Managing Six Sigma
Managing Six Sigma
A Wiley-Interscience Publication
JOHN WILEY & SONS, INC.
New York " Chichester " Weinheim " Brisbane " Singapore " Toronto
Opinions, interpretations, conclusions, and recommendations are those of the authors and are not
necessarily endorsed by the United States Air Force. (J .M .C.)
Copyright © 2001 by John Wiley & Sons, Inc . All rights reserved.
This publication is designed to provide accurate and authoritative information in regard to the
subject matter covered . It is sold with the understanding that the publisher is not engaged in
rendering professional services . If professional advice or other expert assistance is required, the
services of a competent professional person should be sought.
10 9 8 7 6 5 4 3 2 1
To my parents. Thank you for all your support. F.B.
To my family, Prem, and especially Hans. Thank you for your patience,
love, and support . B.M.
CONTENTS
6 Performance Measurement 93
6.1 Measurement Types, 93
6.2 Principles of Measurement, 94
6.3 The Balanced Scorecard, 97
In recent years there has been much interest in the application of Six Sigma
techniques. Within organizations chief executive officers (CEO) are hearing
about the monetary rewards that other firms have achieved through Six Sigma
methodology and are eager to cash in on similar benefits . The book Implement-
ing Six Sigma: Smarter Solutions using Statistical Methods (Wiley, 1999), by
Forrest W. Breyfogle III, was written as a practical guide to help organizations
in both industry and academia with the wise implementation and orchestration
of Six Sigma techniques.
Implementing Six Sigma has been a great success ; however, many people
have expressed the desire for a book that will help management decide if they
should implement Six Sigma and then guide them through the process . To ful-
fill this need, we wrote Managing Six Sigma.
The purpose of this book is to build awareness of the wise application of Six
Sigma tools and how they can be important in the "big picture." Because of this
focus, there may be references to tools with which the reader is not familiar. To
help the reader gain a basic understanding of unfamiliar terms, we provide in
the glossary a brief description of most Six Sigma tools discussed in the text.
Implementing Six Sigma can be consulted for more details about individual
tools or applications (see the outline of that book below).
Six Sigma can improve the bottom line of an organization if implemented
wisely. An organization can get more with less using Six Sigma; for example, it
can use fewer runs and samples and obtain more information . However, if the
techniques are not used wisely, there is a considerable danger that the program
will be counterproductive and frustrating . Organizations can sometimes get so
xii PREFACE
involved in how to count and report defects that they lose sight of the real value
of Six Sigma-orchestrating process improvement and reengineering in such a
way that they achieve significant bottom-line benefits through the implementa-
tion of statistical techniques . Six Sigma efforts need to be orchestrated toward
achieving Smarter Solutions (Smarter Solutions, Smarter Six Sigma Solutions,
and S4 are service marks belonging to Forrest W. Breyfogle III).
If an organization does not apply Six Sigma techniques wisely, the methodol-
ogy will fail. When this occurs, there is a tendency to believe thatthe statistical
techniques are not useful, when in fact the real problem is how Six Sigma as a
program was implemented orhow individualtechniques were effectively applied.
This book uses the term "Smarter Six Sigma Solutions (S4)" to describe our
implementation strategy for Six Sigma. Another description for this S4 activity is
"$marter Six Sigma $olutions" assessments, since a major focus is determining
that the right measurements and actions are being taken relative to bottom-line
benefits . With S4 activities an environment is created where there is knowledge-
centered activity (KCA) focus. KCA describes efforts for wisely obtaining
knowledge and/or utilizing the knowledge of organizations and processes .
GE and othercompanies have usedthe terms "Black Belts" and "Green Belts"
(Cheek, 1992; GE, 1996; GE, 1997; Lowe, 1998 ; Slater, 1999; Pyzdek, 1999;
Harry, 2000) to describe people who actively apply Six Sigma techniques These
people may be assigned to this role either full-time or part-time. in this book we
will use the terms "Black Belt" and "Green Belt" along with the term "Six Sigma
Practitioner" to describe people who implement Six Sigma techniques.
This book refers the reader to Forrest W. Breyfogle III's Implementing Six
Sigma : Smarter Solutions Using Statistical Methods, for more descriptions of
tools, implementation techniques, and strategies . References to the earlier work
appear as Breyfogle (1999), Implementing Six Sigma, which contains the fol-
lowing sections :
Implementing Six Sigma contains more than 100 examples within the fol-
lowing chapters:
To meet the needs ofa diverse readership, we adhered to the following guide-
lines in the writing of this book:
Most chapters and sections are small, descriptive, and contain many illus-
trations . The table of contents can be very useful to quickly locate tech-
niques and examples helpful in solving a particular problem .
Equations and formulas are presented only when we believe they are ab-
solute necessary to describe a methodology.
The symbols and glossary sections are intended as a handy reference that
provides fuller explanations whenever a concise definition or an unfamil-
iar statistical term or symbol is encountered in the text.
The focus of the book is on manufacturing, development, and service/
transactional examples that serve as a bridge between Six Sigma tech-
niques and a variety of real-world situations.
The details ofimplementing Six Sigma are not included in this book, since
these techniques are described in Breyfogle (1999), Implementing Six
Sigma.
Even though Breyfogle (1999), Implementing Six Sigma, includes spe-
cific information for those undertaking the details of implementing Six
Sigma, managers should find the S4 assessments sections very helpful:
These sections, found at the end of most chapters, provide guidance inthe
effective orchestration of Six Sigma implementation .
PREFACE xv
ACKNOWLEDGMENTS
The authors are appreciative of those who helped us define major topics for this
book and develop a structure in which to present them. They were Paul Iglesias,
Mike Kirchoff, John W. Knickel, Alejandra Ajuria Martin, Terry J . O'Connell,
and Roland Schimpke . The authors are also appreciative of Titanium Metals
Corporation (TIMET) for letting us use information from one of their projects
and Mark Wallace for his input to an early manuscript.
The terms key process output variable (KPOV), key process input variable
(KPIV), and the term used to describe a simplified quality function deployment
(QFD) matrix-cause-and-effect matrix were coined by Sigma Breakthrough
Technologies, Inc. (SBTI).
do not necessarily need to use all themeasurements listed (often presented within
typical Six Sigma programs) . It is most important to choose the best set of
metrics for a situation, metrics that yield insight into a situation or process .
Our S4 approach advocates the development of cross-functional teams to
provide a holistic approach to problem solving, encompassing all levelsofcom-
plex processes . We often describe the methodology as a murder mystery, where
practitioners are determining "who done it?" or, equivalently, "what is the ma-
jor cause of defects in a process?" By following a structured methodology,
project teams can determine the "biggest hitters" and make substantial areas for
improvement that provide real benefits to an organizations bottom line.
Subsequent chapters of this book will provide the details of effective Six
Sigma metrics and the importance of implementing Six Sigma as a business
strategy. In this chapter, we first discuss current myths surrounding Six Sigma.
We then provide a brief history of quality leaders and other quality systems that
preceded the creation of Six Sigma. Last, we answer a few of the more fre-
quently asked questions (FAQs) about Six Sigma.
Some people view Six Sigma quality as merely a rigorous application of basic
and advanced statistical tools throughout an organization. There are a number
of Six Sigma consultants and training organizations that have simply repack-
aged the statistical components of their previous TQM programs and renamed
them "Six Sigma." These groups would define Six Sigma quality in terms like
those in the upper half of Figure 1.1.
Others view Six Sigma as merely a sophisticated version of Total Quality
Management (TQM), .as represented by the lower half of Figure 1.1. They see it
as an advanced form of TQM in which various continuous improvement sys-
tems must be put in place with a small amount of statistical analyses added in
for good measure.
The S4 view of Six Sigma emphasizes an intelligent blending of the wisdom
of the organization with proven statistical tools to improve both the efficiency
and effectiveness of the organization in meeting customer needs . The ultimate
goal is not improvement for improvement's sake, but rather the creation of
economic wealth for the customer and provider alike. Our Smarter Solutions
approach recommends thatSix Sigma be viewed as a strategic business initia-
tive rather than a quality program . This implies, not that Six Sigma replaces
existing and ongoing quality initiatives in an organization, but that senior man-
agement focuses on those processes identified as critical-to-quality in the eyes
of customers. Those critical systems are then the subject of intense scrutiny and
improvement efforts, using the most powerful soft and hard skills the organiza-
tion can bring to bear. The success of each and every Six Sigma initiative is
6 HOW SIX SIGMA COMPARES TO OTHER QUALITY INITIATIVES
Is an add-on effort
This is simply the myth "creates a parallel organization" in disguise . Same ques-
tion, same response.
Creates bureaucracy
A dictionary definition of bureaucracy is "rigid adherence to administrative
routine ." The only thing rigid about wisely applied Six Sigma methodology is
its relentless insistence that customer needs be addressed.
8 HOW SIX SIGMA COMPARES TO OTHER QUALITY INITIATIVES
Is not cost-effective
If Six Sigma is implemented wisely, organizations can obtain a very high rate
of return on their investment within the first year.
Numerous individuals have had a significant impact on the Six Sigma quality
movement in the United States and abroad. Unfortunately, we don't have time
or space to acknowledge them all . The six we have chosen to give a diverse
perspective are W. Edwards Deming, William Conway, Joseph Juran, Philip
Crosby, Genichi Taguchi, and Shigeo Shingo . This brief review is more than a
historical perspective . At the conclusion of this section, we will discuss how
Six Sigma relates to the previous generation of quality programs .
W. Edwards Deming
Dr. W. Edwards Deming was born on October 14, 1900 . By 1928, hehad earned
a B.S. in engineering at the University of Wyoming in Laramie, an M.S. in
mathematics and physics at the University of Colorado, and a Ph.D. in math-
ematical physics at Yale. He is best known for revitalizing Japanese industry
after the end of World War II. In 1950, he visited Japan at the request of the
Secretary of War to conduct a population census. While there he was invited by
a member ofthe Japanese Union of Science and Engineering (JUSE) to lecture
on statistical methods for business at ;a session sponsored by the Keidanren, a
prestigious society of Japanese executives . Deming told the Japanese leaders
THE QUALITY LEADERS 9
that they could "take over the world" . if they followed his teachings. The rest is
history. Today the the most prestigious quality award in Japan is called the
"Deming Prize." Deming, who has been heralded as the "founder of the Third
Wave of the Industrial Revolution," oftensounded like a quality crusader when
he issued statements such as "It is time to adopt a new religion in America." He
gained national recognition in 1980 when he was interviewed for the NBC
television ".white paper," "If Japan Can, Why Can't We?"
Dr. Deming is equally well known for his "Fourteen Points" (Table 1 .2) and
"Seven Deadly Diseases" (Table 1 .3) (Deming,1986). He made numerous modi-
fications to the first list throughout his life as he grew in knowledge and wis-
dom. Additional references that provide commentary on Deming's Fourteen
Points are provided in the references section of this book (Breyfogle, 1999 ;
Neave, 1990; Scherkenbach, 1988 ; Gitlow and Gitlow, 1987).
12a. Remove barriers that rob the hourly worker(s) of their right to pride of workman-
ship . The responsibility of supervisors must be changed from sheer numbers to
quality.
12b. Remove barriers that rob people in management and in engineering of their right
to pride of workmanship. This means, inter alia, abolishment of the annual or
merit rating and of managing by objective.
13 . Institute a vigorous program of education and self-improvement.
14 . Put everybody in the company to work to accomplish the transformation . The
transformation is everybody's job.
Lack of constancy of purpose to plan product and service that will have a market
and keep the company in business, and providejobs.
2. Emphasis on short-term profits : short-term thinking (just the opposite from
constancy of purpose to stay in business), fed by fear of unfriendly takeover, and
by push from bankers and owners for dividends.
3. Evaluation of performance, merit rating, or annual review.
4. Mobility of management : job hopping.
5. Management by use only of visible figures, with little or no consideration of
figures that are unknown or unknowable .
6. Excessive medical costs.
7. Excessive costs of liability, swelled by lawyers that work on contingency fees .
William E. Conway
Joseph M. Juran
Joseph Moses Juran (Juran, 1999a) was born December 24, 1904, in Braila,
Romania . In 1920, Juran enrolled at the University of Minnesota. In 1924, he
graduated with a B.S. degree in electrical engineering and took ajob withWest-
ern Electric in the Inspection Department of the Hawthorne Works in Chicago.
In 1926, a team from Bell Laboratories (including Walter Shewhart and Harold
Dodge) made a visit to the factory with the intention of applying the laboratory
12 HOW SIX SIGMA COMPARES TO OTHER QUALITY INITIATIVES
tools and methods they had been developing. Juran was selected as one of 20
trainees, and he was subsequently chosen as one of two engineers to work in the
newly created Inspection Statistical Department . In 1928, Juran authored his
first work on quality, a pamphlet that became an input to the well-known AT&T
Statistical Quality Control Handbook, which is still published today.
In 1937, Juran conceptualized the Pareto principle . In December 1941, he
took a leave of absence from Western Electric to serve in Washington with the
Lend-Lease Administration. It was here that he first experimented with what
today might be called "business process reengineering ." His team successfully
eliminated the paper logjam that kept critical shipments stalled on the docks. In
1945, Juran left Washington and Western Electric to work as an independent
consultant. In 1951, his standard reference work on quality control, Quality
Control Handbook (Juran, 1999b), was first published .
In 1954, the Union of Japanese Scientists and Engineers (JUSE) and the
Keidanren invited the celebrated author to Japan to deliver a series of lectures.
These talks addressed managing for quality, and were delivered soon after Dr.
W. Edwards Deming had delivered his lectures on statistical quality methods .
Dr. Juran has been called the father of quality, a quality guru, and the man
who "taught quality to the Japanese ." He is recognized as the person who added
the "human dimension" to quality, expanding it beyond its historical statistical
origins to what we now call total quality management. Says Peter Drucker,
"Whatever advances American manufacturing has made in the last thirty to
forty years, we owe to Joe Juran and to his . . . work." A few of Juran's impor-
tant quality books are listed in the reference section of this book (Juran, 1964,
1988, 1989, 1992).
Philip B. Crosby
Phil Crosby (Crosby, 1999) was born in Wheeling, West Virginia, on June 18,
1926. He started work as a quality professional on an assembly line at Crosley
in 1952 after serving in WorldWar II and Korea. He took it upon himself to try
to convince management that it was more profitable to prevent problems than
to fix them. He worked for Crosley from 1952 to 1955; Martin-Marietta from
1957 to 1965 ; and ITT from 1965 to 1979. As quality manager for Martin-
Marietta he created the Zefo Defects concept and program . During his 14 years
as corporate vice president for ITT, he worked with many industrial and service
companies around the world, implementing his pragmatic philosophy.
In 1979, Crosby founded Philip Crosby Associates, Inca (PCA), and over the
next 10 years grew it into a publicly traded organization with 304 employees
around the world . In 1991, he retired from PCA and founded Career IV, Inc ., a
company that provided lectures and seminars aimed at helping current and pro-
THE QUALITY LEADERS 13
spective executives grow. In 1997 he purchased the assets of PCA and estab-
lished Philip Crosby Associates II, Inc. Now his "Quality College" operates in
more than 20 countries around the world.
Philip Crosby's lectures provide a thoughtful and stimulating discussion of
managements' role in causing their organizations, their employees, their sup-
pliers, and themselves to be successful. He has published 13 books, all of which
have been best-sellers. His first business book, Quality Is Free (Crosby, 1979),
has been credited with beginning the quality revolution in the United States and
Europe . A few of his other titles are included in the references section of this
book (Crosby, 1984, 1992) .
Philip Crosby's books are easy to read and explain his program for quality
improvement and defect prevention. He emphasizes the importance of
management's role in making quality happen . His approaches team basedbut is
not highly statistical, as are some of the competing approaches to TQM. He
does teach the use of the simple tools and statistical quality control, but in a
very generic, non-mathematical sense . He is an interesting and knowledgeable
speaker who both entertains and educates his audiences . He has certainly had
an impact on the quality revolution over the years.
Shigeo Shingo
Shigeo Shingo was born in Saga City, Japan, in 1909. In 1930, he graduated
with a degree in mechanical engineering from Yamanashi Technical College
and went to work for the Taipei Railway Factory in Taiwan. Subsequently he
became a professional management consultant in 1945 with the Japan Manage-
ment Association . It was in his role as head of the Education Department in
1951, that he first heard of, and applied, statistical quality control. By 1954 he
had studied 300 companies . In 1955, he took charge of industrial engineering
and factory improvement training at the Toyota Motor Company for both its
employees and parts suppliers (100 companies). During the period 1956-1958,
at Mitsubishi Heavy Industries in Nagasaki, Shigeo was responsible for reduc-
ing the time for hull assembly of 65,000-ton supertankers from four months to
two months . In 1959, he left the Japan Management Association and estab-
lished the Institute of Management Improvement, with himself as president. In
1962, he started industrial engineering and plant-improvement training at
Matsushita Electrical Industrial Company, where he trained some 7,000 people .
Shingo's supreme contribution in the area ofquality was his development in
the 1960s of poka-yoke (pronounced "POH-kah YOH-kay") . The term comes
from the Japanese words "poka" (inadvertent mistake) and "yoke" (prevent) .
The essential idea of poka-yoke is to design processes so mistakes are impos-
sible to make or at least easily detected and corrected.
Poka-yoke devices fall into two major categories: prevention and detection.
A prevention device affects the process in such a way that it is impossible to
make a mistake . A detection device signals the user when a mistake has been
made, so that the user can quickly correct the problem . Shingo's first poka-
yoke device was a detection implement he created after visiting Yamada Elec-
tric in 1961. He was told of a problem in the factory that occasionally led to
workers' assembling a small, two-push-button switch without inserting a spring
under each push-button. The problem of the missing spring was both costly and
embarrassing, but despite everyone's besteffort the problem continued. Shingo's
solution was simple. The new procedure completely eliminated the problem:
" Old method : a worker began by taking two springs out ofa large parts box
and then assembled a switch.
" New approach: a small dish was placed in front of the parts ox and the
worker's first task was to take two springs out of the box and place them
on the dish. Then the worker assembled the switch. If any spring remained
on the dish, then the worker knew that he or she had forgotten to insert it.
Although Shigeo Shingo is perhaps less well known in the West, his impact
on Japanese industry has been immense. Norman Bodek, president of Produc-
tivity, Inc., stated : "If I could give a Nobel Prize for exceptional contributions
TQM AND SIX SIGMA 15
The six quality professionals we discussed have a lot in common with respect
to the intellectual content of their quality approaches. They have chosen to em-
phasize different aspects of the body of knowledge known as quality. Deming
and Taguchi emphasize the statistical aspects of quality ; Deming, Crosby, and
Juran focus on management's role in the improvement process ; Conway relent-
lessly drives home the need to eliminate waste wherever it can be found; Shingo
preaches error-proofing. There is no doubt that these individuals had a marked
impact on how we thought about qualityduring the second halfofthe twentieth
century. But does their expertise transfer to bottom-line benefits for organiza-
tions? In a survey of chief executive officers who subscribe to Electronic Busi-
ness, only16% reported improved market share and only 13% reported improved
operating income or profits as a result of their quality efforts (Boyett et al.,
1992) . Mikel J. Harry (2000b) reports that even as increasing numbers of com-
panies jumped on the quality "express" during this time, fewer were reporting
any meaningful impact on profitability.
A study published by the American Society for Quality (Bergquist and
Ramsing, 1999) compares the performance of winners and applicants of both
the Malcolm Baldrige National Quality Award (MBNQA) and state-level award
programs with nonapplicants in the same two categories. The authors are un-
able to "conclusively determine whether quality awardwinning companies per-
form better than others [i.e., nonapplicants] ."
One ofthe problems that plagued many ofthe early TQM initiatives was the
preeminence placed on quality at the expense of all other aspects of the busi-
ness. Some organizations went bankrupt, or experienced severe financial con-
sequences, in the rush to make quality "first among equals." Harry (2000a)
notes that "what's good for the customer is not always good for the provider."
The disconnect between management systems designed to measure customer
satisfaction and those designed to measure provider profitability often led to
unwise investments in quality .
We believe that Six Sigma is more than a simple repackaging of the best
from other TQM programs . Tom Pyzdek (1999c) agrees: "Six Sigma is such a
drastic extension ofthe old idea of statistical control as to be an entirely differ-
ent subject. . . . In short, Six Sigma is . . . an entirely new way to manage an
organization . . . . Six Sigma is not primarily atechnical program; it's a manage-
ment program."
16 HOW SIX SIGMA COMPARES TO OTHER QUALITY INITIATIVES
Ronald Snee (1999x) points out that although some people believe it is noth-
ing new, Six Sigma is unique in its approach and deployment. He defines ix
Sigma as a strategic business improvement approach that seeks to increase both
customer satisfaction and an organization's financial health. Snee goes on to
claim that eight characteristics account for Six Sigma's increasing bottom-line
success and popularity with executives :
Bottom-line results expected and delivered
Senior management leadership
A disciplined approach (Measure, Analyze, Improve, and Control)
Rapid (3-6 month) project completion
Clearly defined measures of success
Infrastructure roles for Six Sigma practitioners and leadership
Focus on customers and processes
A sound statistical approach to improvement
Other quality initiatives have laid claim to one or two of these characteris-
tics, but only Six Sigma attributes its success to the simultaneous application of
all eight.
Mikel Harry (2000b) goes even farther and claims that Six Sigma represents
a new, holistic, multidimensional systems approach to quality that replaces the
"form, fit and function specifications" of the past. In his New world view of
quality, both the customer and provider receive benefits of utility, availability,
and worth:
Utility
" Form (is in a pleasing/effective physical form)
Fit (made to the correct size/dimensions)
" Function (performs all required tasks well)
" Availability
" Volume (production capacity, inventory levels, and/or distribution chan-
nels are adequate to meet varying demand)
" Timing (a short product development cycle time;lot size =1 ; and/or JIT
availability)
" worth
" Intellectual (innovative; sought by early adopters)
" Emotional (satisfies one's pride, passions, or psyche)
Economic (customer perceived value [quality-to-price ratio] is high)
We believe Six Sigma encompasses all of these things and more. As Figure
1.1 depicts, the S4approach encourages the following :
SIX SIGMA AND INTEGRATION OF THE STATISTICAL TOOLS 17
The Smarter Six Sigma Solutions (S4) approach involves identifying projects
that :target the customer's concerns and have the potential for significant pay-
back . This is accomplished using a team-based approach (demanding soft skills)
coupled with a wide range of statistical tools (demanding hard skills) . Zinkgraf
and Snee (1999) have drawn up a list of the eight tools they believe are essential
to getting the job done (Table 1 .4) . These tools represent a subset of a much
larger array of statistical and management tools at one's disposal . Breyfogle
(1999), in Implementing Six Sigma, illustrates a 21-step integration of the tools,
which is shown in Table 1 .5 and further elaborated upon in later chapters.
Process Maps
Cause-and-Effects Matrix
Measurement Systems Analysis (Gage R&R)
Process Capability Studies
Failure Modes and Effects Analysis (FMEA)
Multi-Vari Studies
Design of Experiments (DOE)
Control Plans
18 HOW SIX SIGMA COMPARES TO OTHER QUALITY INITIATIVES
TABLE 1.5 The 21-Step Integration of the Tools [From Breyfogle (1999), with
permission]
Source of
Step Action Participants Information
and other project parameters can be quantified. Later in this book, we describe
how this chart can be used to get organizations out of the "firefighting mode."
The simple tools of Pareto charts, flowcharts, process maps, and so forth also
come in handy at the beginning of a project as ways of displaying and analyz-
ing data. It is important to realize that the assigned Six Sigma practitioners are
involved in every step of the process . They are an integral part of the project
team. but are identified individually in the first three steps, since prior to step 4
the team itself does not yet exist . There is important work to be done before
team members are identified' and trained.
Most of the information in this section was reproduced (with minor additions,
deletions, and modifications) from an article by Denecke (1998), and is repro-
duced with the permission of Honeywell International Inc . (formerly
AlliedSignal) . We decided to share the article with readers of this book. We
have paraphrased it extensively, with permission, and included our own re-
sponse to the article .
It should be noted that in our opinion the differences between how Six Sigma
and lean manufacturing would address a project are even less than what is sug-
gested by this article. We believe that Six Sigma should have metrics, such as
cycle time, that lead to the application of the discipline of lean manufacturing
when it is mostappropriate . In addition, the techniques of Six Sigma should be
used within processes to reduce defects, which can be a very important prereq-
uisite for a lean manufacturing project to be successful.
