
34Types of Impacts - 1
- For most QFD activities, the linkage is
considered to be positive, that is, if the SQC is
moved in the direction of goodness, customer
satisfaction is assumed to increased. - Customer satisfaction performance with respect to
the need is not linked to the SQC. - For changes of any sort (, large or small, in the
amount or degree) of the SQC, no noticeable
change in the customer satisfaction performance
of that need is predicted (by the development
team). - Customer satisfaction performance with respect to
the need is possibly linked to the SQC. - For relatively large changes in the amount of the
SQC, little or no change in customer satisfaction
performance of that need is predicted.
35Types of Impacts - 2
- Customer satisfaction performance with respect to
the need is moderately linked to the SQC. - For relatively large changes in the amount of the
SQC, noticeable but not major changes in customer
satisfaction performance of that need are
predicted. - Customer satisfaction performance with respect to
the need is strongly linked to the SQC. - For relatively small changes in the amount of the
SQC, significant changes in customer satisfaction
performance of that need are predicted.
36Types of Impacts - 3
Symbol Meaning Most common numerical value Other values
(or, ?, b) Not linked 0
? Possibly linked 1
? Moderately linked 3
? Strongly linked 9 10, 7, 5
37Judging Impact of Performance Measures
- The Performance Measure is the ideal SQC for
making impact judgment. - The impact can be represented as the mathematical
relationship between the two variables - Customer Satisfaction A f (Performance Measure
X) - The Performance Measure can be thought of as a
independent continuous variable, whereas customer
satisfaction performance of any need is a
dependent continuous variable. - This assumes a monotonic relationship between the
two variables. - It often, also assumes that the relationship is
linear. - The disproportionate, nonlinear behavior (SQC)
type cannot be easily modeled in QFD. The way to
handle it is to treat such SQCs as if they were
linear (until setting Target Values in section F).
38Monotonic and Linear Relationship
- Monotonic As the performance Measure moves in
the direction of goodness, customer satisfaction
performance continues to improve. - Its a good idea to try to define Performance
Measures that do provide a monotonic relationship
with customer satisfaction performance. - Linear As the Performance Measure moves in the
direction of goodness, customer satisfaction
performance continues to improve as the same
rate. - With Kano model, the relationships for Delighters
and Dissatisfiers are not linear.
39Judging Impact of Other SQCs
- It is generally more difficult for development
teams to judge impacts when the SQCs are not
measurable. - Typically development teams think of
nonmeasurable SQCs as either present or
absent. - That is, try to judge whether customer
satisfaction performance will be high if the SCQ
is present, and low if the SQC is absent. - In fact, it is generally unrealistic and
counterproductive to think of product features or
service elements in this binary fashion. - Example SQC (File OPEN command) vs. Need (Can
mix material from many documents) gt refined SQCs
(stripped down OPEN command deluxe OPEN command)
40Impact Values
- Over time, QFD facilitators felt the need to
create a stronger contrast between strong and
the other impacts, so that strong impacts would
have more influence on the final prioritizations. - Some early QFD applications in the U.S. used 5,
3, 1, 0 for the impacts from strong to none. - The greater the ratio between the values assigned
to strong and moderate, the less likely it is
that an SQC with only moderates assigned to it
will have a technical importance greater than an
SQC with at least one high. (i.e., 9, 3, 1, 0) - There is no specific basis for any of the
choices. - The impact values simply provide a way for the
development team to express its judgment on the
relative impacts of SQCs on customer needs. - The SQCs can then be differentiated in terms of
their overall contribution to customer
satisfaction performance.
41Priorities (Contributions)
42Contribution Calculation
Technical Response X Technical Response Y Raw Weight Normalized Raw Weight
Need A 15 .43
Need B 20 .57
Contributions 5.6 0.4 35
Normalized Contributions .93 .07 6.0
?
3.9
0.4
?
?
?
1.7
0
?
?
Raw Weight Total
Contributions Total
?