Background
Currently there are two premier approaches to improving manufacturing opera-
tions . One is lean manufacturing; the other is Six Sigma. They are promoted as
different approaches and different thought processes . Yet, upon close inspec-
tion, both approaches attack the same enemy and behave like two links within a
chain that is, they are dependent on each other for success . They both battle
variation, but from two different points of view. Lean and Six Sigma integra-
tion takes two powerful problem-solving techniques and bundles them into a
powerful package . The two approaches should be viewed as complements to
each other rather than as equivalents ofor replacements for each other (Pyzdek,
2000):
In theory, lean manufacturing (hereinafter referred to as "lean") is a winning
strategy. In practice, manufacturers thathave widely adopted lean practices record
performance metrics superior to those achieved by plants that have not adopted
lean practices .Those practicescited as lean in a recent industrial survey (Jusko,
22 HOW SIX SIGMA COMPARES TO OTHER QUALITY INITIATIVES
1999) include (1) quick changeovertechniques to reduce setup time; (2) adoption
ofmanufacturing cells in which equipment and workstations are arranged se-
quentially to facilitate small-lot, continuous-flow production; (3) just-in-time
(JIT) continuous-flow production techniques to reduce lot sizes, setup time, and
cycle time; and (4) JIT supplier delivery inwhich parts and materials are delivered
to the shop floor on a frequent, as-needed basis.
Lean evaluates the entire operation of a factory and restructures the manu-
facturing method to reduce wasteful activities like waiting, transportation, ma-
terial hand-offs, inventory, and overproduction . It co-locates the processes in
sequential order and, in so doing, reduces variation associated with manufac-
turing routings, material handling, storage, lack of communication, batch pro-
duction, and so forth. Six Sigma tools, on the other hand, commonly focus on
specific part numbers and processes to reduce variation . The combination of
the two approaches represents a formidable opponent to variation in that it in-
cludes both relayout of the factory and a focus on specific part numbers and
processes . The next section discusses the similarities and differences between
the two approaches.
Lean manufacturing changes the set of problems the organization must ad-
dress and increases the price offailure . The organization is forced to respond to
the lean definition of waste in the process travel distance, wait time, non-
value-added cycle time, setup time, inventory levels, overproduction, and so
on. Setup time is an especially important measure in the lean manufacturing
line. Once the first machine starts setting up for a different part configuration,
the rest of the line becomes starved for parts in other words, setup time be-
comes cumulative. The time involved in setting up the entire cell is the sum of
all the individual setup times. A two-hour setup per machine may result in 20
hours of downtime for the entire cell. A host of problems and opportunities that
the organization could afford to hide in the past now become critical owing to
this cumulative effect. A Six Sigma organization needs to understand the spe-
cial demands of lean manufacturing ifit plans to move toward a quality system
that embraces both approaches.
Synergy
Lean and Six Sigma, working together, represent a formidable. weapon in the.
fight against process variation . Six Sigma methodology uses problem-solving
techniques to determine how systems and processes operate_ and how to reduce
variation in processes . In a system that combines the two philosophies, lean
creates the standard and Six Sigma investigates and resolves any variation from
the standard.
With the lean emphasis on standardization, Six Sigma practitioners can be
10 times more effective at tackling process noise . If a number of traditional
24 HOW SIX SIGMA COMPARES TO OTHER QUALITY INITIATIVES
---------------------------------------------------------------------------------
Activity
Standardization
One-Piece
Flow
Cleanliness
________ ._____________________________ . . _____ ___- _ :____________ :______
SOURCES OF-LEAN-VARIATION
-------------------- ---------------------------------------------------------------------------------------------
Cost of Quality
Scrap & Rework
Poor Yield
Part Number Specifications
Figure 1 .2 Variation as viewed by lean manufacturing and Six Sigma [Adapted from
Denecke (1998) .]
noise factors have been eliminated by the lean activity, the rest are now easily
identifiable through simple observation of the cell. The new system enables
rapid problem solving since the Six Sigma practitioner does not haveto chase
parts through the factory. The hidden factory, now exposed, shows a number of
FREQUENTLY ASKED QUESTIONS ABOUT SIX SIGMA 25
new opportunities for variation for example, cycle time variation, queuing of
parts, downtime of the cell, unpredictable walk patterns, special skill require-
ments, nonstandard operations, quality problems, and so on. The Six Sigma
practitioner can now focus quickly on the problem, assemble the natural work
team, collect data in one contained area, compare the results to the standard,
and begin the improvement project. The Six Sigma practitioner becomes pro-
active, continuously seeking to improve upon the standard by expanding on the
current process knowledge. It is the responsibility of the organization to go lean
so that those implementing Six Sigma are used not as "firefighters" but as pro-
active and productive problem solvers.
Summary
Again, in our opinion the differences between how Six Sigma and lean manu-
facturing define and address a project are even less pronounced than what is
suggested by the article upon which the preceding discussion is based. We be-
lieve that Six Sigma within any organization should have metrics, such as cycle
time, that lead to the application of lean manufacturing when it is appropriate .
In addition, the techniques of Six Sigma should be applied within an
organization's processes to reduce defects, which can be a very important pre-
requisite to the success of a lean manufacturing project.
What Is the Practical Difference Between Three and Four Sigma Quality
Levels?
To a first approximation, a three sigma quality level is equivalent to 1.5 mis-
spelled words per page in a book such as this one . A six sigma quality level is
equivalent to one misspelled word in all of the books in a small library. The
differences are huge. The difference stems from the nonlinear relationship be-
tween these defect rates and sigma quality levels .
Table 1.6 (Texas Instruments, 1992) provides another basis for comparison
among sigma quality levels .
What Kind of Return on Investment (ROI) Can I Expect from Six Sigma
Training?
Average project savings can vary between $50,000 and $200,000. Six Sigma
Practitioners, with 100% of their time allocated to projects, can execute five or six
FREQUENTLY ASKED QUESTIONS ABOUT SIX SIGMA 27
described in more detail later in this book. Note in Table 1 .7 the distinction be-
tween a "supporter" and a "champion" of Six Sigma (Schiemann and Lingle,
1999). Both roles are important, but they differ in several crucial respects.
Champion Supporter
" Recruits new supporters of Six Sigma Maintains support ofthe Six Sigma
process
Anticipates and addresses Six Sigma Reacts to and helps solve Six Sigma
problems before they occur problems that occur
Speaks inspirationally about Six Sigma Speaks supportively of Six Sigma
" Actively seeks opportunities to show Avoids public actions which appear
support of Six Sigma nonsupportive of Six Sigma
Initiates important Six Sigma activities Participates in planned Six Sigma
activities
perience that will allow them to bridge the gap between classroom examples
and the application oftechniques to specific situations such as those that stu-
dents present during training and coaching sessions.
It should be noted that this list does not call for a knowledge of statistical
techniques . Our belief is that trainees with the characteristics listed above will
easily pick up statistical techniques ifprovided with good instructional material
and training.
dominance) lack the abilities and finesse in human relations and team building
(right-brain dominance) . Some important soft skills are the following :
Six Sigma was first espoused by Motorola in the 1980s . Recent Six Sigma
success stories, primarily from the likes ofGeneral Electric, Sony, AlliedSignal,
and Motorola, have captured the attention of Wall Street and have propagated
the use of this business strategy. A Six Sigma initiative is designed to change
the culture in a company through breakthrough improvements by focusing on
out-of-the-box thinking in order to achieve aggressive, stretch goals . The Six
Sigma initiative has typically contributed an average of six figures per project
to the bottom line.
In this chapter we discuss the experiences of other organizations with the
initiative and also discuss a statistical definition of Six Sigma.
In preparing these remarks, I reflected on the beginnings of Six Sigma and the
pioneers who had the courage, intellect, and vision to make it a reality. The father
of Six Sigma was the late Bill Smith, a senior engineer and scientist in our com-
32 SIX SIGMA BACKGROUND AND FUNDAMENTALS
munications products business . It was Bill who crafted the original statistics and
formulas that were the beginnings of the Six Sigma culture. He took his idea and
his passion for it to our CEO at the time, Bob Galvin. Bob could see the strength
of the concept, but reacted mostly in response to Bill's strong belief and passion
for the idea . He urged Bill to go forth and do whatever was needed to be done to
make Six Sigma the number one component in Motorola's culture. Not long af-
terwards, Senior Vice President Jack Germaine was named as quality director and
charged with implementing Six Sigma throughout the corporation. Jack had a big
job and was armed with few resources. So he turned to Motorola University to
spread the Six Sigma word throughout the company and around the world. Soon,
Six Sigma training was required for every employee . The language of quality
became the common Motorola language . Whether, it was French, Arabic,,or Texan,
everyone understood the six steps, defect measurement and elimination, and parts
per million. The training and concepts were not limited to the factory floor. Every
single person was expected to understand the process and apply it to everything
that they did. Some shied away, saying they were exempt from manufacturing
processes. Some challenged the soundness of the statistics . Even so, the company
stood behind the commitment and the mandate.
The result was a culture of quality that permeated throughout Motorola and led to
a period of unprecedented growth and sales. The crowning achievement was be-
ing recognized with the Malcolm Baldrige National Quality Award. At the time,
Bill remarked that Motorola was the only company who could receive the award.
We were, to his knowledge, the only applicant who had the processes, measure-
ments, and documentation in place to tell our quality story.
Motorola launched its "Six Sigma Quality Program" on January 15, 1987 .
The program was kicked off with a speech by Motorola's chief executive of-
ficer, Bob Galvin, that was distributed in the form of both a letter and a video-
tape . The company set a five-year target to achieve Six Sigma. By March 1988,
Motorola University had begun offering a course on implementing Six Sigma
that was aimed primarily at services rather than products . Within a few months
of the initial training, teams were starting improvement projects to reach their
new corporate quality goals.
General Electric (GE) has the unique distinction of being at the head of the list
of Fortune 500 companies based on market capitalization . What does market
capitalization mean? It means that if someone multiplies GE's outstanding shares
of stock by its current market price per share, GE is the highest-valued com-
pany listed on any of the numerous U.S . stock exchanges. This monetary value
exceeds the gross domestic product of many nations in the world.
General Electric CEO Jack Welch describes Six Sigma as "the most chal-
EASTMAN KODAK' S EXPERIENCE WITH SIX SIGMA 33
"Medical Systems described how Six Sigma designs have produced a 10-
'fold increase in the life of CT scanner x-ray tubes increasing the 'up-
time' of these machines and the profitability and level of patient care given
by hospitals and other health care providers ."
"Super-abrasives-our industrial diamond business-described how Six
Sigma quadrupled its return on investment and, by improving yields, is
giving it a full decade's worth ofcapacity despite growing volume-with-
out spending a nickel on plant and equipment capacity."
"Our railcar leasing business described a 62% reduction in turnaround
time at its repair shops: an enormous productivity gain for our railroad
and shipper customers and for a business that's now two or three times
faster than its nearest rival because of Six Sigma improvements . In the
next phase across the entire shop network, black belts and green belts,
working with their teams,: redesigned the overhaul process, resulting in a
50% further reduction in cycle time."
"The plastics business, through rigorous Six Sigma process work, added
300 million pounds of new capacity (equivalent to a `free plant'), saved
$400 million in investment, and will save another $400 million by 2000."
Six Sigma training has permeated GE, and experience with Six Sigmaimple-
mentation is now a prerequisite for promotion to all professional and manage-
rial positions. Executive compensation is determined to a large degree by one's
proven Six Sigma commitmentand success. GE now boasts slightly under 4,000
full-time, trained Black Belts and Master Black Belts. They also claim more
than 60,000 part-time Green Belts who have completed at least one Six Sigma
project (Pyzdek, 1999).
"Eastman KodakCompany has been actively involved in the global quality trans-
formation since the late 1950s (Gabel and Callanan, 1999). In the early 1980s
they introduced their Quality Improvement Facilitator (QIF) Program. In the early
34 SIX SIGMABACKGROUND AND FUNDAMENTALS
1990s they joined a Six Sigma consortium involving Motorola, Texas Instruments,
Digital Equipment Corporation, and IBM. Since that time, they have been ac-
tively pursuing a Six Sigma process performance throughout the company.
"The Office of Six Sigma Programs is part of the Corporate Quality Organization
of Eastman Kodak Company. This office helps define and refine the portfolio of
quality training programs and functions within the company. In 1999, Kodak had
three corporate training and certification programs for building Six Sigma capa-
bilities within the company: the redesigned Quality Improvement Facilitator Pro-
gram, the Black Belt Program, and the Management BlackBelt Program.
"The QIF Program is designed to develop general practitioners in quality that can
support all organizational improvement efforts. Lead operators and l st level su-
pervisors engage in seven weeks of training in the areas of interpersonal skills,
leadership skills, basic quality and statistics, and other technical topics . Project
work is required as part of the certification process. These generalists are then
qualified to coach, teach, and facilitate as active members of any organizational
improvement teams.
"The Black Belt Program is designed to develop quality specialists with advanced
training and skill in statistics . Their role is to lead major breakthrough improve-
ment efforts. Candidates for this program are drawn from the ranks of experi-
enced technical leaders within Kodak . The five weeks of training begin with a
week focused on interpersonal skills, followed by four weeks of training based on
the traditional Six SigmaMAIC model of measure, _analyze, improve and pontrol.
One week is devoted to each of these four MAIC areas. The curriculum provides
extensive coverage in the areas of interpersonal skills, advanced quality and sta-
tistics, and other technical topics. A certification project is also required.
"The Management Black Belt Program is designed to enable 2nd level supervi-
sors up to and including upper management to not only talk-the-talk of Six Sigma,
butalso more importantly to walk-the-walk. Training involves approximately one
week in the classroom covering the general topics of basic quality and statistics .
A certification project is required.
"To date, Kodakhas saved over $100 million on the certification projects associ-
ated with these three programs . The Six Sigma training initiative started at their
Rochester-based operations, but has since been expanded worldwide. The Black
Belts, Management Black Belts, and Quality Improvement facilitators are highly
valued resources within the Kodak community. Most of the original Six Sigma
activity to date has been in the manufacturing area, but has recently expanded
into business and transactional processes as well. The R&D Division is currently
applying Six Sigma philosophy and methods to the organization's internal re-
search and development activity.
"Kodak reports that five factors have been critical to the success of their Six
Sigma training initiative :
" Management support
" Quality of the work environment
ADDITIONAL EXPERIENCES WITH SIX SIGMA 35
A USA Today article presented differences of opinion about the value of Six
Sigma in "Firms Air for Six Sigma Efficiency" (Jones, 1998) . One stated opin-
ion was that Six Sigma is "malarkey," whereas Larry Bossidy, CEO ofAllied
Signal, countered with "The fact is, there is more reality with this [Six Sigma]
than anything that has come down in a long time in business . The more you get
involved with it, the more you're convinced." The following are some other
quotes from the article :
"After four weeks of classes over four months, you'll emerge a Six Sigma
`black belt.' And if you're an average black belt, proponents say you'll
find ways to save $1 million each year."
" Six Sigma is expensive to implement. That's why it has been a large-
company trend. About 30 companies have embraced Six Sigma including
Bombardier, ABB [Asea Brown Boveri] and Lockheed Martin."
"Nobody gets promoted to an executive position at GE without Six Sigma
training. All white-collar professionals must have started training by Janu-
ary. GE says it will mean $10 billion to $15 billion in increased annual
revenue and cost savings by 2000 when Welch retires."
"Raytheon figures it spends 25% of each sales dollar fixing problems when
it operates at four sigma, a lower level of efficiency. But if it raises its
quality and efficiency to Six Sigma, it would reduce spending on fixes to
1%. 11
10 "It will keep the company [AlliedSignal] from having to build an $85
million plant to fill increasing demand for caperolactan used to make ny-
lon, a total savings of $30-$50 million a year."
"Lockheed Martin used to spend an average of 200 work-hours trying to
get a part that covers the landing gear to fit. For years employees had
brainstorming sessions, which resulted in seemingly logical solutions . None
worked . The statistical discipline of Six Sigma discovered a part that de-
viated by one-thousandth of an inch. Now corrected, the company saves
$14,000 ajet."
"Lockheed Martin took a stab at Six Sigma in the early 1990s, but the
attempt so foundered that it now calls its trainees `program managers'
instead of black belts to prevent in-house jokes ofskepticism . . . Six Sigma
36 SIX SIGMA BACKGROUND AND FUNDAMENTALS
is a success this timearound. The company has saved $64 million with its
first 40 projects ."
"John Akers promised to turn IBM around with Six Sigma, but the at-
tempt was quickly abandoned whenAkers was ousted as CEO in 1993."
"Marketing will always use the number that makes the company look
best . . . Promises are made to potential customers around capability statis-
tics that are not anchored in reality."
"Because managers' bonuses are tied to SixSigma savings, itcauses them
to fabricate results and savings turn out to be phantom."
"Six Sigma will eventually go the way of other fads, but probably not
until Welch and Bossidy retire."
As apparent from the vast differences in opinion listed, Six Sigma can be a
great success or failure depending upon how it is implemented .
Recent Six Sigma success stories, primarily from the likes of General Electric,
Sony, AlliedSignal, and Motorola, have captured the attention of Wall Street
and have propagated the use of this business strategy. The Six Sigma strategy
involves the use of statistical tools within a structured methodology for gaining
the knowledge needed to achieve better, faster, and lower-cost products and
services . The repeated, disciplined application of the master strategy on project
after project, where the projects are selected based on key business issues, is
what drives money to the bottom line, resulting in increased profit margins and
impressive return on investment from the Six Sigma training . The Juran Insti-
tute (Godfrey, 1999) emphasizes the point that an organization's vision and
strategy must be the compass directing Six Sigma project selection and mea-
surement. The typical Six Sigma initiative has contributed an average of six
figures per project to the bottom line when properly implemented .
Six Sigma project specialists are sometimes called "Black Belts,"' "top guns,"
"change agents," or "trailblazers," depending on the company deploying the
strategy. These people are trained in the Six Sigma philosophy and methodol-
ogy and are expected to accomplish atleast four projects annually, which should
deliver at least $500,000 annually to the bottom line. A Six Sigma initiative in a
company is designed to change the culture through breakthrough improvements
by focusing on out-of-the-box thinking in order to achieve aggressive, stretch
goals. Ultimately; Six Sigma, if deployed properly, will infuse intellectual capi-
tal into a company and produce unprecedented knowledge gains that translate
directly into bottom-line results (Kiemele, .1998) .
We believe that many organizations will find that Six Sigma implementation
is the best thing that ever happened to them. Unfortunately, we predict that.
THE STATISTICAL DEFINITION OF SIX SIGMA 37
some companies embarking upon the Six Sigma path will find the journey dif-
ficult. :Success depends on how Six Sigma is implemented . The road map in
this book can lead an organization away from a Six Sigma strategy built around
"playing games with the numbers" to a strategy that produces long-lasting pro-
cess improvements yielding significant bottom-line results .
The statistics associated with Six Sigma may initially seem difficult, but they
are relatively simple to grasp and apply, if a little effort is given to understand-
ing them. The reader is referred to the glossary at the end of this book for a
definition of any new terms. To define Six Sigma statistically, we will work
with two concepts, specification limits and the normal distribution.
Specification Limits'
Specification limits are the tolerances or performance ranges that customers
demand of the products or processes they are purchasing. An example of a
specification might be the size of a given circular hole drilled into a circuit
board in a manufacturing plant. For purposes of illustration, we will refer to
this hole as "XYZ."Acircuit board may have hundreds ofholes thathave speci-
fied diameters . Ifhole XYZ is too large, it will tend to accumulate excess solder
that could lead to other problems . If it is too small, the "metal lead" that gets
inserted into the hole and subsequently soldered into place may be too big to fit
properly. The customer desires an XYZ hole diameter of exactly xyz inches,
but will accept XYZ holes with diameters that fall within the range between the
customer-specified lower specification limit [LSL] and upper specification limit
[USL] . Why? Because variability is so ubiquitous in the real world that we
have to allow specification limits that permit some degree ofimprecision in our
ability to drill: xyz-inch-diameter holes.
Figure 2.1 illustrates specification limits as the two major vertical lines in
the figure . The target value, although not shown in the figure, is typically at the
exact center between the USL and LSL . These specification limits are totally
independent of the bell-shaped curve of the normal distribution, also shown in
the figure . The important thing to realize about the range between the USL and
the LSL is that the customer expects every XYZ hole drilled in their circuit
board to have a diameter that falls somewhere within that range and, hopefully,
falls exactly in the center of that range . Is it acceptable for an occasional XYZ
hole to have a diameter equal to one of the specification limits? It depends on
the customer, who will decide whether or not the extreme values at the specifi-
cation limits are acceptable quality levels .
38 SIX SIGMA BACKGROUND AND FUNDAMENTALS
Figure 2 .1 With a central normal distribution between Six Sigma limits, only two
devices per billion fail to meet the specification target. (Copyright of Motorola, used
with permission.)
The bell-shaped curve in Figure 2 .1 is called the normal distribution, also known
as a Gaussian curve . It has a number of properties that make it an extremely useful
and valuable tool in both the statistical and quality worlds . The curve is symmetri-
cally shaped and extends from + to -- infinity on the X-axis . This normal curve
is totally independent of the LSL and USL described above for hole XYZ . The
shape of this normal curve depends solely on the process, equipment, personnel,
and so on, which can affect the drilling of XYZ holes . This normal curve repre-
sents the spread of XYZ hole diameters resulting from the drilling or punching
in a circuit board using current equipment, materials, workers, and so forth. The
normal curve says nothing about the range of XYZ hole diameters acceptable to
the customer. This curve summarizes the empirical quantification for the variabil-
ity that exists within the XYZ hole manufacturing process .
The dashed vertical lines on the curve in Figure 2 .1 represent the number of
standard deviation units (a) a given hole diameter might be from the mean, which
is shown as x on the x-axis . In the example we are using here, the standard de-
viation (a) would be expressed in units of inches. The basic formula for calcu-
THE STATISTICAL DEFINITION OF SIX SIGMA 39
Normal Distribution
Shifted 1 .5a
Lower Upper
specification specification
limit limit
1000000
100000
10000
Average company -- ,.:
1000 -_
100
10-~
Best in class ---
0 .01
Figure 2 .3 Implication of sigma quality level . Part per million (ppm) rate for part or
process step, considers a 1 .5 6 shift of the mean where only 3 .4 ppm fail to meet speci-
fication at a six sigma quality level .
THE STATISTICAL DEFINITION OF SIX SIGMA 41
30x Improvement
100000 66,810 ppm +.
10000
6210 pp~' -.. 70x ImProvem ant
1000
a.a 100 233 ppm"
10
1 3 .4 ppm
D 0.1
0.01 .
0.001 3
4 5 g
Sigma quality level
1 .5 Sigma shift
Centered:
One point that should be emphasized is that sigma quality levels can be
deceptive since theybear aninverse, nonlinear relationship to defectrates. Higher
sigma quality levels mean fewer defects per million opportunities, but the rela-
tionship is not linear. Figure 2.4 illustrates that an improvement from a three to
a four sigma quality level is not the same as an improvement from a five to a six
sigma quality level . The first "unit" shift noted above corresponds to a lOx
improvement in defect rate, and the latter "unit" shift to a 70x improvement,
where these comparisons are based on a ±1 .5 6 shifted process . A unit change
in sigma quality level does not correspond to a unit change in process improve-
ment as measured by defect rates . A shift in sigma quality level from five to six
sigma is a much more difficult improvement effort than a shift in sigma quality
level from three to four sigma.
The sigma quality level metric is very controversial. An organization needs
to understandthe concept behind this metric even if it does not use the metric as
a driving force within the organization. We suggest that organizations select
other Six Sigma metrics over the Sigma Quality Level metric to monitor im-
provement activities within their organizations . However, if the need arises to
convert between defect rates and sigma quality level, Table 2.1 makes the con-
version simple .
Additional Six Sigma metrics are described in Chapter 5.
42 SIX SIGMA BACKGROUND AND FUNDAMENTALS
+/ - Percent Defective
Sigma Percent Defective within spec .: ppm: 1.5
Level at within spec.: ppm: 1.5 Sigma Sigma
Spec. Centered Centered Shifted Shifted
Limit* Distribution Distribution Distribution Distribution
*Sometimes referred to as sigma level or sigma quality level when considering process shift.
An organization that is considering Six Sigma might also consider the implica-
tions of doing nothing . If any one of the current competitors of an organization,
or even a new competitor, achieves six sigma quality levels in the industry or
market segment of the organization, the organization's profitable days could be
numbered. When Six Sigma is implemented wisely, it is not just another quality
program. Six Sigma can bring orders of magnitude improvement over histori-
cal quality levels and can totally change the rules by which an organization
operates . Once someone in an industry implements a successful and enduring
Six Sigma business strategy, the way to become or remain the industry leader is
to implement Six Sigma smarter than the competition . The important question
may not be if, but when, an organization should adopt a Six Sigma business
strategy.
Three sigma quality levels that have been prevalent for the past half:century
are no longer acceptable . Consider, for example, that a 99.9% yield would re-
sult in the following levels of performance in service industries (Schmidt and
Finnegan, 1992; Wortman, 1999):
To emphasize the implications ofa 0.1 defectrate (99 .9% yield), consider
the following : your heart skips a few hundred beats; your checkingaccount was
accidentally . debited $1,000 for an expense you did not incur; your vacation
flight crashed at O'HareAirport; your left kidney was removed whenit was the
right kidney that was cancerous ; your Social Security check was eaten by the
COST OF POOR QUALITY (COPQ) 47
(direct labor and direct materials) . But the initial emphasis on quality costs was
not fully accurate, as only the most obvious categories of quality costs were
addressed. These are shown as the tip of the iceberg in Figure 3.1 .
As the quality movement progressed throughout the latter half of this cen-
tury, it became obvious that the costs associated with quality could represent as
much as 15%-25% of total operating costs (Wortman, 1995) and that many of
these costs were not directly captured on the income statement or balance sheet.
These truly "hidden" quality costs were those shown below the water line in
Figure 3.1 .
The addition of technical specialists within the quality department helped
define and focus on these hidden quality costs. Calculations of large quality
costs caught the attention of the executive level of management . Eventually
these costs were referred to as the "cost of poor quality" (COPQ, since they
represented unsatisfactory products or practices that, if eliminated, could sig-
nificantly improve the profitability of an organization. Over a period of de-
cades a number of surprising facts surfaced about COPQ (Juran, 1988).
" Quality-related costs were much higher than financial reports tended to
indicate (20%-40% of sales) .
" Quality costs were incurred not only in manufacturing but in support ar-
eas as well .
" While many of these costs were avoidable, there was no person or organi-
zation directly responsible for reducing them .
SS Titanic
Management
Figure 3.1 Cost of poor quality (Reproduced with permission of RAS Group.)
COST OF POOR QUALITY (COPQ) 49
Eventually, the COPQ was defined in more detail. At the present time, these
costs are allocated to three major categories: prevention, appraisal, and failure .
In addition, the area of failure cost is typically broken into two subcategories :
internal failure and external failure . Prevention costs are devoted to keeping
defects from occurring in the first place . Appraisal costs are associated with
efforts (such as quality audits) to maintain quality levels by means of formal
evaluations of quality systems . Failure costs refer to after-the-fact efforts de-
voted to products that do not meet specifications or that fail to meet customers'
expectations . Table 3.1 gives examples of individual costelements within each
of these major categories (Wortman, 1995).
An S4 business strategy directly attacks the cost of poor quality (COPQ).
Quality cost issues can dramatically affect a business . Wisely applied Six Sigma
techniques can help eliminate or reduce many of the issues that affect overall
cost. However, management needs to ask theright questions so that these issues
are effectively addressed, or set up an infrastructure designed to inherently ask
these questions . For example, "beating up" workers and supervisors for high
scrap rates is not very effective if the cause of the high rates is inadequate
operator training, outdated production equipment, or poor test equipment for
measuring quality parameters.
PREVENTION
Training
Capability Studies
Vendor Surveys
Quality Design
APPRAISAL
Inspection and Test
Test Equipment and Maintenance
Inspection and Test Reporting
Other Expense Reviews
INTERNAL FAILURE
Scrap and Rework
Design Changes
Retyping Letters
Late Time Cards
Excess Inventory Cost
EXTERNAL FAILURE ,
Warranty Costs
Customer Complaint Visits
Field Service Training Costs
Returns and Recalls
Liability Suits
50 SIX SIGMA NEEDS ASSESSMENT
In summary, the concept ofCOPQ can help identify Six Sigma projects that
(1) have the potential for significant monetary savings, (2) are of strategic im-
portance, (3) involve key process output variable (KPOV) issues, and (4) are of
vital concern to the customer. It would be ideal if a Pareto chart ofthe monetary
magnitude of the 18 COPQ subcategories listed inTable 3.1 could be created so
that areas for improvement could be identified; however, creating such a chart
could involve a lot of effort. As a start, one might begin by quantifying the
COPQ within one strategic area of a business, perhaps on-site rework, and then
later work at quantifying the magnitudes of COPQ within other areas of the
business . In a subsequent chapter, we specifically discuss how COPQ ties in to
effective project selection .
We have addressed the question "Is Six Sigma worth doing?" The flip side of
that coin is "What will it cost us if we don't do Six Sigma? What is the cost of
doing nothing differently?"
When assessing Six Sigma, leadersshould consider the various options: doing
nothing, creating a Six Sigma initiative, and creating a Six Sigma business strat-
egy. Let's consider each of these options.
The "doing nothing" option might be the right choice for an organization;
however, an organization needs to make this decision after comparing the cost
of doing nothing to the cost of doing something .
The "creating a Six Sigma initiative" option typically is viewed by members
ofthe organization as the "program of the month" and risks early abandonment
without much benefit. In the worst case, the initiative is dropped after a lot of
money and resources are spent on it.
The "Six Sigma business strategy" option has the most benefit if it is ex-
ecuted wisely. Our main goal with this book is to give management enough
insight into the Six Sigma approach that they can decide whether or not to
spend the time and money seriously investigating this approach.
One of the most serious mistakes we can make when trying to answer this
question has to do with simple definitions . Many executives confuse efficiency
with effectiveness . Efficiency is "doing things right." Effectiveness is "doing
the, right things." A wise implementation of Six Sigma focuses on "doing the
right things right ."
Often, people want to see a specific example of Six Sigma that closely parallels
their own situation so that they can see how they will benefit. However, often
ASSESSMENT QUESTIONS 51
there are situations where people have difficulty seeing how Six Sigma tech-
niques are directly applicable to their situation no matter how many examples
are shown . This is particularly true of government organizations and govern-
ment contract organizations, where ROI may not be a primary driving force.
Consider an employee who thinks her organization can benefit from Six
Sigma methodology but faces resistance from other employees who do not see
the benefits . This person might ask management and other employees ques-
tions like those shown in Table 3 .2.
Organizations often collect data and then react to the data as though they were
pieces of information. Consider the following when addressing this behavior
(O'Dell and Grayson, 1998) :
Ifdataare not collected and used wisely, their very existence can lead to activi-
ties that are ineffective and possibly even counterproductive . In this chapter we
discuss how to create information from the wise collection and analysis ofdata.
A plot ofthis information is shown in Figure 4.1. From it we can see that the
reaction described above does not improve the process or lessen the likelihood
of having problems in the future . Figure 4.2 shows a replot of the data as an
individuals control chart. The control limits within this figure re calculated
from the data, not the customer specifications, as described inBreyfogle (1999),
Implementing Six Sigma (chapter 10). Since the up-and-down movements are
within the upper and lower control limits, it is concluded that there are no
EXAMPLE 4.1 : REACTING TO DATA 57
10 20 30 40 50
Measurement Sequence
Figure 4.1 Reacting to common cause data as though it were special cause
3.OSL=80.68
Statistically
X=76 .09 Calculated
Process
Control
Limits
-3 .OSL=71 .50
special causes within the data and that the source of the variability is common
cause from the process .
This organization had been reacting to the out-of-specification conditions as
though they were special cause. The focus on "fixing" out-of-specification con-
ditions often leads to "firefighting" (an expression that describes the perform-
ing of emergency fixes to problems that recur) . When firefighting activities
involve tweaking the process, additional variability can be introduced, which
can further degrade the process rather than improve it.
Both public and private organizations frequently look at human and ma-
chine performance data from a perspective similar to those just described and
then make judgments based on the data. Production supervisors might con-
stantly review production output by employee, machine, work shift, day of the
week, product line, and so forth. In the service sector, an administrative assistant's
daily output of letters and memos may be monitored . In call centers around the
world, the average time spent per call may be monitored and then used to coun-
sel low-performing employees . The efficiency of computer programmers may
be monitored through the tracking of "lines of code produced per day." In a
legal department, the number of patents secured on the company's behalf dur-
ing the last fiscal year may be compared to previous annual rates. Whenever an
organization reacts to individual situations that do not meet requirements and
specifications, rather than look at the system of "doing work" as a whole, they
could be reacting to common cause situations as though they were special cause.
To illustrate this point further, let's look at an organization that monitors the
frequency of safety memos . Consider a memo written indicating that the num-
ber of "accidents involving injuries" during the month of July was 16, up by 2
from a year ago . The memo declares this increase in accidents to be unaccept-
able and asks all employees to watch a mandatory, 30-minute safety video by
the end of August . At an average wage rate of $10 per hour, the company pay-
roll of 1,500 employees affects the August bottom line by $7,500, which doesn't
include wasted time as employees get to and from the conference room. Nor
does it take into account the time spent issuing memos reminding people to
attend, reviewing attendance rosters looking for laggards, and so forth. How
effective was the training video at eliminating the common cause of injury?
We are not saying safety and productivity are'not important to an organiza-
tion. We are saying, as did Dr. Deming, that 94% of the output of a person or
machine is a result of the system that management has put in place for use by
the workers. Ifperformance is poor, 94% of the time the system must be modi-
fied for improvements to occur. Only 6% of the time are problems due to spe-
cial causes. Knowing the difference between special and common cause variation
can affect how organizations react to data and the success they achieve employ-
ing a Six Sigma strategy. To reduce the frequency of employees' safety acci-
dents from common cause, an organization needs to look at its systems
collectively over a long period to determine what should be done to improve its
PROCESS CONTROL CHARTING AT THE "30,000-FOOT LEVEL" 59
Consider the view when looking out an airplane window. When the airplane is
at an elevation of 30,000 feet, passengers get a "big picture" view of the land-
scape. However, whenthe airplane is at 50 feet, passengers view a much smaller
portion of the landscape. Similarly, a "30,000-foot level" control chart gives a
60 NUMBERS AND INFORMATION
macro view of a process, while a "50-foot level" control chart gives more of a
micro view of some aspect of the process .
Within training sessions, control charts are most often taught to timely iden-
tify special causes within the process at the "50-foot level." An example of this
form of control is to timely identify when temperature within a process goes
out of control so that the process can be stopped and the temperature-variable
problem fixed before a large amount ofproduct is produced with unsatisfactory
characteristics .
However, control charts are also very useful at a higher level; where focus is
given to directing activities :away from firefighting the problems ofthe day. We
suggest using these charts to prevent the attacking of common cause issues as
though they were special cause . We want to create a measurement system such
that 'ire prevention" activities are used to address common cause issues where
products do not consistently meet specification needs. Our S4 road map gives a
structured approach for "fire prevention" through the wise use and integration
of Six Sigma tools.
Unlike what is called for by "50-foot level" control charts, we suggest an
infrequent sampling to capture how the process is performing overall relative
to customer needs. When the sampling frequency is long enough to span all
"short-term" process noise inputs, such as raw material differences between
days or daily cycle differences, we call this a "30,000-foot level" control chart.
At this sampling frequency, we would examine only one sample and plot its
response on an individuals control chart.
When first introduced to this concept ofinfrequent sampling, an individual's
typical concern is that this measurement does not give any insight into what
should be done to improve the process . The concern is understandable, but the
purpose ofthis type of measurement is not to determine what should be done to
improve the process, but rather:
We do not suggest that practitioners get bogged down, atthe outset, trying to
collect alot of data with the intent ofidentifying a cause-and-effect relationship
that shows the source of their problem. We do suggest that this form of data
collection can be appropriate after a problem has been identified and after teams
have determined what they think should be monitored.
Process characteristics that are deemed important and require monitoring
for the purpose ofdetermining cause-and-effect relationships need to be identi-
fied up front during "brainstorming activities" within the S4 measurement phase .
We refer to this collection ofvarious activities such as process mapping, cause-
DISCUSSION OF PROCESS CONTROL CHARTING AT "30,000-FOOT LEVEL" 61
and-effect diagram, cause-and-effect matrix, and failure mode and effects analy-
sis (FMEA) as "wisdom of the organization" outputs.
To summarize, when selecting a rational subgroup to create a "30,000-foot"
control chart, the organization needs to create a sampling plan that will give a
long-term view of process variability. A sampling plan to create a baseline of a
process might, be to randomly select one daily KPOV response from a data set
compiled during the last year. An individuals control chart could then reveal
whether the process has exhibited common cause or special cause conditions.
Action to investigate and resolve the root cause of special cause conditions may
then be appropriate .
Many special cause conditions appear on a control chart with too frequent
sampling. These conditions could be viewed as noise to the system when exam-
ined using a less frequent sampling approach. We believe that long-term regu-
lar perturbations should be viewed as a common cause of the system/process
and should be dealt with accordingly, if they impact KPOV variation. That is,
they should be addressed looking at the process inputs/outputs collectively rather
than as individual special cause conditions .
For processes that exhibit common cause variability, the next step is to as-
sess the KPOV relative to the needs of customers . This assessment is most
typically made relative to specification requirements for example, process
capability/performance indices such as C , C ~, Pp, and P k. However, we need
to note that indices of this type can be deceiving, as described within Breyfogle
(1999), Implementing Six Sigma (chapter 11).
A probability plot of the data can be a good supplementary tool in better
understanding expected variability. This form of output can also be very ben-
eficial in describing the "process capability/performance" of a transactional/
service process . A probability plot for this process might indicate that 80% of
the time it takes between two days and six weeks to fulfill orders. For an S4
project in this area, someone might thenestimate the cost impact ofthis process
variability on an organization and/or on general customer dissatisfaction level.
These values could thenbe the baseline against which S4 projects are measured.
When this measurement and process improvement strategy is implemented,
true change can be detected as a shift of the "30 ;000-foot level" control chart.
This chart should be constantly monitored with an eye toward detecting a shift
in the control chart pattern as a result of S4 improvements .
techniques .In an attempt to allay this concern, this section will elaborate on the
technique further.
The classic approach to creating control charts is that subgroups be chosen
so that opportunities for variation among the units within a subgroup are mini-
mized. The thinking is that if variation within a subgroup represents the piece-
to-piece variability over a very short period oftime, then any unusual variation
between subgroups would reflect changes in the process that should be
investigated .
Let us consider a process that has one operator per shift and batch-to-batch
raw material changes that occur daily. Consider also that there are some slight
operator-to-operator differences and raw material differences frombatch to batch,
but that raw material is always within specification limits . If a control chart is
established where five pieces are taken in a row for each shift, the variability
used to calculate x and R control chart limits does not consider operator-to-
operator and batch-to-batch raw material variability. If the variability between
operators and batch-to-batch is large relative to five pieces in a row, the process
could appear to be going out of control a lot. When a process goes out of con-
trol, we are told to "stop the presses" and fix the special cause problem . Much
frustration can occur in the manufacturing process, because employees will
probably be trying to fix a special cause problem over which they have no
control .
One question that might be asked is whether the manufacturing line should
be shut down because of such a special cause. It could even be questioned
whether out-of-control conditions caused by raw material should be classified
as special cause. Note that this point does not say that raw material is not a
problem to, the process even though it is within tolerance. The point is whether
the variability between raw material should be treated as a special cause . It
seems that there is a very good argument to treat this type of variability as
common cause. If this is the case, control limits should then be created to in-
clude this variability .
To address this problem, we suggest an infrequent sampling plan where only
one KPOV is tracked on a control chart. To address whether different variabil-
ity sources are causing a problem, the long-term variability of the process could
then be compared to specification needs, as illustrated in Figure 4.3. Math-
ematically, this comparison could be expressed in process capability/perfor-
mance units such as C~, C ~, P~, and P k. In addition, this variability could be
described as percent beyond specification, or in parts-per-million (ppm) defect
rates. Ifthere is no specification, as mightbe the case in transactional or service
processes, variability could be expressed in percent of occurrence through the
use of normal probability plots. An example of this application occurs when
80% of the time a purchase order takes between 10 days and 50 days to fill.
No time sequencing
of data
Time sequencing
No time
4 of data-
sequencing
of data
,AA--AAA f
Lower Upper
spec . spec .
limit limit
,A- A
Measurement sequence
Figure 4.3 Control charting and process capability/performance assessment at the "30,000-foot level" view
64 NUMBERS AND INFORMATION
If the process is not capable of meeting specification needs, then further analy-
sis of the process using additional tools could show that the process is not robust
to raw material batch-to-batch variations, as illustrated in Figure 4 .4 . Graphical
tools described in Breyfogle (1999), Implementing Six Sigma (chapter 15), that
give insight into these differences include multi-vari charts, box plots, and mar-
ginal plots . Analytical tools that mathematically examine the significance of these
observed differences include variance components analysis, analysis of variance,
and analysis of means .
Lower Upper
Spec. Spec.
Limit Limit
When we react to the "problem ofthe day," we are not structurally assessing
this type of problem . With an S4 strategy we integrate structural brainstorming
techniques (e.g., process flowcharting, cause-and-effect diagram, and FMEA),
passive analyses (e.g., multi-vari analysis, ANOVA, regression analysis), pro-
active experimentation (e.g., fractional factorial Design of Experiments), and
control (e.g., error proofing and control charting at the "50-foot level") to iden-
tify and fix the process problem and prevent its recurrence.
It should be noted that the discussion above does not suggest that an infre-
quent sampling plan using individual charting is the best approach for all pro-
cesses. What it does suggest is that perhaps we often need to step back and look
at the purpose and value of how many control charts are being used. Surely
control charts are beneficial in indicating when a process needs to be adjusted
before a lot of scrap is produced . However, control charts can also be very
beneficial at a higher plane of vision that reduces day-to-day firefighting of
common cause problems as though they were special cause .
We are trained to think that goals are very important, and they are . Goals should
be SMART (i.e., simple, measurable, agreed to, reasonable, and time-based) .
However, these guidelines are often violated. Arbitrary goals that are set for
individuals or organizations can be counterproductive and costly when indi-
Figure 4.5 Control charts at the "30,000-foot level" : Attribute response
GOAL SETTING, SCORECARD, AND MEASUREMENTS 67
viduals or organizations have no control over the outcome . Consider the fol-
lowing two situations within your personal life.
Situation 1
You have a lot of work that needs to get done around your house or apartment.
Setting a goal to get these tasks out of the way on a Saturday would solve the
problem. Consider these two alternative scenarios:
" You can set a goal for the number of tasks you would like to accomplish,
create a plan on how to accomplish these tasks most efficiently, and track
how well you are doing during the day to meet your goal. You would
probably get more done with this strategy than by just randomly attacking
a list of to-do items .
" Your spouse or friend can set an arbitrarily goal for the number of tasks
you are to accomplish. If they don't create a plan and the goals are set
arbitrarily, you probably will not accomplish as much as you would under
the first scenario, since there is no plan and you do not necessarily have
buy-in to the tasks that are to be accomplished.
Situation 2
You would like to make an overall rate of return of 50% on your financial
investments next year. Consider these two alternative scenarios :
You purchase mutual funds that have tracked well over the last few years .
However, their yield to date is not as large as you had hoped. Because this
goal is important to you, you decide to track the yield of your portfolio
daily and switch money in and out of various funds in order to meet your
goal.
You evaluate the performance of a wide range of mutual funds that have
multiyear performance records. You determine that your initial goal was
too aggressive since the majority offunds have not produced the yield that
you desire . Furthermore, funds that yield high rates of return one year
often have trouble repeating this level of performance . You alter your plan
and choose an ` investment strategy of buying and holding a diversified
portfolio of index mutual funds .
Situations one and two are very different relative to goal setting . In the first
situation, goal setting can be very useful when there is buy-in to the plan, be-
cause we do have some control over the outcome.
In the second situation, goal setting could be very counterproductive since
we do not directly control the outcome of the process (i .e., how well the stock
68 NUMBERS AND INFORMATION
market will do next year). The first scenario in situation 2 will probably add
variability to your investment process and may cause significant losses . The
second plan for addressing this goal is the best, because we did research in
choosing the best process for our situation and took the resulting yield, even
though we might not achieve our goal.
Management can easily fall into the trap of setting arbitrary goals where the
outcome is beyond the employee's control . Consider a manufacturing process
that derives 90% of its problems from supplier quality issues . Often the ap-
proach in this situation is to "beat up" on suppliers whenever quality problems
occur or to add inspection steps. This may be the only course of action manu-
facturing employees can take, but it will not be very conducive to meeting im-
provement goals.
However, if management extends the scope of what can be done by employ-
ees to fix problems, real productivity improvement can result . For example,
manufacturing employees can become more involved with the initial supplier
selection . Or perhaps manufacturing employees can be permitted to work with
engineering to conduct a design ofexperiments (DOE) where several two-level
factors are evaluated within an experiment for the purpose of improving the
settings in the manufacturing process . Through this DOE, perhaps a change in
aprocess temperature orpressure setting can be made that will result in a higher-
quality product using raw material of lesser quality .
We do not mean to imply that stretch goals are not useful, because they are .
Stretch goals can get people thinking "out ofthe box ." However, goals alone
without a real willingness to change and a road map to conduct change-can be
detrimental .
There is another consideration when setting goals : the scorecard : Vince
Lombardi, the highly successful American football coach of the Green Bay
Packers, said that "if you are not keeping score,you arejust practicing." Orga-
nizations should not just measure against sigma quality levels and/or process
output defect rates . Ifthey do, they can be trapped into "declaring victory" with
one metric at the expense of another metric . For example, the financial mea-
surements of Xerox in the 1970s were excellent in part due to their servicing of
the copiers that they developed and manufactured. However, customer satisfac-
tion was low because of frequent paper jams and servicing . When less expen-
sive and more reliable copiers were made available by the competition, the
company had a severe financial crisis. In a later chapter we discuss consider-
ations for a balanced scorecard .
The success of Six Sigma depends upon how goals are set and any bound-
aries that are set relative to meeting these objectives. In addition, if the scorecard
is not balanced, people can be driving their activities toward a metric goal in
one area while adversely affecting another metric or area. The result is that
even though an employee may appear successful since he has met his goal, the
overall organization can suffer.
SUMMARY 69
4:6 SUMMARY
Metrics can be very beneficial within an organization when they are used in
such a way that they give direction to what is the most appropriate activity for
a given situation . However, some metrics can have no value or might even
counter productive activity. For example, little value is realized when an orga-
nization reports metrics merely as a status and never makes an orchestrated
effort to improve them. In addition, counterproductive results can occur when
organizations choose to use one type ofmetric throughout their organization. In
other words, creating a "one-size-fits-all" metric, such as sigma quality level,
as described earlier in this book, can lead to frustration, fabrication ofmeasure-
ments, and "playing games with the numbers ."
Organizations should consider using only those Six Sigma metrics that are
appropriate for each situation. Chapter 9 of Breyfogle (1999), Implementing
Six Sigma, not only describes many possible Six Sigma metrics alternatives
including DPMO, rolled throughput yield, and process capability/performance
(Cp, C k, PY and P k) but also addresses the confusing and controversial as-
pects of metrics . Next we will describe some of these Six Sigma metrics fol-
lowed by example strategies .
Process Capability
Process capability is a commonly used metric within Six Sigma. The following
equations are used to determine the process capability indices C and C k:
71
72 CRAFTING INSIGHTFUL METRICS
where USL and LSL are the upper specification limit and lower specification
limit, respectively, a is standard deviation,, and [t is the mean of the process,. .
Figure S.1 and Figure 5.2 illustrate how the distribution of a. dimension could .
be related to a specification, yielding various process capability indices of C
P
and Cpk. A process is said to be equal to at "six sigma quality level" if C = 2.00
: P
and CA = 1 .5, which equate to 3.4 PPM using the 1 .5 standard shift, as previ-
ously described within this book.
For a visual image of the meaning inherent to C11 consider that the numera-
tor of the CP equation represents the width of a garage door opening. Likewise,
consider the denominator of the CPequation to be the width of a new sports car.
If the door width and the car width are identical, then CP is equal to a numerical
value of 1 . It also means that the driver needs to be extremely careful when
pulling into the garage. Any slight miscalculation will lead to a scratch or dent.
A process with a CP value of one is not considered capable of consistently meet-
ing specifications, because any slight shift in the process mean, or any increase,
in process variation, would lead to the production of defective product based on
a ±3(y processes requirement .
A CP value of two means that the garage door is twice as wide as the car, or
126 units wide. A process capability index of 2 means that even if the process
Lower
Upper
Specification
Specification
Limit (LSL)
Limit (USL)
Lower
Upper
Specification
Specification
Limit (LSL)
Limit (USL)
variability were to double to an exact width of 126 units, the process still would
not exceed the customer's specification limits if it were centered.
Care must be exercised when determining and utilizing C and Ck process
capability indices . First, the previously described equations are appropriate when
data re normally distributed. Second, the sample size and how the data are
collected from a process can dramatically affect estimates of these 6 values . If
data are collected infrequently from a process and then analyzed, we might get
a value forprocess capability data dramatically different from that gotten when
the data re collected over a very short period of time.
In general, there are several methodologies for data collection and proce-
dures for calculating process capability values that can result in very different
results for C and C k values. Breyfogle (1999), in Implementing Six Sigma
(chapter 11), describes many procedures for the calculation of standard devia-
tion for process capability/performance equations . These procedures vary from
determining a short-term value to determining a long-term value using variance
components techniques. Some people insist that a short-term view of standard
deviation is to be used with respect to these equations ; others hold that a long-
team view of standard deviation is proper. Obviously, these methodologies can
give very different results :
Another problem with requiring thatall organizations use process capability
metrics within Six Sigma is that typically service/transactional processes do
not have published specifications that are similar to those within manufactur-
ing. To calculate C and Ck, specifications are needed; hence, people could be
74 CRAFTING INSIGHTFUL METRICS
that were parts of the hot-water heater assembly process, we may get a value of
only 4 parts. Ifwe were to calculate a DPMO rate using this number (i.e., 4) for
failure opportunities, our sigma quality level would be very poor. In an effort to
make our numbers look better, we might get ridiculous and "count the number
of0.001 inches of welding joints within the hot-water heater assembly process"
as opportunities for failure . Ifwe were to do this, our DPMO rate could perhaps
improve such that a 25% yield (75% defective rate) appears to have a seven
sigma quality level . We don't mean to suggest or condone such activity when
calculating a Six Sigma metric; however, this is the type ofactivity that actually
occurs whenever an organization requires a metric such as sigma quality level
across all organizations and processes .
Another situation that can also lead to "playing games with the numbers"
occurs ifa company requires sigma quality level measures for its service/trans-
actional processes . For this situation we might have basic desires of customers
but no specification requirements as they would typically exist within manu-
facturing. For example, within a service/transactional process someone might
have chosen a target that all deliveries are to be within ±1 day of their due date;
however, this specification does not have the same importance as a tolerance
for a piston that is to be part ofan automobile engine assembly. When we calcu-
late many of the Six Sigma metrics, such as process capability/performance,
DPMO rates, and sigma quality levels, we need specification values in order to
perform the calculations. If we are to force any of these Six Sigma metrics for
example, sigma quality level within service/transactional processes the orga-
nization responsible will often need to fabricate specification values in order to
perform the calculations . This activity can lead to the adjustment ofthese "speci-
fication values" so that their overall sigma quality value "looks good."
Hopefully, our hot-water heater and service/transactional examples illustrate
the dangers that can occur when an organization succumbs to the temptation of
using one-size-fits-all metrics across organizational disciplines . The metrics
associated with Six Sigma can be very valuable ; however, it is important to
determine, use, and report the most appropriate metric(s) for any given situa-
tion/process . In addition to reporting and tracking the important metrics within
a process, we suggest also reporting the monetary impact of the metrics:
Figure 5.3 graphically illustrates the difference between process yield and
RTY. We will use the left side of this figure to illustrate RTY and the right side
of this figure to illustrate process yield. In both cases we will assume that our
master schedule calls for us to produce the "1,000 units" highlighted at the top
of the figure . Notice that our process consists of five subprocessing steps, each
with its own distinctive yield (Y) . The yield ofthe first subprocess, Y1 , is 0.92,
or 92°10. Focusing again on the left-hand side of this figure, we notice the num-
ber of units decreasing from 1,000 to 920 to 754 to 633 to 519 to 493. This is a
direct result of applying each subprocessing step's yield directly to the number
of units that initially enter the work cell for batch processing . When 1,000 units
enter subprocess Yl; only 920 units exit without some form of further process-
ing. Looking at the right side of the figure, one can tell where those 80 units
(1,000- 920) went: 40 were scrapped and 40 were reworked and put back into
the production cycle at some later time. For this simple example we will as-
sume that all reworked material successfully passes to the next subprocess with-
out further incident andthat the percent reworked and scrapped are the same at
each process step. The 920 units that avoided scrap and rework during subpro-
cess Y1 enter the second subprocess, Y2 (with a measured yield of 82%), result-
ing in 754 good parts . And so on until all five subprocessing steps deliver 493
good units . This` 49.3% rolled throughput yield (RTY) value can also be esti-
mated by merely multiplying the yields of the five sequential subprocessing
steps . The numbers are identical when the round-off error of the numbers is
ignored .
The right side of Figure 5.3 illustrates how process yield can be calculated.
During the first subprocess the initial 1,000 units is reduced to 960, since we
had to scrap 40 units that could not be salvaged. The reworked units (4%, or 40
units) subjected to rework will eventually show up in inventory and hence are
counted as good product. The only deviation from perfection is the scrap (40
units), resulting in a process yield through subprocess Y1 of 96%, or 960 units.
By the time we get to the end of the fifth subprocess, the process yield has
declined to 71 .4%.
The two methods of calculating yield for this example are 71 .4% and the
49.3% . The accounting systems of organizations typically ignore rework and
report the higher value. The rework process that we are missing within our
calculations is referred to as the "hidden factory."
In summary, rolled throughput yield is a measure of how well companies
"process quality." Process yield measures how well one "processes units." It is
not uncommon to have a multistage process where yield at each step is quite
high (say, at or above 90%) but the rolled throughput yield is at 50% or below.
An RTY of 50% indicates that only one out of two units completes the entire
production process without being reworked or scrapped, giving more insight
into process performance .
78 CRAFTING INSIGHTFUL METRICS
Rolled
/000 1111 i1 .s
Throughput
SCRAP(4%) =
Yield (RTY)
YI = 0.92
REWORK (4 %)
920...1.
units.-.- .-. ...-.- .-.- .-.- .-.-.-.-.- .-.- .-.- .-.- .-.- .-. .. .-.-.-.-. 960 units
.... ... ... . ... . ... . ....
SCRAP (9 %) a
0 - 86 units
Y2 = 0:82
REWORK (9
519
-.- .- .- .- .- - .- .- .- .- .- .-units 732
.------- .- .- .- .- .- .- .- .- .- .- .- .- .- .- .- .- .- .- .- .-units
.- .- .- .- .- .- .- .- .-
.., SCRAP 2.5 % = _ 18 units
Ys= 0.95
REWORK (2 .5 %) =
. . .. . . .. .. . .. .. .. . . ..493
. . .. units
. . .. . . .. . . .. .. . . .. . . .. . . .. .. .. .. . . .. . . .. . .714 units
.. . . .. . . . . . ._ . . . .. . . .. .
Executive direction can sometimes lead to the creation and mandatory use of
metrics that have no long-term or even short-term benefit . For example, the
vice president of manufacturing might indicate that she wants a weekly listing
of defectives produced on the ABD production line, and a subsequent action
plan for fixing these defectives so they don't recur. This type of direction often
RIGHT QUESTION, RIGHT METRIC, RIGHT ACTIVITY 79
leads to firefighting, where common cause variation is treated as ifit were spe-
cial cause. Amore appropriate question along these same lines might be: "Pro-
vide me a control chart showing the defect rate over time. If our process is
in-control and the average failure rate is too high, provide a Pareto chart de-
scribing the frequency of defect types over several months and then describe an
actionplan to reduce the magnitude ofthe categories thathave the largest value ."
Rather than engaging in firefighting, management should consider asking
for a "30,000-foot level" control chart of defectives, as illustrated in Figure 5.4
and as described previously in this book. The control chart on the left side of
this figure shows an average defect rate slightly above 20% . This process is
considered to be in statistical control, since there are no special causes present
based on these data. Assuming that this reject rate is unacceptable, a reasonable
next step is to organize observed defects into logical categories. A transforma-
tion of this data into a Pareto chart is also illustrated in Figure 5.4. A potential
next step would be to request a project plan to address the most frequent cat-
egory of defect, using Six Sigma tools. Also, it might be helpful to quantify
how much each type of defect is costing the company in financial terms . This
would help prioritize this potential Six Sigma project relative to other improve-
ment opportunities .
The example used above involved percent defective attribute data. Figure
5.5 illustrates what a completed process improvement scenario might look like
when plotting continuous data in the form of an individuals control chart. This
control chart indicates that some sort of process change has created a signifi-
cant shift in the process mean to a "better (smaller)" value . When we superim-
pose customer specification limits on the ±36 normal distribution, we get a
visual representation indicating that the frequency ofnonconformance has gone
down significantly as a result of the process change .
These two examples show how the "30,000-foot level" control chart can be
used to give meaning and direction within anorganization. The other Six Sigma
metrics described earlier in this chapter can be beneficial fortracking ; however,
they are not appropriate for every situation . In many cases their use can be a
terrible waste oftime and effort . For example, rolled throughput yield (RTY) is
not something that should automatically be measured whenever there is a trouble-
some product or process . It takes a lot of time and money to collect the data
necessary to estimate rolled throughput yield on a process of even a modest
size (five to ten steps) . At eachindividual step in the process one would need to
keep track of scrap, rework, number of defects, number of opportunities for
defects, and so forth. In addition, one needs to permit enough product to flow
through the process so that reasonable estimates of throughput yield can be
obtained for each process step. One would only go down this path for data
collection when there is evidence that a process is producing a lot of rework
and/or scrap and ifthis data collection can give insight to meaningful improve-
ment/tracking activities.
UCL=0.3813
P=0.2096
LCL=0.03694
Measurementsequence
Customer
Customer
needs
needs
Frequencyof
Frequency of
nonconformance
nonconformance
reaction to the situation is to immediately assume that the problem is the result
of special cause variation (i.e., let manufacturing identify and fix the problem).
Let's look at the situation froma "30,000-foot level" perspective by examin-
ing all of the data as if it came from the same continuous process . We are going
to ignore the individuality of the distinct product lines, since all products basi-
cally encounter the same process of design, design qualification testing, and
manufacturing. In other words, we, are considering product differences to be
noise to our overall process . From this point of view, we can hope to determine
if the failures are special or common cause in nature. If the failures are due to
common cause variation, we might be able to collectively view the common
cause data to gain insight to what should be done differently within the process
to reduce the overall frequency of failure . If the failures are due to special cause
variation, we might be able to examine this/these special cause condition(s)
individually to determine why these data differ from our normal process and
then take appropriate action(s) .
In lieu of compiling and reporting the data in attribute format, we will col-
lect and report conformance data in the form of continuous data. The raw data
used to construct the XmR control charts in Figure 5.6 are listed below. The
time sequence in which the data were collected can be reconstructed by reading
the raw data from left to right, starting with the first row and continuing to rows
two and three, using the same left-to-right pattern as in row one . The govern-
ment specification simply states that raw data values at or below zero are all
equally acceptable, whereas positive values are unacceptable, with the degree
of unacceptability increasing as the positive values increase. In other words, 0
is good, 0.1 is bad, 1.0 is worse, 10.0 is terrible, and so forth. Any raw data
values greater than zero for this sampled product represent an out-of-specifica-
tion condition that must be resolved within the manufacturing line.
The XMR chart in Figure 5.6 indicates that the seventeenth sequential data
point, the bold value of 11.2 in the raw data table, is "out of control" or statisti-
cally different from the other readings from the process . Our initial reaction to
the magnitude of this point is that a special cause condition may now be occur-
ring within the manufacturing process that needs to be resolved. It should be
noted that the number(s) 1 plotted within Figure 5 .6 are placed there by the
statistical analysis software to indicate that rule number 1 has been violated.
Rule number 1 indicates that the process is out of control because a single data
point is outside of the. ±3a control limits . See Chapter 10 in Breyfogle (1999),
Implementing Six Sigma, for additional information on these and other control
chart rules.
EXAMPLE 5 .1 : TRACKING ONGOING PRODUCT COMPLIANCE 83
O 10 3.OSL=9 .424
9
co =-5 .355
c -20 -3 .OSL=-20 .1
Subgroup
20
O 3.OSL=18.16
c
to
10
c
=5 .557
O
-3 .OSL=0 .000
Figure 5 .7 shows a normal probability plot of the same raw data. This plot
indicates two important results : First, only a single data point in the upper right-
hand corner of the normal probability plot seems to be inconsistent with the
maximum-likelihood straight line that is determined by the statistical analysis
software. From this, one might conclude that only one of the 31 data points
appears to be an outlier. This conclusion is consistent with our results from the
XMR charts . Second, if we look where the normal probability plot maximum
likelihood regression line crosses the 0 point on the X-axis (83%), we estimate
that 17% of product fails to meet specifications . This means our overall manu-
facturing process is not very capable ofconsistently meeting the customer speci-
fication requirements. In other words, we might consider that the outlier or
special cause event is only part of our problem . We have a process that is not
very capable and we need to make one or more process changes if we want to
improve quality and reduce the amount of scrap and rework.
When we went to the manufacturing line, we identified that there was a
special cause problem for the seventeenth data point from this data set, which
later was resolved. So we need to reevaluate the process with this data point
removed. The resulting normal probability plot for this adjusted set of data is
shown in Figure 5.8. This plot shows a much better fit, where we would now
conclude that about 10% of overall product from the manufacturing process is
beyond the specification limit, whenever there are no special cause conditions.
We would nextliketo determine what improvement(s) should bemade within
our process . Using existing data, we could create a Pareto chart of what had
84 CRAFTING INSIGHTFUL METRICS
99 - ML Estimates
Mean: -5.35484
95
StDev. 5.44319
90
80
70
80
50
W 40
0- 30
20
10
5
-10 10
Data
99 ML Estimates
Mean: -5 .90667
95
StDev: 4.60173
90
80
70 ----------------------_L______________1
60 ----------------------------------
50
40
30
20
10
5
-15 -5 0
Data
Figure 5.8 Normal probability plot of data without special cause condition
EXAMPLE 5 .2 : TRACKING AND IMPROVING TIMES FOR CHANGE ORDERS 85
18 2 0 0 0 0 0 0 0 0
0 0 0 14 0 0 7 3 0 41
0 0 0 0 0 0 17 0 0 0
0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 11 0 0 17
26 0 0 0 0 0 21 0 0 0 0
6 17 0 0 0 0 0 0 0
The XmR chart of this data shown in Figure 5.9 does not give much insight
into the process, since there are so many zeros. We should note that an XmR
chart was chosen to analyze this data so that between-sample variability would
be considered within the calculation of the control limits, as described within
Breyfogle (1999) in Implementing Six Sigma (chapter 10).
A histogram of this data indicates that it might be better to consider the
situation as bimodal in nature . That is, we might consider "less than one day
(i.e., 0)" as one distribution and "one day or more" as another distribution. An
estimate of the proportion of ECOs that take "one day or more" to process is 14/
85 = 0.16 . As shown in Figure 5 .10, the XmR control charts of the "one day or
3 .OSL=13 .13
X=2 .3fi5
-3 .OSL=-8 .400
40
a
30
cc
20
c 3 .OSL=13 .22
10
R=4 .048
0 -3 .OSL=0 .000
Figure 5.9 XmR chart of the time taken to process engineering change orders
EXAMPLE 5 .3 : IMPROVING EFFECTIVENESS : EMPLOYEE OPINION SURVEYS 87
more" values do not indicate any evidence of the existence of special causes of
variation. The normal probability plot shown in Figure 5.11 is another way of
viewing the data when ECOs "take one day or more" to process .
We can combine the information contained in these two figures to give an
estimate of the percentage of change orders from the process that take longer
than any specified time period of our choosing . To illustrate this, let's estimate
what percentage of the time ECOs take 10 or more days to process . We have
already estimated that the proportion of the ECOs that take "one day or more"
to process is 14/85 = 0.16. Since we are interested in a target value of 10 or
more days, this ratio of 0.16 applies to our situation. We can use the normal
probability plot in Figure 5.11 to estimate that the proportion of ECOs that take
one day or more to process is 0.66, or 66%,since from the figure 1- 0.34 = 0.66 .
Therefore, the estimated percentage of engineering change orders taking 10
days or longer to process is [0.16] x [0.66] = 0.11, or 11 %.
With this information we could categorize the characteristics of change or-
ders that take a long period of time. We could also create a Pareto chart that
would give us a visual representation that leads to focus areas for process im-
provement efforts . Also, we could calculate the monetary implications of late
change orders as they impact production and/or development deliveries.
Many employee opinion surveys are conducted periodically for example, an-
nually or semiannually. These surveys typically involve the use of a Likert scale
with associated numerical responses, such as the following :
Strongly agree-5
Agree-4
Uncertain -3
Disagree-2
Strongly disagree-1
Much time and effort can be spent trying to understand the meaning of the
results. And even after considerable analysis, we might question the value of
the survey. When interpreting the results from the survey, for example, one
might ask several questions :
" Is the rating scale appropriate for what we are trying to accomplish?
" Are there any significant positive or negative trends evidenced by the data?
Figure 5.10 XmR chart of nonzero values
EXAMPLE 5 .3 : IMPROVING EFFECTIVENESS : EMPLOYEE OPINION SURVEYS 89
99
ML Esfimates
Mean: 14.3571
95 -I
StDev : 10.4171
90 -I
80
70 _n------- r------------- _~___-__I
60
50 ------- ------ ------ -------
1
40 _ _ 1_
30 _ __--____=y--_______~..______~
.
, 1 ,
-10 0 10 20 30 40
Data
Might the results be different if the survey were taken at a different time
for example, right after the release of performance appraisal data to em-
ployees, after the U.S. Internal Revenue Service income tax filing deadline
ofApril 15, during the Christmas holidays, and so forth?
" What could or should be done differently during the next survey cycle?
Let us examine what might be done differently so that survey design and
analysis will focus on gaining insight into company internal processes as op-
posed, for example, to ranking individuals and/or departments and then taking
"appropriate corrective action as required."
The use of the Likert scale with integer responses on a 1-to-5 rating scale
and with distinct upper and lower bounds can lead to interpretation problems.
For example, if most or all employees rate one or more of the survey questions
at a 5 level (i.e., strongly agree), it will be virtually impossible to measure
subsequent improvement. This issue could be addressed by asking respondents
to answer the question(s) relative to changes from the previous year. This would
involve modifying the rating scale to:
Significant improvement-5
Some improvement--4
No change- 3.
Some decline-2
Significant decline-1
90 CRAFTING INSIGHTFUL METRICS
Since there are no upper bounds with this approach, there is much more
flexibility. For example, management could set a stretch goal, consistent from
year to year, of 3 .5 or greater average response, with 80% believing improve-
ment has not taken a turn for the worse, which equates to a score less than 3.0.
Because of the phrasing ofthe question, only those who had been employed for
one year or more could serve as respondents . This mightrequire the addition of
a sixth response category, such as "not employed last year" or "not applicable."
To address the bias in the results due to "noise" such as weather and fear of
layoff, we could randomly divide the workforce into 12 separate groups and
assign each group to be surveyed in a different month of the year. A control
chart could then be used to evaluate overall and specific question responses .
Other Six Sigma tools, such as normal probability plotting and analysis ofmeans
(ANOM), can give insight into variability and other areas.
Valuable information for improvement opportunities is often contained in
the "comments" section of the survey. However, it can be very difficult, if not
impossible, to classify and rank-order these suggested improvements without
also being able to ask respondents how much they value the proposed changes .
To address this concern, consider adding a section that lists improvement ideas.
Each person can vote on a certain number of listed ideas. This information can
then be presented in a Pareto chart format. Ideas can come from previous years'
write-in comments or they can be created ahead of time by a designated com-
mittee of cross-functional experts presumed to be sensitive to employee needs
and concerns .
Although it makes survey design, execution, and analysis considerably more
difficult, consider having different lists of improvement ideas fordifferent groups
within the organization . For example, the administrative staff may be highly
satisfied with their computer workstations, whereas the engineering group is
clamoring for a supercomputer to conduct complex finite-element analysis and
three-dimensional modeling ofcomplex structures . Asingle question on "ready
access to required information technologies" could lead to considerable confu-
sion when attempting to explain strong differences ofopinion between the two
groups of workers .
Invoices were selected randomly over time from an accounts receivable depart-
ment. Figure 5 .12 shows a histogram of the number ofdays that these invoices
were overdue in payment. Negative numbers indicate that payments were made
before the due date. It is quite evident that the data are not normally distributed,
since the distribution is skewed to the right. We could try to use a three-param-
eter Weibull distribution or Box-Cox Transformation (Box et al., 1978) to bet-
EXAMPLE 5 .4 : TRACKING AND REDUCING OVERDUE ACCOUNTS PAYABLE 91
ter describe the data. It should be noted that for this study the measurement was
changed from attribute to a continuous measurement relative to due date. Con-
tinuous response data are preferable to attribute data, such as a yes or no re-
sponse, that address whether invoices have been paid on time or not. With
continuous data we can obtain more information about a process with less data.
In addition, we can quantify the amount of time overdue accounts are delin-
quent, which can yield a better monetary impact for the current situation .
When collecting the "days overdue" data, other potentially useful variables
were simultaneously recorded. Variables that were collected for this example
included company invoiced, amount of invoice, whether the invoice had errors
upon submittal, and whether any payment incentives were offered to the cus-
tomer. This supplemental data permits other kinds of useful analyses . For ex-
ample, by knowing the monetary amount of the invoice and the days overdue, it
was possible to calculate the time value of money thecompany was forgoing by
not being paid on time. This number could then be used as a COPQ baseline for
justifying an S4 project.
Analysis of variance (ANOVA) and analysis ofmeans (ANOM) studies in-
dicated that the monetary amount of the invoice did not significantly affect the
delinquency. It did reveal, however, that certain specific companies were con-
sistently more delinquent than others . Also, invoices were frequently rejected
100
50
Figure 5.12 " Histogram showing the number of days invoices are delinquent
92 CRAFTING INSIGHTFUL METRICS
In recent- years, much has been written about performance and performance
measurement. Increasingly, organizations are hiring consulting firms to help
with these performance measures. Six Sigma project selection should take these
metrics into account. However, we must keep in mind that some performance
measures can lead to short-term activities that are not beneficial. In this chapter
we discuss measurement types, the principle of measurement, and the balanced
scorecard. These topics can be beneficial when executing projects .
Thor (1988) has found that there are three main reasons for measuring things in
organizations: to plan, to screen, and to control . Planning measures, normally
the domain of senior executives, attempt to answer thequestion "Are we achiev-
ing our long'-term strategic goals?" They are typically expressed in the lan-
guage of executives, which is money and ROI . Therefore, it is appropriate that
one Six Sigma category of measurement is planning. Because planning mea-
sures focus on long-range performance issues that tend to change slowly, they
may be tabulated infrequently, perhaps once or twice a year.
Screening measures address the question "Are the functional areas perform-
ing in ways that support the organization's strategic goals?" They are expressed
in both monetary and nonmonetary terms . Their focus tends to be on short- to
intermediate-range performance ; hence, they may be measured monthly or quar-
94 PERFORMANCE MEASUREMENT
Quality planning
Team processes
Education and training
Recognition and reward
Communications
Measurement
Know why measurements are made and how they will be used
" Measure only what is important
PRINCIPLES OF MEASUREMENT 95
Know why measurements are made. The use of measures to control ma-
chines, processes, and products, for example, can be beneficial. However, per-
formance measures to control individuals can be counterproductive . It should
be noted that control measures are intended to "lock in" the quality gains that
have been achieved following changes in a product or process . We suggest
using high-level performance not as a control metric but as one that helps with
project selection and the wise implementation of Six Sigma.
Measure only what is important. We say that financial metrics should be
considered as part ofthe infrastructure when selecting Six Sigma projects . The
success of S4 projects relative to financial measures can breed upon itself in
such a way that more S4 projects are created, leading to additional successes .
Also, excessive measurements can be expensive both in the time and money
needed to collect data, and in the forgone resources when the data is responded
to in an inefficient manner. To address this situation, we suggest giving empha-
sis to identifying and tracking only KPIVs and KPOVs forprocesses . GE made
reference to these terms as X's (i.e., KPIVs) and critical-to-quality (CTQ char-
acteristics, or Y's (i.e., KPOVs), where CTQs are the items essential for cus-
tomer satisfaction in a product or service and KPIVs are the key drivers .
Measure causes (drivers) ofgood performance. Historically the most im-
portant metrics have been financial measures such as return on investment (ROI),
net profit, price to earnings (P/E) ratio, and stock price. These measures are
outcomes, not drivers ; they are effects, not causes . Six Sigma can guide organi-
zations through the determination, measurement, and improvement of the re-
sponse of KPOVs that affect financial measures . Six Sigma can then give
direction on the determination and control of KPIVs that affect these KPOVs.
Use a family ofmeasures. The "family of measures" concept is simply the
belief that performance is multidimensional. Most processes worth measuring
are too complex to be assessed using a single variable. A single measure of
success can lead to the optimizing of one factor at the expense of others . The
American Productivity and Quality Center (1988) asserts that measuring the
three variables of productivity, quality, and timeliness captures the essence of
group performance for knowledge workers . If multiple performance measures
are chosen with skill, any false improvement recorded as aresult of focusing on
only one of the variables will be neutralized or minimized by a corresponding
loss of achievement in a different variable from the same family.
96 PERFORMANCE MEASUREMENT
service resulting from data entry errors . If one were to plot these monetary
values over a certain time frame, the data could be deceiving unless an allow-
ance were made for inflation .
In today's business climate, the term "balanced scorecard" can refer strictly
to the categories originally defined by Kaplan and Norton (1996), or it can refer
to the more general "family of measures" approach involving other categories .
General Electric, for example, uses the balanced scorecard approach but devi-
ates from the four prescribed categories of the balanced scorecard when it is
appropriate . Godfrey (December, 1999) makes no demands on the balanced
scorecard categories other than that they track goals that support the
organization's strategic plan.
PART 3
Most of the Six Sigma statistical tools and methodologies are not new. From an
organization's point of view, the traditional approach to the deployment of sta-
tistical tools has not been very effective. Most complex statistical analysis has
been left to an internal or external statistical consultant or has been part of a
segregated quality department.
An engineer may approach a statistical consultant with no real understand-
ing of how he can benefit from the wise use of statistical techniques . The statis-
tician should first learn the technical aspects of the dilemma presented by the
engineer in order to give the best possible assistance. However, most statisti-
cians do not have the time, background, or desire to understand all the en-
gineering dilemmas within their corporate structure. To be effective, therefore,
engineers need to have some basic knowledge of statistical concepts so that
they can first identify an application of these concepts and then solicit help, if
needed.
Easy-to-use computer software has made the process of statistical analysis
readily available to a larger group of people. However, if the practitioner does
not have a sound understanding of the assumptions and limitations of a statisti-
cal methodology, computer-derived models can be erroneous or misinterpreted.
Also, even when an engineer has recourse to such software, the issue of prob-
lem definition and the application of statistical techniques still exists.
When the Six Sigmabusiness strategy is followed and combined with the
wise use of metrics and statistical tools, the strategy can result in dramatic im-
provements to an organization's bottom line. In this chapter and the others in
102 DEPLOYMENT ALTERNATIVES
Often organizations do not look at their problems as the result of current pro-
cess conditions. However, if they did, their problems could often be described
in terms similar to those in Figure 7.1 . They could determine key process out-
put variables (KPOVs) of their process such as a critical dimension, overall
cycle time, DPMO rate, and customer satisfaction. By sampling this output
over time at a frequency that gives them a "30,000-foot level" view, they could
gain insight into how KPOVs are changing, detect trends in the process, and
distinguish between common and special cause variation . This data could then
be transformed and combined with data on customer needs to get a picture of
the accuracy and precision of the overall process .
Instead of looking at their situation as a process, organizations often react
over time to the up-and-down movements oftheir products, reacting to KPOV
levels in a "firefighting" mode and "fixing" the problems of the day. They often
have multiple "process improvement" projects working simultaneously. Not
realizing that "noise" (material differences, operator-to-operator differences,
machine-to-machine differences, and measurement imprecision) can impact a
KPOV to a degree that results in a large nonconforming proportion, organiza-
tions may make frequent, arbitrary tweaks to controllable process variables .
Practitioners and management might even think that this type of activity is help-
ing to improve the system. However, in reality they may be spending a lot of
resources without making any process improvements. Unless holistic process
changes are made, the proportion of noncompliance, as shown in Figure 7.1,
will remain approximately the same.
Organizations that frequently encounter this type of situation have much to
gain from implementing the S4 business strategy. Often, when organizations
assess their Six Sigma needs, they do not give adequate attention to the costs of
not making a change. The concept of "the cost of doing nothing,". described
earlier in this book, addresses these costs and leads to a better appreciation of
the potential gain from the implementation of Six Sigma.
DEPLOYMENT OF SIX SIGMA: PRIORITIZED PROJECTS 103
noise
conhilable n~io se contollable
noise noise variable variable noise
Rework
A,
Start B c
no contollabie-
Rework Sq.a
.variable noise
11
noise
no
Customer
needs
Frequency of
non-conformance
Figure 7.1 Example of a process with akey process output variable [Reproduced from
Breyfogle (1999), with permission]
As illustrated in Figure 7.2, the strategy after assessment and kickoff con-
sists of a series of phases :
" Deployment phase, which includes the Define Phase for projects : deter-
mining what is important to the customer
" Measure Phase : understanding the process
104 DEPLOYMENT ALTERNATIVES
Project Implementation
ease vase
s c
noise
n oise
Bottom-line Benefits
Figure 7.3 Example of process improvement and impact to a key process output
variable [Reproduced from Breyfogle (1999), with permission]
106 DEPLOYMENT ALTERNATIVES
Traditional Approach
" Offers bigger impact through projects tied with bottom-line results
" Utilizes tools in a more focused and productive way
" Provides a process/strategy' for project management that can be studied
and improved
" Increases communication between management and practitioners via
project presentations
Facilitates a detailedunderstanding of critical business processes
Gives employees and management views of how statistical tools can be of
significant value to organizations
Deploying Six Sigma through projects instead of through tools is the most
effective way for an organization to benefit from its investment in Six Sigma.
$4 Approach
however, the strategy and/or training material do not match the needs of the
organization. When deciding which Six Sigma provider to partner with, we
think that it is essential that decision makers view one day of actual Black Belt
training from each provider under consideration. During this time these deci-
sion makers can also talk with current Black"Belt candidates to get firsthand
experience about the Six Sigma provider.
We also suggest that the following questions be presented to Six Sigma pro-
viders who are being considered:
date)?
During Six Sigma training do you use multimedia equipment with a pre-
sentation program such as PowerPoint, or do you use transparencies only?
What is the basic format of your Six Sigma course for executive training
and champion training?
How do you address business and service processes?
ESSENTIAL ELEMENTS OF AN S4 IMPLEMENTATION PLAN 1o9
" What topics do you cover within your Six Sigma workshops?
" What is the format of the handout material?
" Is licensing of instructional material available for use by our internal in-
structors
Is the material in a form that others can easily teach from?
" Have all the courses that your organization has sponsored used your mate-
rial (some providers subcontract much of their workshop instruction to
other organization that use their own material)?
Describe the frequency of exercises (i.e., manual, computer software, and
other hands-on exercises) within the Six Sigma workshop.
How do you address application of the techniques to real-world situa-
tions?
Are attendees required to have computers?
" what software do you use? (We suggest using a general-purpose package
such as Minitab .)
Is the software easy to use and learn?
What have others said about your training/consulting in the application of
Six Sigma? Can they be contacted?
What is the experience level of the person(s) responsible for developing
the courseware?
" What companies have you helped successfully implement Six Sigma?
Applying Six Sigma so that bottom-line benefits are significant and lasting
change results requires a well-thought-out and detailed plan. Once a Six Sigma
provider is selected, it should assist the organization with the development of
an effective implementation plan. Figure 7.6 shows a sample implementation
schedule for the kickoff of a Six Sigma plan. Although every plan is unique, it
should consider the following essential elements, which are described in detail
in subsequent chapters:
January February March April May June July Augu
ID Task - Name 2 9 162330 6 11312012715 112119 26 2 9 16 23 30 7 14 21 28 4 11 18 25 2 9 16 23 30 6 13
1 Choose a Six Sigma Provider
2 - Executive Training
3 Champion training
Select 84 Projects
6 Six Sigma Training: Measurement
Project Report-outs
Although difficult to schedule, ongoing top-down and down-up communica-
tions are also essential elements of a deployment plan. Communication plans
112 DEPLOYMENT ALTERNATIVES
are often the first thing to get overlooked on a busy manager's schedule but are
essential to break down barriers between management and practitioners .
It is important that executives ask the right questions that lead to the right
activity, sometimes referred to as the "five whys." This is an informal tool that
management can use to help practitioners get at the root cause of the defect. By
continuing to ask "why" after the team develops solutions for the problem, one
can help practitioners identify areas that need more focus or analysis . It will not
always take five iterations of "why does this defect occur" to get at the root
cause. Leaders must take care when using this approach so that it comes across
as a mentoring tool instead of a tool used to judge the effectiveness ofthe team.
Questions should always be focused on understanding the process better, not
judging the team's analysis of the process .
The quality community has complained for years that management has not
listened and supported their qualityinitiatives . To gain the awareness of upper-
level management to the value of statistical analysis, effective project reports
are needed that are scheduled up front. Quality professionals often need to do a
better job communicating to executives in clear language. Project report-outs
should tell a simple story, using easy-to-understand graphs that highlight the
root cause of the problem .
Project report-outs should also be structured up front. Management needs to
define the frequency and expected content of these reports . Project report-outs
serve the purpose of supplying periodic feedback on the status of the project,
which will aid in its timely execution. Initially, consider report-outs after each
phase in order for management and the attendees to learn the methodology
together. Later, one might consider streamlining the process . General Electric
went from reporting out after each phase to reporting out before any project
improvements were made and upon completion of the project . Also, as part of
an overall communication plan, results on successful projects should be posted
for all to see; this will serve to generate enthusiasm for Six Sigma by showing
its possibilities .
Postmortem
Successful deployment of Six Sigma is best achieved in a series of waves, each
wavefocusing on strategic change areas. Between waves, there is time for evalu-
ating effectiveness, compiling lessons learned, and integrating improvements
into the infrastructure . When a Six Sigma infrastructure is created in waves,
one can take time between waves to assess progress and incorporate critical
lessons learned for lasting change.
At General Electric, when transitioning from wave 1 into wave 2 during the
implementation of Six Sigma, "postmortem" discussions were conducted. Sub-
sequent brainstorming and planning sessions led to the integration of the fol-
lowing valuable lessons learned in the deployment strategy:
ESSENTIAL ELEMENTS OF AN S 4 IMPLEMENTATION PLAN 113
" Master Black Belts were aligned with strategic business units (SBUs) and
worked togetherto leverage projects and create strategic focus areas.
The number of Black Belts and Green Belts was significantly increased .
Mentoring relationships were established between Black Belts from wave
1 and new Green Belts/Black Belts.
The Define Phase was added to the Measure-Analyze-Improve-Control
deployment training to assist the team in scoping manageable-sized
projects .
" Project report-outs were more structured and less frequent.
" Larger-theme projects were developed that focused on key areas, with
numerous structuredprojects relating to the theme being addressed simul-
taneously.
" An annual Six Sigma week was held that allowed quality practitioners to
share the past year's successes; communicate lessons learned, and elect
the project of the year.
Training was made mandatory for promotion.
CREATING A SUCCESSFUL
SIX SIGMA INFRASTRUCTURE
Executive leadership
" Customer focus
" Strategic goals
116 CREATING A SUCCESSFUL SIX SIGMA INFRASTRUCTURE
Black Belt
Selection Deliver Results
215 1111
Metrics
Resources
213
412
Planning
114
Training and
ommunications
112 Execution
412
Figure 8.1 An interrelationship digraph assessing Six Sigma success factors and
their interrelationship
Project selection
" Training and execution
" Resources
Black Belt selection
Metrics and feedback
" Culture
Communications
" Planning
Results
An ID for any one company may produce an order different from above;
however, many of the factors will be like those listed. This chapter discusses
the critical success factors depicted in Figure 8.1, excluding the factors out-
lined in bold, "Project Selection" and "Training and Execution," which are cov-
ered in other chapters . This chapter also includes a sample implementation
schedule and lessons learned from other companies that have created success-
ful Six Sigma. infrastructures. .
EXECUTIVE LEADERSHIP 117
Focusing on the needs of customers goes hand inhand with creating a success-
ful Six Sigma infrastructure. The factorsthat are important to customers are a
necessary input for the process improvement team's true success . Therefore,
evaluating the customer's perception of quality shouldbe at the forefront of an
implementation process and a foundation block of an infrastructure .
Complaints from customers should be viewed as an opportunity for growth
and increased market share. These complaints should serve as a spotlight on
areas needing process improvement . The key to success in this initial step is to
make it easy for customer input to be heard. There are various methods to do
this, including the following :
Customer surveys can be an easy and effective means of inserting the voice
of the customer into the project-selection process (an employee opinion survey
was described in Example 5.3) . When General Electric initiated its Six Sigma
strategy, each division performed intense customer service surveys . Key cus-
tomers were asked to rate GE on a scale of 1 to 5 on several critical-to-quality
issues . GE also asked its customers to rate the best in class in the same catego-
ries, measuring a defect as anything less than 4, and tracking customer satisfac-
tion on a quarterly basis. Areas with low scores were viewed as potential Six
Sigma projects.
The following are general process steps for conducting effective surveys
within an organization (Folkman, 1998):
a o a
Please mark an X in the column that most closely
matches your view for each of the statements below : °~ 1 0 ~ ° an
v~A A Z v5~
1 . Book organization was logical and cohesive . 0 0 (0
2 . Glossary at the end of the book was helpful .
3 . My organization will use the book as a
guide to implementing Six Sigma.
4 . "Frequently Asked Questions" section was
useful.
5 . Explanation of Six Sigma in statistical terms
was understandable.
6 . Satisfied with the purchase of this book .
7 . Development applications of Six Sigma were
explained well.
8 . Cross-references to Breyfogle (1999),
Implementing Six Sigma, were useful .
9 . Length of the book was appropriate for
managers and executives.
10. Written text was concise and informative.
11 . Service/Transactional applications of Six
Sigma were explained well.
12. Education and training requirements of Six
Sigma were adequately covered.
13 . Manufacturing applications of Six Sigma were
explained well .
14. Relationship of Six Sigma with other quality
systems was insightful. 0 171
15 . Index at the end of the book made it easy to
find key Six Sigma topics . 171 173
16. Figures and tables were clear and useful in
explaining text information.
17 . Key Six Sigma roles for people were clearly
defined and differentiated.
18 . Flow of topics within the book was good. 173
19 . Infrastructure requirements for a Six Sigma
strategy were well articulated .
20 . The project integration of the tools and
examples in the application sections were
explained well and serve as good references
to project execution .
Comments/Improvement Recommendations:
Occupation :
Please send completed copy to : Smarter Solutions, Inc ., 1779 Wells Branch Parkway, #110B-281, Austin, TX
78728. Or complete an online version at www.smartersolutions.com.
STRATEGIC GOALS 121
Learning through customer feedback what works and what doesn't will es-
tablish the mind-set of continual process improvement within an organization.
Jack welch himself has been quoted as saying that a business strategy alone
will not generate higher quality throughout an organization. Depending on the
size of an organization and its core values, the word customer can take on many
different definitions . When collecting feedback, care should be taken to main-
tain a comprehensive view of customers . By combining external feedback with
such things as internal business strategies, employee needs, and government
regulations, organizations can obtain a balanced list of customer needs .
Once the needs of key customers are collected and analyzed, a method is re-
quired to transform them into strategic goals for the organization . At this stage
in the implementation process, quality function deployment (QFD), or the
"House ofQuality," is a useful tool to drill down from customer needs to strate-
gic focus areas that have abigger impact on the bottom line of a company.
Figure 8.2 summarizes the customer satisfaction House of Quality process .
The overall approach is an iterative process in which a cross-functional team
completes a series of "houses" via the following high-level steps or guidelines:
5. The process continues until the ranked "hows" are specific enough to
assign strategic goals and measurements . The subsequent further scoping
of the final "hows" into S4 Improvement projects is described in Chap-
ter 10 of this book, under "Project Selection."
Upon the successful completion of this process, there will be a list of key
focus areas, which can be assigned strategic goals and objective measures that
serve as S4 project opportunities . This process provides confidence that project
resources are focused on meeting critical needs of the business and the needs of
customers .
The following are tips to consider when creating a customer satisfaction
House of Quality :
The more specific the outputs from customer surveys, the less time the
QFD process will take.
Combine customer input with internal business strategies, employee needs,
and government regulations to obtain a comprehensive list of customer
needs.
Consider first organizing all the "whats" into the categories with similar
detail, and then perform the iterative process, inserting new "whats" at the
appropriate "house" level.
Bias is easily injected into the process through the wording of survey ques-
tions or the choice of team members.
The process can require a lot of time and resources to conduct .
Hold separate meetings for the brainstorming sessions so the output is
meaningful and participants don't feel rushed.
Sometimes it takes many repetitions of the process to achieve meaningful
results .
RANKED y STRATEGIC
HOWS r"V GOALS
0-
Figure 8.2 Customer satisfaction House of Quality: Relating customer needs to critical business processes
12 4 CREATING A SUCCESSFUL SIX SIGMA INFRASTRUCTURE
8.4 RESOURCES
The methodology of Six Sigma needs to be combinedwith the right people for
real results (Harry, 2000) . GE has had much success with the deployment of ix
Sigma techniques through individuals called Black Belts, who work full-time
on executing Six Sigma projects (Slater,1999) . These individuals are the project
leaders, but they cannot do the work of Six Sigma alone.
For process teams to achieve results that are meaningful, the priorities of
team members need to be aligned to the strategic goals of the projects . Manage-
ment needs to create a supportive environment and realign resources to priori-
ties when needed. Listed below are the roles and responsibilities of a well-staffed
team as it existed within GE. Smaller companies may need to combine roles if
resources are limited . Also, larger project themes may require additional
resources.
Facilitate lessons learned session between Black Belts and other Master
Black Belts
Motivate others toward a common vision
Coordinate activities to drive project completion
Participate in multiple projects
Remove barriers to success
Aid Black Belts in formulating effective presentations to upper manage-
ment
Approve completed projects
Become certified as a Master Black Belt
Team Member
$BU #2
Champion
Black Belt I I Black Belt I I Black Belt I I Black Belt I I Black Belt I I Black Belt
Quality Leader
B
Master Black Belt I I Master Black Belt I I Master Black Belt
Black/Green Belt Black/Green Belt Black/Green Belt BIac klGreen Belt BlackfGreenBelt Black/Green Belt
Team Members Team Members Team Members Team Members Team Members Team Members
tionships of an effective Six Sigma organization within GE. In this model, the
company appointed an overall quality leader to drive program effectiveness .
Master Black Belts report to the quality leader and are responsible for multiple
Black Belts or Green Belts within specific departments or functional areas.
Champions, Sponsors, Master Black Belts, and Black Belts are the leaders
of Six Sigma change, and care should be taken to select the right players . How-
ever, having dedicated teammembers isjust as important as a well-trained leader.
Free up time and resources for improvement projects team members should
not feel overburdened by their Six Sigma responsibilities . Black Belts need a
supportive environment that prioritizes their projects . Up-front work needs to
be done within the team to communicate expectations and establish how to
meet them. For additional information, refer to the section "Create ProjectCharter
and Overall Plan" in Chapter 10.
People selected to become Black Belts need to be respected for their ability to
lead change, effectively analyze a problem, facilitate a team, and successfully
manage a project.
8.5 METRICS
Teammembers
Key process output variables
Key'process input variables
Estimated completion date by phase
Actual completion date by phase
Six Sigma tools utilized in the project
Open action items
Defect definition
" Baseline measurements
130 CREATING A SUCCESSFUL SIX SIGMA INFRASTRUCTURE
From this database, quality leaders were able to glean insight into the effec-
tiveness of Six Sigma.
In general, metrics need to capture the right activity and be designed to give
immediate feedback. Successful metrics focus on the process rather than the
product or individuals . If employees are forced to enter additional information
and don't immediately see how this information is utilized, the quality of re-
ported data in all likelihood will drop dramatically.
8.6 CULTURE
The launching of Six Sigma affords an opportunity to assess the current culture
of an organization . Consider the following questions :
" How has your company historically dealt with change initiatives?
" Does your company make consistent changes that don't last?
" How effective are your project teams?
" Are you frequently focusing on the same problem?
" How do youremployees attack problems when conducting their daily work?
What is required within your company culture to make continual process
improvement a lasting change?
" What will prevent your company from achieving success with Six Sigma?
(+)RIVERS
{-)RESTRAINTS
LACK OF FOLLOW-
CREA TI VE f THROUGH
MOTIVATED .1 OVERBURDENED
The next step is for a company to create strategic plans to overcome their
restraining farces and enhance their drivers . Listed below are examples of com-
pany cultures that exhibit some of the key drivers and restraints listed in
Figure 8.4:
Creativity
Motivation
Most companies that readily adapt to change have a high level of motivation
throughout their workforce. Employees are personally motivated for different
reasons, but most employees are motivated in a professional context by the
prospect of reward. Remember, reward does not necessarily mean money . A
simple, personal message of sincere thanks can sometimes be more effective
than a check in the mail . What motivates an employee to risk change? The
answer may be different for different departments within acorporation, and can
vary from time off to verbal recognition or from new equipment to a weekend
getaway. However, all goodreward systems are fair and honest and have follow-
through .
At GE, motivation was a key driver to success with Six Sigma. In 1995, Jack
Welch sent a memo to his senior managers stating that every employee must
have begun Six Sigma training to be promoted, leading the Six Sigma charge
by connecting quality to compensation (Slater, 1999). To reinforce this charge,
he told top executives that 40% of their bonuses would be tied to the successful
introduction of Six Sigma. The memo launched the new quality initiative with
a fierce momentum . A direct connection between outcome and reward was es-
tablished; in other words, reward was tied to achievement of breakthrough im-
provements . GE was consistent with rewarding desired outcomes, and itbrought
the company true success in its Six Sigma initiative. After a small net loss on
GE's initial investment, the company's payoffin a short period oftime reached
more than $750 million in 1998.
Empowered
Powerful results come from synergistic teams. Without common purpose and
shared methods on how to work together, the output of the team will merely be
a sum of the outputs of the individual contributors .
In his book Managing Upside Down, Tom Chappell (1999) describes how
his family-run natural personal-care company (Tom's ofMaine) achieved rapid
growth and profitability by "managing upside down." Mr. Chappell invented
the idea of "acorns," which were small, flexible teams of no more than three
people responsible for new idea generation, market research, and scientific in-
vestigation . The company had previously gone three years without the intro-
duction of a new product ; however, once acorns were invented, the company
increased its product line from 27 to 117 in two years, proving it could create
profitable products by empowering people to think and act creatively.
Frequently, the failure of a Six Sigma project can be attributed to the inabil-
ity of the company to embrace and sustain change . The only effective way to
stimulate change is to link immediate and unbearable pain to our old behaviors
while linking immediate and pleasurable sensations to new ones (Robbins,1991).
The challenge ofleaders beginning the Six Sigmajourney is to create a culture
that is able to absorb the change required to continuously improve .
8.7 COMMUNICATIONS
Typically, company leaders implement Six Sigma because they possess a clear
vision of what their company can achieve . Frequently, however, they do not
realize the power of effectively communicating this vision throughout the cor-
poration. As mentioned in chapter 7, traditional approaches to deploying statis-
tical methods often fail because of communication issues between management
and employees . Often, managers do not ask the right questions to strategically
focus the application of statistical methods on process improvement instead of
firefighting.
To avoid this situation, executives need to get everyone engaged and speak-
ing the language of Six Sigma. A shared vision of how Six Sigma fits he stra-
tegic needs ofthe business should be created. Awell-thought-out vision has the
power to move people and create shared values . Communicating this vision
effectively is an ongoing process and is integral to a well-thought-out Six Sigma
implementation plan. The Four C's of effective communication plans are as
follows :
Concise
Words should be carefully chosen to communicate the specific goals of the Six
Sigma business strategy . These words should become the common language
for Six Sigma activities.
Consistent
Communication should be sincere and consistent thatis, it should repeat the
same message.
Complete
Communication plans should reach all levels of the organization . If necessary,
consider the use of different media. A brainstorming session can help to deter-
mine the most effective communication techniques for different groups . One
should not underestimate the power of communicating one-on-one, especially
with team members working on the shop floor.
LESSONS LEARNED 135
Creative
A creative method is needed to acquaint employees with the Six Sigma mission.
The first Black Belts trained at GE were asked to define what Six Sigma was
and what it meant to the organization . In a facilitated session, each leader re-
created a make-believe one-minute elevator ride with Jack Welch. During the
ride, Jack asked what Six Sigma was, what it meant to them, and its importance
to the company.
At first, many thought the exercise to be a silly waste of time. But once a few
Black Belts went through the process, they realized it was difficult to commu-
nicate what Six Sigma meant in a meaningful and concise fashion . After spend-
ing time in smaller groups and brainstorming definitions, the Black Belts went
through the process again . This time they produced clear, well-thought-out an-
swers that were personally meaningful and easily repeated to other team
members .
Communication plans should be carefully considered and executed with
enthusiasm. If successful, these plans will be the biggest ally in key stakeholder
buy-in. On the other side of the coin, practitioners should be able to construct
project reports that speak the language of executives and tell a concise, under-
standable story. Both employees and management need to make sure that they
are asking questions thatsupport finding the rootcauses onevery critical project.
Chapter 10 of this book, "Project Selection, Sizing, and Other Techniques,"
discusses this topic further.
The success of Six Sigma depends on the existence of a solid infrastructure and
executive leadership that supports the commitment. The roles ofthe Six Sigma
champions will quickly transform into "watchdogs of quality actions" ifexecu-
tive energy is not infused into the Six Sigma infrastructure . It is crucial that
project champions express enthusiasm and belief in Six .Sigma . Leaders need to
take personal responsibility for driving Six Sigma efforts, including participat-
ing in and even running Six Sigma projects . True leaders show belief in a vision
that then becomes believable to their employees . True leaders show their em-
ployees the future and, through their own actions, motivate them to achieve
goals.
Six Sigma was continually communicated at GE as "the way we do busi-
ness" instead ofjust another initiative (Slater, 1999) . Jack Welch's goal was to
have the Six Sigma methodology embraced within all daily activities. He knew
that for this company-wide initiative to be a success, everyone had to be in-
volved. He pulled out the slogans and banners in full force, then closely moni-
tored the situation and provided meaningful follow-up when necessary. The
136 CREATING A SUCCESSFUL SIX SIGMA INFRASTRUCTURE
following are lessons learned from GE's executive leaders responsible for imple-
menting the Six Sigma initiative (Slater, 1999):
When assigning strategic goals within Six Sigma, it is important that the
goals focus on meeting the comprehensive needs of customers . These goals
must also be SMART: Simple, Measurable, Agreed to, Reasonable, and Time
based. Also, a few carefully selected projects that meet certain initiatives will
bring more respect and ultimately more buy-in than a shotgun approach toproject
selection . With respect to strategic goals, Robert K. Phelan, Senior Vice Presi-
dent of Quality Programs at GE American Communications, shares the follow-
ing lesson : "We experienced false starts when we engaged all employees [in
Six Sigma practices] quickly. We quickly reviewed every project and made
sure they were aligned with our business objectives" (Aviation, 1998) .
As mentioned earlier, many times a project fails because a team is not em-
powered and adequately supported. Thejobs ofSix Sigma leaders are infinitely
more difficult if teams do not have the correct balance of structure, expertise,
and resources . However, companies can go overboard with appointing Black
Belts. Making every employee either a full-time or part-time Black Belt dimin-
ishes the power of the program . If everybody is required to lead a project, that
leaves few resources for dedicated team members . A rule of thumb is that full-
time Six Sigma practitioners within an organization should number no more
than 2% of the total number of employees . Most important, people who be-
come successful full-time Six Sigma Black Belts have a "fire in their belly" to
take on process improvement and reengineering challenges. Steve Hollans, a
Six Sigma applications expert from AlliedSignal, warns, "An early barrier for
us was selecting some people on the basis oftheir availability instead of people
who were capable of being change agents" (Aviation, 1998) .
At the outset, expect results and develop metrics that lead to the "right"
activity. Next, determine what should be done to obtain insight into the source
LESSONS LEARNED 13 7
Go forward with large goals, but also respect the paths taken by other com-
panies, integrating their most critical lessons learned into a continual process
improvement strategy.
TRAINING AND IMPLEMENTATION
A simple but significant distinction between education and training is that edu-
cation teaches why, training teaches how. Companies need to hire people who
already know why and then empower them with a new vision of how. That new
vision of how is Six Sigma.
As discussed previously in this book, an education and training phase should
be centered around projects. The education component of this endeavor is fo-
cused on executives and managers, while the training portion is designed for
Six Sigma Black Belts. In this chapter we will focus on the training of full-time
Six Sigma Black Belts.
In-depth training for full-time Black Belts is typically delivered atthe rate of
one week of training per month, followed by three weeks of the participant
applying what has been learned to a project within his or her area.
Many tools and techniques should be covered in depth during four weeks of
instruction spread over the four-month training period. Statistical techniques
can include multiple regression, analysis of variance, statistical process con-
trol, Gage repeatability and reproducibility (R&R), process capability and per-
formance capabilitystudies, probability, reliability testing, designed experiments,
response surface methodology, basic quality tools, variance components analy-
sis, mixture experiments, and so forth (Breyfogle, 1999). Many of these topics
are taught in semester-long courses at a college or university, which can make
for an intense four-week course .
To address this, we believe that it is not necessary for trainees to learn de-
tailed theoretical "why" aspects of statistical techniques ; it is more important to
140 TRAINING AND IMPLEMENTATION
focus on the "how" aspects. Also, we believe inunderstanding the basic prin-
ciples behind the tools and the wise application/integration of these tools with
other techniques. The labor-intensive portion of "how" to implement the tools
is then addressed through the use of statistical software such as Minitab . With
this training strategy, much less knowledge of mathematics/statistics is needed
as a prerequisite .
A major goal of S4 training is to explain how statistical and other tools fit
into the overall improvement methodology designed to address strategic issues
within an organization . With this approach, more people have the opportunity
to apply techniques directly within their own area of the business . We should
note that a university or college degree does not imply that a person will be a
successful Black Belt candidate . We have seen many trainees with no univer-
sity or college training have great success implementing Six Sigma.
In truth, training is a lifelong commitment to learning. It is most effective
when acquired "just in time" on the shop floor, in the accounting department, or
in the bullpen of specialists answering technical-support phone lines. Poor de-
livery of training material can turn even the best material into a boring exercise
thatfails to impart useful know-how. Conversely, marvelous trainers are worth-
less if their materials are outdated or are taught at a level beyond the students'
ability to understand. The true test of effective training is not an enthusiastic
student evaluation, but rather the students' subsequent ability to perform new
tasks effectively on thejob.
9.1 STRATEGY
In the training of Six Sigma practitioners, the curriculum needs to provide fre-
quent opportunity to apply Six Sigma tools and techniques . Flexibility, dynam-
ics, and power are added to training sessions when each student has a notebook
computer with a statistical software package such as Minitab installed, along
with a software suite such as Microsoft Office . A very effective teaching tech-
nique is to intermingle computer exercises with presentations of the material
using software such as PowerPoint . This technique allows computer exercises
to utilize examples, exercises, and data sets that are prepared and loaded onto
each student computer before the training sessions begin. For this type of train-
ing, we use the examples and exercises described within Breyfogle (1999),
Implementing Six Sigma, as a foundation .
The material should be organized and delivered in such a way that trainees
can immediately apply the concepts to the project they have been assigned
that is, during the three weeks before a new round ofinstruction begins . During
this on-the-job application of Six Sigma tools and techniques, it is important
for the instructor and other members of the training faculty to provide students
STRATEGY 14 1
It should be noted that Six Sigma tools and techniques are applicable to a
wide variety of situations . Therefore, the same Six Sigma course can be used
for training people involved in manufacturing, product/process development;
service, and transactional processes . However, although many of the same tools
can be used in different projects, the sequence of their application may vary.
This is one reason why some organizations prefer that training sessions be com-
posed of students from the same discipline to facilitate common group
discussions .
We believe it is very beneficial for each student to make brief presentations
to other class members explaining how they have applied Six Sigma tools and
techniques. In our training programs, each student presents, during the one-
week-a-month training sessions, an overview of his project and estimates of
bottom-line savings to the organization as a result ofthe project . These presen-
tations, each limited to about 10 minutes, are scheduled for weeks two, three,
and four of the Six Sigma training. Students are strongly encouraged to use the
instructor's computer/projector system and theirown presentation software . The
instructor may provide writtenfeedback of presentations to help students present
their project results to their internal management.
There are legitimate differences of opinion on the appropriate guidelines to
follow in selecting projects for people training to be Six Sigma Black Belts.
Each organization should establish its own guidelines. Some issues to consider
when determining these guidelines are listed below. More in-depth information
on the project selection process is included in the next chapter.
Projects
Create a process for generating potential S4 projects that includes the person
who will be championing the project. For each S4 project from a list of possible
projects an organization might determine (1) potential cost savings and (2) the
likelihood of project success. The product of these two numbers (cost savings
and likelihood of success) can then be used to rank-order the list of projects .
Draman and Chakravorty (2000) have reported that selecting process-improve-
ment projects with the greatest potential for improving systemthroughput based
on the Theory of Constraints, not just the local quality level, had the greatest
impact on the bottom line. They found this approach superior to three tradi-
tional TQM-based approaches to project selection: (1) processes producing the
highest number of scrap units, (2) processes producing the highest monetary
volume of scrap, and (3) treating all processes equally.
An S4 project should be large enough that its successful completion is sig-
nificant to the organization, but not so large that it cannot be handled within
four to six months . We agree with Jonathan Kozol (Pritchett and Pound, 1997)
that one should "pick battles big enough to matter, small enough to win." When
STRATEGY 143
an ongoing S4 project appears to be too large or too small based on these guide-
lines, rescope the project as soon as possible.
An S4 project is considerably more likely to succeed when there is some-
one from within the student's organization championing the project and who
has responsibility for the process performance of the area that the project im-
pacts. This helps make the project more visible and integratable into the busi-
ness area. Students should review the project periodically with this person and
important members of the process team. Structured project report-outs with
a predefined list of attendees is an effective method of achieving this
integration .
Upon completion, the S4 project should be carefully documented and made
available internally to other interested employees. Consider creating a standard
project report format that can be used by future students . At a minimum, the
format should call for an executive overview, a description of the problem,
baseline metrics before work begins, and a flow diagram showing the process
before and after improvement . Metrics at the end ofthe improvement process,
results of all tools and techniques that contributed to the problem resolution,
and estimated S4 project benefits may also be included. More information on
project report-outs is included in the next chapter.
strategy, we can consider multiple metrics where one of the primary measure-
ments considered should be monetary savings. When a strict ruleof counting
only "hard" money is applied for this consideration, the wrong message can be
sent. Atrue "hard" money advocate would probably not give financial credit to
a Six Sigma project that reduced development cycle time by 25%, since the
savings would be considered "soft ."
Recognition/Certification
Carefully consider the process requirements for certification . A multidimen-
sional approach should be used in lieu of only monetary savings for example,
demonstrated project savings of $100,000 or more. Also consider requiring the
following : (1) a demonstrated proficiency in the wise use of Six Sigma tools
and techniques: (2) the ability to interact appropriately with team members in a
variety ofroles ; and (3) proficiency in documenting all aspects ofproject activity.
Organizations should also consider how participants within an S4 business
strategy will be recognized in nonmonetary terms . The relative emphasis placed
on compensation versus recognition should be looked at carefully.
There are a number of unique roles and responsibilities associated with Six
Sigma. It is naive to think that Six Sigma can develop within an organization
through a massive hiring of people thatare "certified" Six Sigma experts. For
Six Sigma to be successful, practitioners need a supportive infrastructure where
they can build their expertise .
We will start by describing what a typical four-month . Six Sigma training
session looks like; our model includes one week of in-class training during
each of the four monthly periods . This model will serve as a baseline or refer-
AGENDAS FOR SIX SIGMA TRAINING 145
ence point against which to describe and quantify the requirements for other
types of training .
It is possible to move some material from the first two weeks to the last two
weeks, perhaps eliminating some optional topics from the last two weeks as
necessary ("reliability" and "pass/fail functional testing" are considered op-
tional) . The four weeks oftraining may also be reduced to three weeks ifcertain
topics are not consideredapplicable "design of experiments (DOE)" and "mea-
surement systems analysis" may be deemed irrelevant to students from an orga-
nization based on transactional or service processes (keep in mind, however,
that DOE techniques and measurement systems analysis are very applicable to
certain transactional or service processes) . Hahn et al. (1999) provides a list of
key lessons learned at GE relative to statistical tools (see Table 9.1) . Consider
incorporating these lessons into the training curriculum.
Control variability
Ensure integrity of measurement system (Gage R&R)
Key role of designed experiments (and preplanning)
Usefulness of simulation
Importance of reliability
Need to use right tool for the rightjob
Existing databases are inadequate for the job
Don't "oversell" statistical tools
Avoid diversions
146 TRAINING AND IMPLEMENTATION
Executive Training
We recommend that S4 executives receive one to two days of training. This
training not only describes the basics of Six Sigma but also emphasizes the
establishment and management of a Six Sigma infrastructure . Emphasis should
be given to the importance of wisely applying Six Sigma techniques, where
one-on-one dialogue is encouraged to address specific issues and implementa-
tion questions .
There is debate over the merits of in-house training versus training using exter-
nal sources. Organizations often become confused when addressing this topic.
Let us consider some of the issues.
Organizations interested in initiating an S4 business strategy from scratch do
148 TRAINING AND IMPLEMENTATION
not have off-the-shelf materials available to use even if they have qualified
instructors on staff. Most people drastically underestimate the time and cost
associated with developing training materials of the scope and complexity re-
quired for effective Six Sigma training. Developing a world-class course can
take many years of effort. The achievement of an effective S4 business strategy
requires a much higher level of quality in instruction and materials than is pro-
vided by a typical business workshop. The advanced statistical material de-
manded by an S4 approach to process improvement is complex and voluminous .
It is essential that the materials be carefully designed and masterfully delivered.
Most organizations, we believe, should not attempt to create their own Six Sigma
training materials.
If an organization has a "corporate university," with qualified faculty and
curriculum development specialists, it may be feasible to develop the material.
However, for Six Sigma to be successful there is more involved than simply
combining previous statistical process control (SPC), design of experiments
(DOE), and other course material and calling it Six Sigma. Six Sigma courses
should address not only the tools of Six Sigma but also organizational infra-
structure support issues.
Another option is to hire a company to deliver Six Sigma training and coach-
ing. This type of service can be very valuable during the initial stages of a Six
Sigma rollout . Typically, the material can later be licensed for presentation by
a core group of individuals within the company. Six Sigma training or coaching
may initially appear expensive, but it is not nearly as expensive as trying to
accomplish the task without having the resources and timeto do so. A major
portion of the initial investment in a Six Sigma business strategy is the time
people devote to training and implementation . Consequently, it is not wise to
cut corners by attempting to quickly develop Six Sigma in-house training ap-
proaches and material to fulfill training needs. When an initial Six Sigma train-
ing wave is not successful because of poormaterial ortraining, the organization's
future chances of success with Six Sigma are decreased. This does not reflect
well on those who are sponsoring the implementation.
PROJECT SELECTION, SIZING,
AND OTHER TECHNIQUES
149
10PSAOT5R
Determine the Needs
of KeyCustomers and
Strategic Business
Needs
Apply
QFD to Drill Down to
Project Opportunities
Complete Project
Assessment
Matrix
Figure 10 .1
Determine aBaseline
and
Bottom Line
Benefits
Attack the
Low-Hanging Fruit
Project Execution :
Applying the S4
Improvement
low the process map shown in Figure 10.l, each section highlighting important
Different companies have different strategic goals associated with Six Sigma.
Strategy
In a recent article, Jerome Blakeslee (1999) defines the essence of Six Sigma as
"being able to examine and close the gap between what a business process
produces and what customers demand ." He then recommends that the width of
the gap be used to prioritize Six Sigma efforts .
As mentioned in Chapter 8, "Creating a Successful Six Sigma Infrastruc-
ture," one ofthe methods executives or champions can use to begin to prioritize
their Six Sigma efforts is the Customer House of Quality. To progress, Six Sigma
practitioners from specific departments can take the output of the House of
Quality process (i.e., a list of project opportunities) and brainstorm specific
projects that address the strategic needs of the organization .
PROJECT ASSESSMENT 151
Organizations might also consider having within their first wave oftraining
a project filter of "easy to implement." Such a filter will make it more likely
that the first project assigned for training is manageable and has a high likeli-
hood of success . This will provide the Six Sigma practitioner with the experi-
ence and confidence needed to tackle more complex issues . It will also yield
early success stories to share with employees, increasing the company's confi-
dence in the S4 methodology.
Guessworkcan be taken out of the project selection process by using aproject
selection matrix as shown in Table 10 .1, which is similar to the cause-and-
effect matrix described in Breyfogle (1999), Implementing Six Sigma (chapter
13). This sample matrix relates to a company that defined "efficiency of their
purchasing department" as a strategic goal for the first wave of their Six Sigma
implementation . In this example, a Black :Belt with sourcing experience was
assigned to the group . During the first week of the assignment, the Black Belt
and the manager, who was champion for the project, held a brainstorming ses-
sion to determine the project she would take into training . They used a tool
similar to the project-selection matrix and determined a project to focus on,
having established that it would be best to work on a standardized process for
purchasing first.
When creating a similar matrix within your organization, each column of
this matrix should contain the critical criteria your company uses in selecting a
Six Sigma project . The column headings should be predefined, with rankings
appropriate for the needs of the business. It is then up to project sponsors and
Black Belts to apply the project-selection matrix to opportunities in their func-
tional areas and to develop a list of prioritized projects . By focusing on projects
152 PROJECT SELECTION, SIZING, AND OTHER TECHNIQUES
Supports High
Project Complete Strategic Meets Return Motivated Meets
Assessment within 6 Business Customer on Team Data Budget -4 Process
Matrix months Goals Needs Investment Available Available Constraints Outputs
10 10 7 6 -4 Importance
Projects Total
Standardized
Process for
Purchasing 9 9 3 9 3 348
Improve
Customer
Service 9 9 1 9 336
Improve
On-Time
Delivery 9 1 9 308
Reduce
Purchased
Labor and
Material Costs 3 9 3 288
Supplier
Management
Improve
Employee
Training 1 3 3 9 1 214
the success ofthe teams and the company. Managers should assign performance
goals that are demanding but flexible. Teams can then use creativity and be-
come empowered into shaping their own metrics .
Project metrics should focus on measuring the process, not the product. Suc-
cessful project metrics contribute to process knowledge, help to create a baseline
for strategic processes, and are characterized by the following traits :
Customer Focused
The metrics used to evaluate the success of process improvement teams should
relate to customer values, such as product quality, on-time delivery, and cus-
tomer service. Reviewing the project selection House of Quality can be benefi-
cial to a brainstorming session to create customer-centered metrics.
Cross-functional
Traditional measurements are functional; they were established to measure
progress toward goals. When metrics fail using traditional measurements, a
team is apt to forget its goals and return to the old "functional" way ofworking .
Historically, most companies have not implemented cross-functional metrics
except in the area of financial measures . A primary project metric should be
cross-functional, helping teams clearly evaluate how well they are producing a
desired result.
Informational
Metrics should be simple and straightforward, requiring a minimum of ma-
nipulation to become meaningful. Metrics should measure performance across
time so that they can be analyzed for trends and improvement opportunities.
Relative numbers are usually more informative .
Metrics should also create a common language among diverse team mem-
bers. Metrics that are created by teams should have a purpose, buy-in, and em-
powerment . When creating project metrics, one should consider how these
metrics link to key business metrics . Figure 10.2 shows how project selection
metrics at the production and factory levels can relate to key business metrics .
Typically, there is no one metric that fits all the requirements for a particular
situation . Teams might conduct a brainstorming session to decide what metrics
will help them to better their performance. The team can then review these
metrics with executive management to ensure that they are in accord with the
overall strategy of the business .
Care should be exercised in determining what is measured. Metrics should
make sense to all team members and be easily captured and calculated. Metrics
should not be set in stone; rather, they should be periodically revisited for value
added to the process . Some organizations give focus only to the rate of defects
at the end of a process, using a "one-size-fits-all" metric that may not make
Corporate Level
Goals Measures
Profitability % earnings growth
Market share Market share product index
Store presence Linear shelf space r store
Invents control Inventory turns
Setfmans ed teams 1% teams operational
Factory Level
Goals Measures
Revenue rowth % safes growth
Scrap rate over expected
New customer orders I%
% increase
Invent control Inventory turns
On-time shipping % on-time shipments
Self-man ed teams % lo ees certified
Safe OSHA reportables
Figure 10.2 Selecting projects that link to key business metrics [Reproduced from Wise (1999), with permission]
DETERMINE A BASELINE AND BOTTOM-LINE BENEFITS 155
sense. As discussed previously, this can happen when a sigma quality level
metric is used to drive change . Measuring only defect rates or sigma quality
level can lead to playing games with the numbers and ultimately turn a quality
initiative into a bureaucratic game with numbers . In this situation, project lead-
ership can degrade into an effort to make numbers meet artificial goals. With
this approach, real change will be sporadic .
A balanced scorecard approach can be used within a Six Sigma business
strategy. The balanced scorecard can address issues of quality, cycle times, and
decreased costs . One balanced scorecard approach, described previously in this
book, considers the categories of financial, customer, and internal business pro-
cesses, and learning and growth. We believe that a balanced scorecard approach
is needed ; however, we suggest also considering reporting ROI values within
summary reports whenever possible, even if assumptions need to be made when
converting, for example, customer satisfaction to monetary units. The value of
projects can be better understood when money is also used in reporting metrics .
We believe that emphasis should be given to the search for smarter solutions
thatreduce the cost of poor quality (COPQ, as opposed to "playing games with
the numbers," which often happens in Six Sigma programs .
Consider developing a project scorecard approach to metrics, since improve-
ment in one area might degrade another. Figure 10.3 shows a sample project
scorecard, including examples of measurements that an organization may want
to reference whendeveloping its own project metrics . Some items under "Cus-
tomer" and "Learning and Growth" may seem difficult to measure . Consider
using surveys for items that are nebulous and where objective data collection is
difficult . The glossary in this bookcan be referenced for terms that are unfamil-
iar to the reader.
Project Scorecard
FINANCIAL
CUSTOMER
Inventory Levels
Customer Satisfaction (CTQ)
Cost per Unit
On Time Delivery
Hidden Factory
Product Quality (KPOV)
Activity Based Costing
Safety Communications
Cost of Poor Quality
Overall Project Savings
firefighting within an organization and can help refocus common cause issues
toward process improvement activities. Other statistical tools can then be used
to identify the amount of common cause variability and what should be done
differently to improve the process . For identified Key Process Input Variables
(KPIVs), control charts can be created and utilized at a "50-foot level" to track
the variables and identify when special causes occur so that immediate action
can be taken.
Care should be taken in how data are collected. The effectiveness of control
charts is highly dependent on the rational subgrouping that is used during data
collection. Sampling plans can make a big difference in what is reported as
special cause or common cause .
When creating "30,000-foot level" control charts, parts should be selected
so that time between samples captures the entire variability of the process. In
addition, a control chart strategy should be selected that uses subgroup variabil-
ity when determining control limits . Breyfogle (1999), in Implementing Six
Sigma (chapter 10), describes different control chart techniques and how an
XmR chart, for example, can in many cases be a better charting alternative than
a p chart or an x and R chart.
DETERMINE A BASELINE AND BOTTOM-LINE BENEFITS 157
Many projects start out with goals and objectives that are too broad. Project
leaders may feel overwhelmed with the task at hand. They may flounder, revis-
iting the measure phase and rethinking their project approach in frustration .
This can wear on the confidence ofthe Six Sigma practitioner and performance
levels of the team. This frustration often stems from the fact that leaders are
working on a project that is too big and are actually trying to solve multiple
projects at once. A strategy is needed to help organizations narrow the scope of
projects and focus on areas that have the biggest impact first .
Frequently, the initial project goals that managers set for their teams serve as
"parent projects," with multiple projects that roll up to meet the desired perfor-
mance metrics. A method should be established to keep track of"child projects"
to be completed at a future date so they are part of the overall plan for selecting
projects .
If a project has been properly scoped, an organization should be able to
answer the following questions when the project is finished:
Consider having project leaders fill out a.checklist before submitting a project
for approval. One section could deal specifically with project bounding, asking
if the project has customer impact; does not try to "boil the ocean"; and can
specifically answer "what is the defect?" An organization is on the right track if
it can answer the questions listed above before establishing a project team and
creating a charter.
Process Mapping
Creating a process flowchart as a team is a great way to display an accurate
picture of the process and to gain insight into opportunities for improvement .
Later in the project, the flowchart can be used for maintaining consistency of
process application and subsequent improvement/establishment of standard
operating procedures.
Process mapping can be laborious if the process is not bounded and the de-
tail level agreed upon beforehand. The following are tips for proficient process
mapping :
SCOPE THE PROJECT 159
The Six Sigma practitioner should embrace any conflict that arises during
this process . The dialogue that emerges from a team during a flowcharting ses-
sion is as important as the output of the session. Conflict in the discussion may
pinpoint problems in the actual process . For further details and the mechanics
of creating process flowcharts, refer to Chapter 4 of Breyfogle (1999), Imple-
menting Six Sigma .
Pareto Chart
Pareto charts can help identify sources of common cause variation within a
process . Pareto charts are based on the principle that a "vital few" process char-
acteristics cause most quality problems, while the "trivial many" cause only a
small portion of the quality problems . When nonconformance data are classi-
fied into categories, most of the time it is evident where initial project focus
should lie .
Cause-and-Effect Diagram
When the process map is complete, it is time to narrow down the focus further.
A brainstorming session can be held to determine the key causes that are creat-
ing defects within a process . A helpful tool in determining these causes is the
cause-and-effect diagram .
When creating a cause-and-effect diagram, a team should keep in mind the
following : The output of a cause-and-effect diagram depends to a large extent
on the quality and creativity of the brainstorming session. Agood facilitator can
help teams stay creative and ensure that feedback is received fromall participants .
One approach when conducting a brainstorming session is to have team
members spend 10 to 15 minutes writing their ideas on notecards in silence .
Next, shuffle .the cards, reading each aloud as the cards are organized within
similar categories in plain view of all team members . Blank notecards can then
be used to capture the new ideas that will be created during the discussion of
each card.
160 PROJECT SELECTION, SIZING, AND OTHER TECHNIQUES
Once all new ideas have been submitted, spend some time reorganizing the
cards and combining similar ideas. Often, facilitators use the following catego-
ries to group the data:
Personnel
" Materials
" Measurements
" Methods
" Machines
" Environment
This approach sometimes helps teams to focus their thinking if the session is
generating few ideas. However, it can stifle the creativity that goes into think-
ing about causes . We believe it is more effective to assign natural categories for
the causes once brainstorming sessions are complete.
Cause-and-Effect Matrix
When the cause-and-effect diagram is complete, a method is needed to rank the
list of causes and determine critical focus areas. A cause-and-effect matrix is a
simplified QFD (quality function deployment) that gives focus to the inputs
that are most likely to affect output (KPOVs) .
In the cause-and-effect matrix shown in Table 10.1, outputs are listed in the
top row and are assigned values according to importance to customers and stra-
tegic business goals. Ahigher number indicates more importance . The row con-
tains a list ofinputvariables fromaprocess map orcauses from a cause-and-effect
diagram . At the intersection of each row and column, values are entered that
quantify the amount of correlation thought to exist between the row and col-
umn factors . Relationships that are thought to be stronger can be given more
weighting . One might, for example, use the following scale:
0 = no correlation
1 = remote correlation
3 = moderate correlation
9 = strong correlation
Numbers in each input row are cross-multiplied with the importance num-
ber at the top of the column and summed across each rove. A ranking of the
magnitude of the cross-multiplied values gives focus to the input variables the
team thinks are most important . The highest-ranked values are then considered
to be KPIVs, and should be given the most attention for additional study and/or
process improvement efforts .
CREATE PROJECT CHARTER AND OVERALL PLAN 161
The project charter is an important activity ofthe planning process for any new
team. It acts as the preliminary project plan and increases teamwork. Many
teams skip over this important step because they view it as a "soft skill" and
don't understand its value in helping to create a well-functioning team.
Perhaps the most beneficial outcome of a charter is team cohesion . Creating
a charter as a team communicates goals and individual responsibilities clearly.
It is a good team-building exercise and is an opportunity to get potential road-
blocks out in the open for discussion . When developed as a group, a charter can
create a sense of team ownership of the project.
The second major benefit ofa project charter is that it clearly defines expec-
tations between managers and team members, preventing delivery of mixed
messages . A charter should be reviewed with upper management at a kickoff
meeting to ensure that the overall project supports company strategy.
At a minimum, the project charter should document the following key items :
Lack of administrative
Too many buyers 0 Many sites independently ;
support 0 Purchase Cards G purchasing f om common
sup liers © ! mpact
Standard procedures Face Approvals 0 ,
not defined/ Field prefers local IHigh Low
communicated at T&L Cards ® / suppliers (3 Easy
a-
field level 1 2
Too many suppliers '.Implement
3
Hard
PROJECT CHARTER
KEY MILESTONES
Define Phase Completion January 14, 2000
Measure Phase Completion March 10, 2000
Analyze Phase Completion May 5, 2000
Improve Phase Completion June 30, 2000
Control Phase Completion August 2000 (TBD)
Verify Phase Completion TBD
Along with the project charter, a good plan is essential to the success of any
project . Figure 10.5 shows a sample project plan, also from a team at Titanium
Metals Corporation . They created this plan in the measure phase of their first
Six Sigma project, and it served as a template for future project work. Again,
this example is presented to show the items and detail included in an effective
project plan. Duration dates for any particular project can differ significantly
from these, depending on the scope of the project and team/management com-
mitment, among other things. The ID nomenclatures within the figure as they
relate to Six Sigma phases are as follows :
who not only have the ability to understand the statistical methodology but also
possess excellent leadership and facilitation skills . To be truly successful, Six
Sigma team members must be able to relate to each other as co-owners of the
project . Relationship patterns in teams encompass the following six categories,
ranging from adversarial to partnering (Weaver and Farrell, 1999) :
A good facilitator can help a team progress into a less adversarial relation-
ship. Weaver and Farrell describe a practical method for facilitators to use in
promoting this growth (Weaver and Farrell, 1999). The approach, called TAR-
GET (Truth, Accountable, Respect, Growth, Empowered, Trust), can be sum-
marized as follows :
Truth: Group members must know the truth about expectations and expe-
riences . Facilitators can help create a climate of truth within group meet-
ings by honestly reflecting their own experiences .
Accountable: Group members need to be accountable to each other for
their work. Facilitators can create a structure for accountability in meet-
ings, such as reviewing action item lists and periodically revisiting the
group's charter.
Respect : Group members must experience others acting with integrity and
honesty. Generating respect in a group is a tough task for a facilitator
because it is so closely tied with trust. The best approach is to try to name
the underlying problem directly and generate a plan to understand and
address the issue.
16
Growth : Team members and the teamitself needto grow as a whole through
learning to become more efficient at tackling difficult tasks. Facilitators
can promote growth by helping groups recognize and value it.
Empowered : Successful groups depend on independent action. Facilita-
tors can help by identifying opportunities for independent action and can
suggest methods for reflecting independent work back to the group.
10
Trust: Effective groups have team members who trust each other and can
count on each other to complete the necessary work of the team. Facilita-
tors can help teams identify areas where trust is- the strongest and the
weakest.
w
.J
~ ~ ~ ~ W W W ~ W ' W IL W W ~ W ~ W
W
_a a ~
r
~
~ r
tL ~
~ C
o
m a ~
x ..
m
A
E ~ a
P W
r N !7 ~! W f0 h q C r N N ~ h i0 h A CI ~ r
r r r r r r r r r' r N
3 ~ g ti I ti
~ ~ r w ~ M ~ ~ r r w ~ r r
m
O
Ad
o "r a
d v
9 ~,
g .E
" r3.
d ~ d d
a
U
N N N M a M
17 0 PROJECT SELECTION, SIZING, AND OTHER TECHNIQUES
Cross-Functional
Teams should have cross-functional representation and possess complemen-
tary skills . It is best to have only a few participants within each team. When a
cross-functional team is assembled, most members have hidden agendas, de-
pending on the department they are from. As a result, a professional facilitator
is sometimes needed to keep a team on track and focused on one agenda. Facili-
tators can also help improve communications between departments and elicit
feedback from reluctant team members.
Performance Challenges
A team's likelihood of success increases if its leaders clearly communicate chal-
lenging performance objectives based on the needs of customers, employees,
and shareholders .
Meaningful Goals
Arbitrary goals set for individuals or organizations who have no control over
the outcome can be very counterproductive and very costly. Each team member
must feel that the purpose of the team is worthwhile and important to the orga-
nization as a whole. Successful teams have a shared purpose that can be articu-
lated by all members .
Clear Approach
Team members must speak a common language . They need to understand and
agree upon the approach to solving the problem at hand. Every member must
understand the process they are undertaking.
Open Communications
Open, honest, and frequent communication among team members should be a
primary goal. Members should feel free to speak their mind and give honest
feedback .
Feedback
What gets measured and rewarded gets done, but few companies recognize or
reward good performers effectively. To be truly motivating, the recognition pro-
cess needs to be well thought out and personal . It should be done in a way that
PROJECT EXECUTION CONSIDERATIONS 171
It was noted that many projects tried to take on a "solving world hunger"
mind-set and needed to be reworked into manageable sizes. Also, other projects
seemed to be picked out of the blue and didn't have substantial benefit poten-
tial. Because of this problem, a phase to define projects was added, which re-
sulted in the define-measure-analyze-improve-control (DMAIC) nomenclature
for the overall Six Sigma implementation process . There was also a refocus on
the training of Black Belts and Champions in project scoping, which signifi-
cantly improved the success rate ofprojects.
When a project is properly scoped, sometimes only very simple changes are
needed to make the process much better. Strict process improvement tactics
that are applied rigidly can yield benefits from the "low-hanging fruit" that is
prevalent on many projects at the start; however, "big hitters" in the process,
172 PROJECT SELECTION, SIZING, AND OTHER TECHNIQUES
those that are further up the tree, can require much creative thinking . Many
initial projects can be simple reductions of unplanned slack and redundancy.
But when tackling more persistent process problems, the statistical tools of a
Six Sigma methodology can provide a wide array of options .
There are many combinations of Six Sigma tools and charts that will help
determine the root cause of variation within processes . However, care must be
taken to avoid becoming overwhelmed by the data and slipping into a "data
analysis paralysis" mode. Also, many companies need to "clean up" their data
collection and measuring processes before data can be analyzed in a meaning-
ful way. For example, in many cases a Gage R&R study needs to be conducted
early when investigating process improvement opportunities. Many companies
initially find that their measurement systems are not accurate and require modi-
fication.
Figure 10.6 illustrates the concept of passive data collection of input vari-
ables that were thought to be significant from wisdom-of-the-organization as-
sessments during the measurement phase of an S4 project, using such tools as a
cause-and-effect diagram, a cause-and-effect matrix, and FMEA. Subsequent
activity, part of the analysis phase, can involve graphical tools such as multi-
vari charts, box plots, and marginal plots. Testing significance and making pre-
dictions involve statistical tools such as regression, analysis of variance,
correlation studies, and variance components.
Consider the possibility that the measure and analyze phases did not yield
significant improvements through passive assessment. This could occur because
the input variables were not large enough to detect a difference or because care
was not taken to establish a data collection system that produced meaningful
results . ADOE factorial methodology can be usedto assess the impact of changes
to the process . This approach involves making proactive changes for the pur-
pose of determining what long-term changes are most beneficial for process
improvement.
Six Sigma training provides a toolbox of statistical techniques and excellent
problem-solving/project management skills that when combined can truly make
breakthrough improvements in business processes . This approach can facilitate
creativity and the willingness of individuals to let he "good" data lead them
down the path to appropriate improvements . Again, the point of failure inmany
Six Sigma projects is rushing to solutions. Repeatedly, we see practitioners
begin a project with the answer in mind. They go through the motions of apply-
ing the Six Sigma tools, but often they do not address the biggest opportunities
for improvement and bottom-line benefits .
In an earlier chapter we discussed Six Sigma deployment. Figure 7.2 high-
lights our suggested overall business strategy of this deployment through projects.
In the chapters that follow, we give specific examples that follow our "21-Step
Integration of the Tools," a road map of how individual tools fit into the big
picture.
Yes End
se uencin of data
Six Sigma practitioners often feel pressured to apply all the tools they have
learned and include them in presentations to executive management. This can
result in a presentation that is difficult to follow.
Finding the root cause of a problem requires diligence and forethought . A
cursory view of data is not sufficient, because it can lead to detecting mere
symptoms of the true underlying problem . To be efficient, data analysis must
use sophisticated statistical tools for the purpose of determining a concise and
simple definition of the problem's key causes . It should tell a story that is well
thought out and ends with a simple conclusion . An effective presentation can
consist of as little as a few good charts :
We suggest that companies investigate how they can establish a periodic
review process for all projects. When creating this "project report-out" process,
a company should consider holding a brainstorming session with employees
who will participate in all areas of the process . The goal is to obtain group
consensus on how frequent project report-outs should take place and who should
be involved.
We also suggest that the team spend time developing templates that Black
Belts can follow in presenting theirmaterial to upper management. A report-out
can take the form shown in Figure 10.7 through Figure 10.11, which describe
how someone might present the work of a project to management in a closeout
report. Graphs that tell the best story were selected from each phase.
Project Number :
Project Title :
Improve Phase
Black Belt
Master Black Belt
Sponsor
Champion
Date
Team Members :
Cause-and-Effect Diagram
sartstoottai"l R
ab Stan R Ira It Ie tae
split
LCDd trip
Owen *so a
Cods C8Wga
Ikaord Estt
C
Process hlap Cause and Effect Di ram
MW CapaWAlai(kUdMbee(31
Lateral
Distance
Projection Projection From edge
Distance Distance of tape
Process Outputs »»»» (Average) ariabili CycleTime measurer <<<<Process Outputs
Measurement Rounding 3
O I -
Error 3 0 60
Measurement Impact
Reference Point 3 3 0 0 1 so
ion . .
J
r
.-ft)
meana ants -~-sue
start Angle Wrong setting .7 6 None
-
_ offtwget _ Wig. ._......____..._._..
Lockin ... . .. ... _
7 5 None
mechanism slips
.,
distance
Unstable variability 3 holding 9 No. 9 243 Redesign
exoesais. rnachanism
_...___._._ ... . ___. .__...._.__._.__ _ . ..__._____._. . .._.. w_..____.__..._. . .___ __..3 -pie"ssablli{y ..5 ._....__ ..__.._..___._ ._9 .... . ... ....._._.. .._. ..... . ._ ...... ......_...._ _
.__
None 135 -...~. . . ..__
of tubing
rowemente:
I
-Use aluminum foil and mesured io center of ball maikwithin 74" crurc.xM -~ - . 1t! .
i~ . .f
-Change start angle to solid posilon. To.pet ?6" approsdmately n r! f0
Needed to readjust from 2D degrees start angle. 78 degrees stop
Angle . 9" arm Iength13.Qcm rubber band pivot ID other settings. 1rroNnrerr lance
-Documented and periodically ahedad settings Rrr'rlRk~enna
sax+r ~!
yr l .earn 5T
lyl IT
Irl "rJrrr
i<+Kas
r»err rr
rr"" rA'1 hrrrN
f! rN'1 r.rN""
~K~ as
srsw " .Yrlr 1i y a NYlr*
o b r..c
are r.N
H" r.w
"~1L+tJcO 4L i.M
arm ,r "~ .r
r.wxaya .r .wrt
~r r.o rrr .lrl r.rr err+nl " .+crrJr rrr~ln rrw" .r.
nr r.a nr" vrl ear rrr,vtl " .ra.n rrr " vrl r .r "rar
m . r.a rrr' .m r.a rrr ..rl " rr" rr.a rrr' .m .no" .u
rev r.a
ProcessCamblMyAfter Charge
Control Chart After Change
Inspection costs = 10 10
Rework cost = 100 100
Scrap cost = 200 200
Median of data= 70 74
Upper spec limit = 73 77
Lower spec limit = 67 71
Rework upper limit = 74 78
Reworklower limit = 66 70
Number to rework = 7 9
Number to scrap = 42 5
-. .
reworked = 7a/Q 9%
Annual cost of rework .= 700000 900000
% scrapped = 42% 5%
Annual cost of scrap = 8400000 1000000
CO PQ * $10,100,000 ; $2,900,000
__$_7,200,000
Includes inspection costs
In this example we use data collected from catapult exercises used in our
workshops to illustrate how the Six Sigma tools can be integrated to solve project
problems using an S4 road map, as described in Figure 7.2. This report-out
focuses on the "low-hanging fruit" fixes that the team completed initially ; there-
fore, the report does not represent the full tool usage that project report-outs
might contain .
The details ofthe catapult exercises are described in Breyfogle (1999), Imple-
menting Six Sigma. The flow of tool usage and the people involved at various
points is similar to what is shown in section 11 .1.
Other slides to consider in a presentation of basic topics are the following :
As a mentor ofBlack Belts at General Electric, one ofour authors (Becki Mead-
ows) found that many of the GE project leaders were working on similar projects
and often had the same frustrations. There was a lack of synergy between Black
Belts and no established method for them to communicate. The need for
mentoring and communicating lessons learned led to the creation of a quality
web page. A new intranet site was created, containing the following elements:
Cleveland Service Shop Generates Most Defects With A Failure Rate of 89% Late Deliveries
There are many possible alternatives and sequencing patterns of Six Sigmatools
for S4 manufacturing projects . Ourchart called "21-Step Integration of the Tools :
Manufacturing Processes," which follows, is a road map to how individual tools
fit into the larger manufacturing picture . Specific tools are highlighted in bold
print to aid the reader in later locating where a tool may be applied. The glossary
at theback ofthisbook will help clarify terms. Breyfogle (1999), Implementing
Six Sigma, provides details on the mechanics of using the tools.
21-STEP INTEGRATION OF THE TOOLS : MANUFACTURING PROCESSES 185
15 Assess statistical significance Six Sigma practitioner and Passive data analysis
of relationships using team
hypothesis tests .
21-STEP INTEGRATION OF THE TOOLS : MANUFACTURING PROCESSES 187
16 Consider using variance Six Sigma practitioner and Passive data analysis
components analysis to gain team
insight into the source of
output variability. Example
sources of variability are day-
to-day, department-to-
department, part-to-part, and
within part.
17 Conduct correlation, Six Sigma practitioner and Passive data analysis
regression, and analysis of team
variance studies to gain
insight into how KPIVs can
impact KPOVs .
18 Conduct factorial DOEs and Six Sigma practitioner and Active experimentation
response surface analyses . team
Consider structuring the
experiments so that the levels
of KPIVs are assessed relative
to the reduction of variability
in KPOVs. Consider
structuring the experiment for
the purpose of determining
KPIV settings that will make
the process more robust to
noise variables such as raw
material variability.
19 Determine optimum operating Six Sigma practitioner and Passive data analysis
windows of KPIVs from DOEs team and active
and other tools. experimentation
20 Update control plan . Six Sigma practitioner and Passive data analysis
Implement "50-foot level" team and active
control charts to timely experimentation
identify special cause
excursions of KPIVs .
21 Verify process improvements, Six Sigma practitioner and Active experimentation
stability, and capability/ team and/or collected data
performance using
demonstration runs . Create a
final project report stating the
benefits of the project,
including bottom-line benefits .
Make the project report
available to others within the
organization . Monitor results
at 3 and 6 months after project
completion to ensure that
project improvements/benefits
are maintained.
188 MANUFACTURING APPLICATIONS
The following examples highlight how to utilize the steps outlined above to
gain insight into and improvement in manufacturing processes .
0.4
0.2
150
120
.100
Figure 11.2 Pareto diagram of solder defects [From Messina (1987), with permission]
After changes are made to the process and improvements are demonstrated
on the control charts, a new Pareto chart can be created. Perhaps the improve-
ments to the insufficient-solder noncompliance characteristic will be large
enough that blowholes will be the largest of the "vital few" to be attacked next.
Process control charts can also be used to track insufficient solder in isolation
so that process degradation from the "fix level" can be identified quickly.
Improvements should be made to the manufacturing process after a con-
firmation experiment verifies the changes suggested by the experiments . Be-
cause of these changes, the data pattern of the control chart should now shift
downward in time to another region of stability. As part ofa continuing process
improvement, the preceding steps can be repeated to identify other areas for
improvement .
190 MANUFACTURINGAPPLICATIONS
This approach is one way to define the process measurement and improve-
ment strategy. However, often in this type of situation each of the company's
manufacturing lines assembles various product types, which consist of differ-
ing circuit layouts, component types, and number of components. For this situ-
ation one might wish to create ameasurement system and improvement strategy
as described above for each product type. Amajor disadvantage to the approach
described above is that emphasis is given to measuring the product versus mea-
suring the process . Since most manufacturing improvements are made through
adjustments to the process, as opposed to the product, a measurement system is
needed that focuses on the process and what can be done to improve it. Acouple
of Six Sigma metrics may prove to be very valuable for this situation ; these
metrics are DPMO and rolled throughput yield.
Let us consider first DPMO. This metric can also bridge product type and
expose the hidden factory of reworks . As described earlier, the defects per op-
portunity (DPO) calculation can give additional insight into a process because
it accounts for number of opportunities for failure . Consider, for example, that
there were 1,000 opportunities for failure on an assembly, consisting of the
number of solder joints and the number of assembly components . If 23 defects
were found within the assembly process after assembling 100 boards over a
period of time, the estimated ppm rate for the process for that period of time
would be 230 (i.e., 1,000,000{ 23/[100 x 1,000] } = 230). If someone were to
convert this ppm rate to a sigma quality level, they would report a 5.0 sigma
quality level for the process using Table 2.1. The sigma quality level of 5 ini-
tially seems good, since average companies are considered to be at a 4.0 sigma
quality level, as noted in Figure 2.3. However, if these defects were evenly
distributed throughout the 100 assemblies produced such that no more than one
defect occurred on each assembly, 23% ([23/100] 100) of the assemblies would
exhibit a defect somewhere within the manufacturing process . We believe most
readers consider this an unsatisfactory failure rate for the manufacturing pro-
cess, which is one reason why we discourage the reporting and use ofthe sigma
quality level metric to drive improvement activities.
However, a reported ppm rate can be a very beneficial metric in that it can
create a bridge from product to process and can expose the hidden factory of
reworks. We should note that it would be difficult to examine the magnitude of
this metric alone to quantify the importance of project activities within this
area. For example, the manufacturing of a computer chip might have millions
ofjunctions that can be considered opportunities for failure . For this situation,
we might report a sigma quality level much better than six and still have an
overall defective rate of the assembled unit of 50% .
Another assembly situation might have only a few parts that are considered
opportunities for failure . For this situation we might report a sigma quality
level of 4.0 but an overall defective rate ofthe assembled unit of0.1 %. Because
of this discrepancy we suggest tracking ppm rates directly over time using a
EXAMPLE 11 .2 : BETWEEN- AND WITHIN-PART VARIABILITY 191
control chart . When a process exhibits common cause variability, the centerline
of the control chart can be used to describe the capability/performance of the
process . Cost-of-poor-quality calculations can then be made using this value
and the cost associated with the types ofdefects experienced in the manufactur-
ing process . This cost of poor quality can be used to determine if a project
should be undertaken, or as a baseline for determining the value of process
improvement activities resulting from a project .
Rolled throughput yield is another metric that can be useful in this situation .
This metric, which quantifies the likelihood of product progressing through a
process without failure, can expose major sources of reworks within a factory.
To create this metric, the process is subdivided into process steps such as pick-
and-place of components, wave solder, and so forth. The yield for each step is
multiplied :together to determine rolled throughput yield. Steps that have low
yields are candidates for initial S4 process improvement efforts .
Other areas of this manufacturing process thatcould be considered for projects
are inventory accuracy, frequency and cause of customer failures, expense re-
duction, effectiveness of functional tests, and functional test yields .
Chapters 10, 34, and 35 ofBreyfogle (1999), Implementing Six Sigma, describe
a variety of control charting techniques . The example described here is an ap-
plication of the three-way control chart and improvement efforts .
Plastic film is coated onto paper. Three samples are taken at the end of each
roll (Wheeler, 1995a) in such a way that the time between samples captures the
entire variability in the process . Care needs to be exercised when evaluating
this type of situation . Sometimes practitioners erroneously create x and R con-
trol charts using within-part variability as though they were~measurements from
different part samples . This type of data should be plotted as a three-way con-
trol chart, which can then be used to understand the variability between and
within rolls, 'as shown in Figure 11 .3.
The chart for subgroup means from this figure shows roll-to-roll coating
weights to be out of control . Something is apparently happening within this
process to allow film thickness to vary excessively. The sawtooth pattern within
the moving range chart suggests that there is a large change in film thickness
every other roll. The sample range chart is indicating stability between rolls;
however, the magnitude of this positional variation is larger than the average
moving range between rolls, which is an opportunity for improvement .
We are now at step number 6 of our chart called "21-Step Integration of
Tools: Manufacturing Process ." We need to assess the process to determine the
process capability/performance from the needs of the customer, as described by
192 MANUFACTURING APPLICATIONS
3.08L-11 .75
10 ._:
"
5
" -3.595
O "
0 " -USL-0.000.
Range Chart of AI Data
o) so , 3ASL-5921
c 50 ,
12 40 _. , ly
,
aD 30 "
a 20 --r " P-23.00
10 --
0 - -3ASL-0.000
4-
Subaroup 10
11 .4 SHORT RUNS
Note that the emphasis in these three application scenarios is product mea-
surement. Our thinking about the applicability ofcontrol charting to these situ-
ations changes when we use a methodology that links these product scenarios
to process measurements . Chapter 34 of Breyfogle (1999), Implementing Six
Sigma, describes short-run control charting, which makes a bridge from prod-
uct to process measurements . Future part numbers from a process can benefit
from the current wise application of control charting and process improvement
activities.
Manufacturing and business process applications for short-run charts include
Consider a situation where parts are produced by a tool that needs periodic
sharpening, adjustment, or replacement . These situations are application ex-
amples ofexponentially weighted moving average (EWMA) control charts with
engineering process control (EPC).
In the Shewhart model for control charting, the mean is assumed to be con-
stant . Also, errors are assumed to be normal, independent with zero mean, and
with a constant variance 62. In many applications, these assumptions do not
hold true. EWMA techniques offer an alternative that is based on exponential
smoothing, sometimes called geometric smoothing .
Taking the weighted average ofpast observations withprogressively smaller
weights over time gives the EWMAcomputation the effect of a filter. The addi-
tion of a weight factor to achieve balance between older data and more recent
observations gives EWMA flexibility of computation.
EWMA techniques can be combined with engineering process control (EPC)
to give insight into when a process should be adjusted. The mechanics of con-
ducting an EWMA and EPC are described in Chapter 36 of Breyfogle (1999),
Implementing Six Sigma.
The experiment's results led to a process optimization that cut the number of
rejects in half as compared to the performance of the previous equipment . In
addition, data analysis revealed that the pin protectors (i.e., cover/shield) were
not needed . This permitted an employee to be moved to another process.
The mechanics of setting up and conducting fractional factorial DOEs are
described in Chapters 27--33 of Breyfogle (1999), Implementing Six Sigma.
12
SERVICE/TRANSACTIONAL
APPLICATIONS
Many companies have learned akey lesson intheir implementation ofSix Sigma:
successful outcomes are not limited to manufacturing processes . A number of
our clients have had great success applying S4 methodology to their key ser-
vice/transactional processes . Consider the following cases :
195
19 6 SERVICE/TRANSACTIONALAPPLICATIONS
There are many possible alternatives and sequencing patterns of Six Sigma
tools for S4 service/transactional projects . Our "21-Step Integration ofthe Tools :
Service/Transactional Processes," shown below, is a concise summary of how
Six Sigma tools can be linked or otherwise considered for service/transactional
projects . This road map can be referenced for insight on how individual tools fit
into the "big picture ." Specific tools are highlighted in bold print to aid the
reader in later locating where a tool may be applied. The glossary at the back of
the book will help to clarify terms. Breyfogle (1999), Implementing Six Sigma,
provides details on the mechanics of using the tools.
of the process . Table 12.1 shows the data sampled for creation of the control
chart. Note that the control chart limits are a function of sample size and that
variability between samples is not taken into account when calculating the lim-
its. The data were collected on a late/not-late criterion and were therefore tracked
using a p chart. As Figure 12.1 shows, the process is in statistical control but
has a presumably unacceptable overall late-delivery rate ofapproximately 61 %
(Chapter 10 ofBreyfogle [ 1999], Implementing Six Sigma, discusses other con-
trol charting possibilities) .
The Black Belt compiled a database of all pertinent details on the sampled
purchase orders. She analyzed the data, looking for trends in late deliveries .
Figure 12.2 shows a Pareto chart of the data by service center. She facilitated
the team in a brainstorming session to compile a list of possible causes for late
deliverieswithin a cause-and-effect diagram . During the meeting, service cen-
ter managers denied having such a drastic late-delivery problem, but the data
proved them wrong. Headquarters thought the service center managers were
202 SERVICE/TRANSACTIONAL APPLICATIONS
0 .8
UCL=0 .7596
0 .7
0
'C
0 P=0.6135
0 0 .6
0 .5
LCL=0.4674
10 20
Sample Number
not motivated to change . Team members thought the project was a waste of
time. Overall buy-in was low. The Six Sigma practitioner expended a lot of
time looking for trends by service center, suppliers, materials, and time ofyear,
but no trends that would account for late deliveries were immediately apparent.
The Black Belt decided to create a detailed map of the purchasing process .
By mapping the purchase-order entry process in detail, the team determined
that under certain circumstances material request dates were defaulting to the
same day the order was placed . Owing to incomplete purchase-order forms,
more than 60% of the late purchase orders were placed in the emergency order
category that is, orders needed the same day. Before a project could be under-
taken that truly improved on-time customer delivery, improvements to the
purchase-order entry module were needed .
The team decided to mistake-proof the process and made the material re-
quest date a required field . If same-day delivery was entered on the order form,
the order-entry person was prompted to select a reason for the emergency order.
Through changes in the computer system and training of order-entry personnel,
the process went from 1.02 sigma to 2.71 sigma . The before and after outputs to
the process that illustrate this dramatic improvement are shown in the Pareto
chart in Figure 12.3.
At this point charts of the new data, similar to Figure 12.4, can be re-created.
Any points beyond the control limits should be investigated. The process de-
EXAMPLE 12 .1 : IMPROVING ON-TIME DELIVERY 203
Los Angeles
Atlanta
Boston
Detroit
NewYork
Richmond
Salt Lake
New Orleans
Dallas
Philadelphia
Tucson
Charlotte
Pittsburgh
Denver
Cleveland
Jacksonville
San Francisco
30
Los Angeles
Atlanta
Boston
Detroit
New York
Richmond
Salt Lake
New Orleans
®After
Dallas
" Before
Philadelphia
Tucson
Charlotte
Pittsburgh
Denver
Cleveland
Jacksonville
San Francisco
20 40 60 80
Figure 12.3 Pareto chart of late-delivery data before and after process change
204 SERVICE/TRANSACTIONAL APPLICATIONS
scribed above can then be repeated using the "good" data, and a root-cause
analysis of late deliveries can again be expressed in Pareto chart format to pro-
vide direction for further process improvement efforts. Control chart tracking
can then be used in greater detail to monitor the largest Pareto chart items and
provide insight as to when process improvements focused on these items are
effective (i.e., should go out of control again for the better) . The team should
also translate any ofthese quantifiable defect-rate improvements into monetary
benefits for presentation to upper management .
While it proved difficult to calculate "hard" monetary savings to this initial
project, altering the order-entry process was the necessary foundation for im-
proving on-time delivery in the service centers, and it led to improvements in
the bottom line with respect to working with specific suppliers. It was the first
"big hitter" that needed to be attacked, but this was difficult to see in the data.
Now that the team had "good" data to work with, objectivity and buy-in in-
creased .
It should be noted that for this project the data could have been transformed
from attribute (i.e., order was late) to continuous (i.e., number of days late) .
The response charting criterion could be changed from pass/fail to number of
days late. A team could then create a "30,000-foot level" control chart, ran-
domly selecting one order per day, to obtain a long-term view of the process
output .
0.6
0.5
0 0.4
UCL=0 .3660
0
0 0.3
a. P=0 .2382
0.2
LCL=0.1104
5 10 15
Sample Number
A company was unhappy with the number of "hits" its website had received
and the website's rank with search engines. The company formed an S4 im-
provement team to increase traffic on the website. Upper management gave the
team a budget and an aggressive goal: doubling the number of hits within the
next four months .
The team compiled "hit-rate" data in time series format from the website,
looking at hit rates by week over the past three months . The data used for the
project are shown in Table 12.2, and an XmR control chart of this data is shown
in Figure 12.5. Note that weeks 4 and 5 are beyond the control limits . Investiga-
tion of-these two data points showed that they reflected website traffic during
the last two weeks ofDecember, when all websites generally have lower activ-
ity owing to the holiday season . Since these data points could be accounted for,
they were removed from the data and the control charts were re-created.
6000
UCL=5742
5000 -
Mean=4233
4000 -
3000 -+
LCL=2724
2000 -
Subgroup 0 10 15
2000 -i
UCL-1854
N
t0
DC 1000 --
1567.4
O .
0 LCL=O
Increase
Website Hit Easy to Process
Rate Implement Outputs
10 10 -4 Importance
Process Input Correlation of Input to Output Total
No data were available on the relationships of the KPIVs to the website hit
rate, so the team decided to run an experiment. To test the critical factors, they
established different gateway pages with links back to the main site. Instead of
testing these factors one at a time, the Six Sigma practitioner opted for a DOE
approach. He knew that a DOE approach would make the team more efficient
and allow them to gain a better understanding of how the critical factors inter-
acted with each other and affected website hit rate.
Toreduce the size of the experiment and the numberofgateway pages needed,
the team reviewed the list of factors from the cause-and-effect matrix and pared
itdown to factors that wouldyield insight specifically into website traffic . Next,
a meeting of peers from all affected areas was conducted to consider expansion
or deletion of selected factors, outputs, and limiting assumptions in the experi-
ment. The teamthen grouped the factors into two levels, chosen to represent the
extremes ofeach setting . The revised list, with settings, is shown inTable 12.4.
Factor +
Designation Setting Setting
The Six Sigma practitioner usedthe procedure listedin Chapter 42 and Table
M2 of Breyfogle (1999), Implementing Six Sigma, and came up with the test
matrix shown in Table 12.5. Eight gateway pages were created, each page rep-
resenting one of the trials listed in this table. Hit rates were then collected on a
weekly basis to determine which combination of factors produced the most
website traffic .
208 SERVICE/TRANSACTIONAL APPLICATIONS
Trial Number A B C D E F G
1
2
3
4
5
6
7
Based on the results of the experiment, the team could determine which
factors to focus on in its attempt to increase traffic on the company website .
Using the results they could, for example, test the hypotheses developed by
Web Consulting Honolulu concerning the most important things a company
can do to increase website traffic (Web Consulting Honolulu, 1999) . The con-
sulting firm believes that having the right key words and offering a "gift" (e.g.,
a discount or a download) are key contributors or KPIVs to increase website
traffic .
As part of the control plan, the team may choose to checkthe ranking of the
website against those of other sites weekly, using a combination of keywords
they have deemed important to the increasing interest in the site. Whenever the
site ranking degrades, adjustments are considered and/or another DOE is con-
ducted to address any new technology issues that may have a bearing on website
traffic.
" The bottom line can be improved through the customer invoicing process :
Example 43.8.
" An improvement in the tracking metric for change order times can give
direction to more effective process change needs : Example 43.6.
" Illustrating how S4 techniques improve the number in attendance at a local
ASQ section meeting (the same methodology would apply to many pro-
cess measurement and improvement strategies) : Example 11 .5 and Ex-
ample 19.5 .
1~3
DEVELOPMENT APPLICATIONS
Figure 13.1 Measurements across products developed over time with information feed-
back from manufacturing'to development
There are many possible alternatives and sequencing patterns of Six Sigma
tools for S4 development projects. Our "21-Step Integration of the Tools : De-
velopment Processes," which follows, is a concise summary of how Six Sigma
tools can be be linked or otherwise considered for development projects . This
road map can be referenced to give insight as to how individual tools fit into the
"big picture ." Specific tools are highlighted in bold print to aid the reader in
later locating where a tool may be applied . The glossary in the back of the book
214 DEVELOPMENT APPLICATIONS
can help clarify terms. Breyfogle (1999), in Implementing Six Sigma, provides
details on the mechanics of tool utilization.
7 When measuring the Six Sigma practitioner and Current and collected
development process, team data
specifications are not
typically appropriate.
"Long-term" process
"capability/performance"
of KPOVs can then be
described in "frequency of
occurrence" units. For
example, 80% of the time
the development cycle
time of a product is
between 6.1 and 9.2
months . Another example :
80% of the time the
number of reported
problems for a developed
product is between 2 .8 and
216 DEVELOPMENT APPLICATIONS
13 Collect data from previous Six Sigma practitioner and Collected data
development projects to assess team
the KPIV/KPOV relationships
that are thought to exist .
14 When enough data exist from Six Sigma practitioner and Passive data analysis
previously developed products, team
use multi-vari charts, box
plots, and other graphical tools
to get a visual representation of
the source of variability and
differences within the process .
15 When enough data exist from Six Sigma practitioner and Passive data analysis
previously developed products, team
assess statistical significance
of relationships using
hypothesis tests.
16 When enough data exist from Six Sigma practitioner and Passive data analysis
previously developed products, team
consider using variance
components analysis to gain
insight into the source of
output variability. Example
sources of variability are
supplier-to-supplier and plant-
to-plant variability.
17 When enough data exist from Six Sigma practitioner and Passive data analysis
previously developed products, team
conduct correlation,
regression, and analysis of
variance studies to gain
insight on how KPIVs can
impact KPOVs .
19 Determine from DOEs and Six Sigma practitioner and Passive data analysis and
other tools optimum operating team active experimentation
windows of KPIVs that should
be used within manufacturing .
20 Update control plan, noting Six Sigma practitioner and Passive data analysis and
that fractional factorial DOE team active experimentation
experiments, pass/fail
functional testing
methodologies, and wisely
applied reliability assessments
can be an integral part of the
control plan of the
development process. Also,
consider within this control
plan the involvement of
development and other
organizations whenever field
problems occur for the purpose
of immediate problem
resolution and improvements
to the overall development
process for future products .
EXAMPLE 13 .1 : NOTEBOOK COMPUTER DEVELOPMENT 219
to conduct the testing of new instructions ; the factors from this matrix would
include items such as experience of the person setting up the new computer
design .
For the second-highest category, "no trouble found," in the PAreto chart a
cause-and-effect diagram, a cause-and-effect matrix, and then FMEA also indi-
cated other control measures that should be in place. Examples ofthese types of
control measures found in Breyfogle (1999), Implementing Six Sigma that
could be identified through S4 projects are:
Testing for thermal design problems that are causing NTF: Example 30.2
Creating a more effective test for reliability issues using fewer resources
through DOEs: Example 43.2
Improving software/hardware interface testing : Example 42.4
19
Managing product development of software: Example 42.6
Tracking of change orders : Example 43.6
Creating a more effective ORT test strategy : Example 40.7
Using a system DOE stress-to-fail test to quantify design margins : Ex-
ample 31.5
Using a DOE strategy to size the future capability of a product by testing
a small number of preproduction samples . Methodology can also deter-
mine if any key process manufacturing parameters should be monitored
or have a tightened tolerance: Example 30 .1
Using a QFD in conjunction with a DOE to determine manufacturing set-
tings : Example 43 .2
NEED FOR CREATIVITY,
INVENTION., AND INNOVATION
For more than a decade Fortune magazine has been ranking major public U.S.
companies on "innovation ." Their findings are insightful (Jonash and
Sommerlatte, 1999). The top 20% of firms rated high on innovation also have
twice the shareholder rate of return as compared to other firms in their industry.
The bottom 20%, on the other hand, report shareholder returns 33% below those
of their industrial peers. These data suggest that innovation likely plays a con-
siderable role in corporate profitability and shareholder wealth .
Six Sigma needs to be more than metrics and improvements to existing pro-
cesses. To maintain customer relationships, a company must create a culture in
which innovation occurs continuously within the company's services and prod-
ucts. The needs of the customer are dynamic . Product features that were con-
sidered "wow" features in the past are now taken for granted . For example, a
person seeing for the first time a car that does not need a "crank start" would
consider this a "wow" change ; however, electronic starters are now taken for
granted . Noritaki Kano's description of this concept is illustrated in Figure 14.1
(King, 1987).
The diagonal arrow :in the middle, "one-dimensional quality," shows the situ-
ation where the customers tell the producer whatthey want and the producer sup-
plies this need. The lower arrow represents the items that are expected; customers
are less likely to mention these, but they are dissatisfied if they do not receive
them. Safety is an example ofitems in this category. The top arrow represents
wow" quality ; these are the items that a normal customer will not mention . Inno-
11
222 NEED FOR CREATIVITY, INVENTION, AND INNOVATION
Physically fulfilled
condition
Creativity
Creativity is the discovery or elicitation of something new, usually in the form
ofknowledge. Invention is the creation ofnew and useful things. An innovation
is the embodiment of those "things" into prototypes with the potential for be-
coming profitable commercial goods and services . Gilman (1993) would add a
fourth category : commercial activity. And if he were forced to measure each
one ofthe four using only a single metric for each, he would measure (1) basic
research (creativity) by "papers published," (2) applied research (invention) by
"patents received," (3) development (innovation) by "prototypes/artifacts" cre-
ated, and (4) commercial activity by "profits ."
"The primary objective of industrial research is to generate profits" not
knowledge (Collier and Gee, 1973) . However, there is no assurance that the act of
creation or discovery will lead to an innovative product or service that will sell at
aprofit. The presumption that there is a strong relationship between basic research
and commercial success or supremacy is simply wrong (Gomory, 1990) . "New
knowledge-especially new scientific knowledge is not the mostreliable or
most predictable source of successful innovations" (Drucker,1985a) . Are these
findings at odds with the inferences drawn from the first paragraph in this chap-
ter? They are not. We began this chapter talking about innovation. The preceding
comments in this paragraph talk about creativity, basic research, and scientific
knowledge . This entire subsection deals with the topic ofcreativity. There are
important differences among creativity, invention, and innovation, as we will
describe. Subsequent sections will address the topics of invention and innovation.
Creativity consists ofthree components : expertise, imaginative thinking, and
intrinsic motivation (Hughes et al., 1993). Expertise serves as the bedrock for
creativity. A little knowledge provides the same stimulus to the mind that a
small clump of "sourdough starter" adds to a damp lump of kneaded bread
dough. Imaginative thinking is just another name for creativity. And intrinsic
motivation can strongly promote and influence the creative acts of individuals .
Stated bluntly, creative individuals are intelligent, imaginative, and motivated.
If any one of these three traits is missing, creativity is not likely to flourish .
There is a difference between one's creative ability and the innovativeness
of one's output (Pelz and Andrews, 1976) . Creativity is a personal trait defined
as the ability to envision associations between ideas and/or objects that others
do not see. This creativity, however, is merely an input to the creative process.
Creative outputs are products with varying degrees of innovativeness and pro-
ductiveness . Innovative outputs create new possibilities. Productiveoutputs yield
advances along an already existing path. Innovation equates to Juran's "break-
through" thinking . Productivity equates to continuous improvement, which is
sometimes referred to as kaizen.
DEFINITIONS AND DISTINCTIONS 225
link between Six Sigma and creativity is that Six Sigma can improve the over-
all process of nurturing creativity.
The measurement ofcreativity within R&D in terms ofthebalanced scorecard
as defined by Kaplan and Norton (1996) could lead to the list below. In this
case we are using an R&D example, but the :name principles apply to any cre-
ative function marketing, stock analysis, and so forth.
Invention
One measure of whether or not someone has invented something is whether or
not it can be patented . To qualify for a patent wi invention must be "new, useful,
and non-obvious" (Lubar, 1990). Sternberg and Lubart (1995) use the terms
"novel and appropriate." The National Science Foundation (NSF) prefers the
term "applied research" to the term "invention," but the terms are synonymous .
For several decades, the NSF (1989) has stated that the aim of applied research
is "gaining the knowledge or understanding [necessary] to meet a specific, rec-
ognized need." By contrast, the NSF states that "the objective ofbasic research
is to gain more comprehensive knowledge or understanding of the subject un-
der study, without specific applications in mind."
Comparing these two definitions, we again see the dichotomy between cre-
ating knowledge and inventing things. Creativity is totally enamored with dis-
covery and couldn't careless about the utility,ofwhat is learned . Artistic people
create that they might have joy. Likewise, invention is enamored with creative
prodding and probing but acquires a structural skeleton in response to "cus-
tomer" needs frequently a physical need. If we look ahead to the process of
innovation, we find the skeleton of invention. now covered by a tough layer of
rawhide in response to the abrasive human clawing and clamoring for utility.
This layer of skin tends to hide much of tree underlying creativity that gave
DEFINITIONS AND DISTINCTIONS 227
innovation its initial breath of life. The prototypes and artifacts of innovation
must be useful or they are discarded as useless .
Managing invention is incredibly difficult (Buderi, 1993) . Success on com-
plex tasks still depends on the talents of a few key individuals . Identifying those
key individuals is also difficult, because inventors tend to excel in many disci-
plines . Their cross-functional knowledge seems to be the source of their unique
ability to tackle difficult problems. Most organizations excel in those areas they
value and measure . Inventive companies reward inventive behavior; hence pat-
ents. If an organization wants to be inventive, it needs to measure and reward
inventive behavior.
Below is a sample "invention" balanced scorecard as defined by Kaplan and
Norton (1996) . This is a university faculty example the Department of Busi-
ness Administration in a College of Business . For the sake of brevity, we offer
only one metric per category.
Financial : Average weekly contact hours per faculty member spent with
students. This metric needs to be normalized by an instructor's course load.
Customer: The faculty's median, student course evaluation scores on the
question "My instructor is an effective and accomplished educator" (1-10
scale) . This median score will be compared with that ofall other academic
departments . A goal might be a ranking above the 90th percentile .
Internal: Average annual number of published journal articles or invited
conference presentations per faculty member addressing the topic of "im-
proving the quality of higher education" .
Innovation and Learning: The 3-year running average percentage of the
faculty that take sabbaticals working in for-profit businesses for at least
six months .
We should stress again how difficult it is to find, mentor, and manage inven-
tive personalities . "No great breakthroughs have ever been achieved by reason-
able men" (Magaziner and Patinkin, 1989) . In the example of a university
professor described above, it may be impossible for "unreasonable men/women"
to achieve tenure. It also takes an extremely accomplished leader to manage
creative, inventive, innovative people .
Innovation
Innovation is the creation of useful artifacts . The NSF refers to artifacts as pro-
totypes. These artifacts are initially too crude or too symbolic to be of immedi-
ate interest to an external customer. The innovations that flow from the R&D
process do not necessarily result in commercial products. They are a necessary
but insufficient condition for commercialization . A few artifacts flourish, while
the majority do not survive .
228 NEED FOR CREATIVITY, INVENTION, AND INNOVATION
vations made by Michael Schrage (2000) in his book Serious Play. Anyone
interested in Six Sigma should read this book, and specifically the portions
dealing with the "innovation and learning" aspects of Six Sigma activities. We
do not put quotes around the next two paragraphs, but much of it is paraphrased
from Schrage's book.
Wheneveran organization wants to enhance its innovative culture and prac-
tices, it soon discovers that it must significantly change how it prototypes. One
must literally change how people play with the prototypes they build. An
organization's prototyping methods determine, to a large degree, its ability to
profit from innovation. Except for the unforgiving veracity of the marketplace,
there is no better way to promote value-creating behavior than through the use
of prototypes . And as prototyping media become cheaper and more powerful, it
becomes increasingly difficult for management to resist the desire to explore
one more design option, to playjust a little bit longer. An organization's ability
to create customer and shareholder value now depends on its ability to use its
prototyping tools effectively. It is not a question of more and better informa-
tion, or effective management and sharing of the knowledge we already have.
Rather, we must determine the kinds of prototypes we need to create in order to
create new value for the customer. Some experts are claiming that the number
and quality of prototypes an organization develops and interrogates are directly
proportional to the ultimate quality of the product under development . And the
recent explosion of computational choice and speed, and the array of modeling
options at everyone's disposal, make the organizational culture the critical suc-
cess factor in managing innovation .
Virtually all product and service innovation results fromthe constant struggle
between the customer's wish list (specifications) and the prototypes built to
articulate those desirable traits . The ongoing corporate dialogue between speci-
fications and prototypes is the heart and soul of innovative design . But even the
most technically accurate model cannot guarantee effective communication or
productive interaction . There are also profound differences between organiza-
tions that build prototypes primarily to create questions and those that do so to
answer questions. Once again, culture has an important impact. As a rule, the
technical beauty of a final product is a function of the number of prototypes
tested and the prototyping cycles per unit of time. Prototyping cycle time is
becoming an important metric in today's competitive environment. This em-
phasis on time is radically redefining what a good prototyping culture looks
like. Some organizations have moved on to "periodic prototyping," which re-
quires design teams to produce a given number of prototypes on a fast, fixed
schedule. One suggested indicator of the health and well-being of an internal
prototyping culture is to measure the time interval between initial creation of
the first prototype and the first "demo" of a subsequent prototype to senior
management. Serious Play contains more about exploring the intimate rela-
tionship between innovation and Six Sigma.
230 NEED FOR CREATIVITY, INVENTION, AND INNOVATION
The Environment
Considerable research has been conducted on the influence ofthe environment
on creativity. Unfortunately, the results draw no definite conclusions . For every
study identifying X as important, one can find another study rejecting X as
important . This is most likely due to the modest but important differences in the
research methods used to evaluate creativity. Sternberg and Lubart (1995) sug-
gest that_this contradictory evidence results from the existence of important,
nonlinear variables involved in these studies that interact with one another in
such a way that different outcomes are reached, or conclusions made, depend-
ing on the magnitude of the variables being studied. For example, supervision
(one common factor in these types of studies) can inhibit creativity if its use is
excessive, but it can be a motivator if its use is reasonable .
Taylor (1961) interviewed 12 department heads in 10 research laboratories .
Ten of the 12 identified the working environment as the most important vari-
able affecting creativity. We hasten to add that department heads may view
creativity differently from practicing scientists and engineers . In another study
of 165 scientists engaged in R&D (Amabile, 1988), a number ofvariables were
consistently identified as being motivators (9) or inhibitors (11) of creativity ;
these results are shown in Table 14.2. We have taken the liberty ofattempting to
align factors in the table that represent opposites of one another. The blank
spaces in the table indicate where such "opposites" were not reported. There
are many, many lists like this in the literature that contain similar conclusions
on this type of research. One aspect of Table 14.2 worth noting is that salary
does not show up anywhere on the list. One should not conclude that salary is
not important to scientists and engineers . Rather, a firm's technical staff con-
232 NEED FOR CREATIVITY, INVENTION, AND INNOVATION
Motivators Inhibitors
A number of the motivators listed in Table 14.2 are consistent with a Six Sigma
strategy . Scientists and engineers want challenging problems, a sense of urgency,
collaboration across functions, sufficient resources, encouragement of new ideas,
and recognition . All of these motivators should be present on Six Sigma projects .
The motivator "sufficient time to think" may pose somewhat of a problem for
scientists, but it should be much less of a problem for engineers. Scientists like
to read the literature, go to conferences, ponder problems, play in the lab, experi-
ment, and so forth. Those are all appropriate in the pursuit of R&D . However, we
are not suggesting that an organization reenginecr the R&D process as an S 4
project. Rather, we want scientists to improve the new-product development
process . Much of that work will require scientists to work on S4 projects that focus
R&D,
on the support activities associated with as delineated earlier.
We will leave this topic with a quotation from Warren Bennis : "There are
two ways to be creative . One can sing and dance . Or one can create an environ-
ment in which singers and dancers flourish ." Six Sigma activity requires sig-
nificant amounts of both . Creative leaders provide creative environments for
all workers, including themselves .
ENCOURAGING CREATIVITY 233
Intelligence
We have been focusing on the effect of the environment on creativity. We will
finish this section by briefly commenting on creativity's dependence on the
additional factors of intelligence, knowledge, personality, and motivation.
There is considerable evidence that some minimal level of intelligence is
necessary for creativity, but intelligence alone is an insufficient condition for
creativity (Barron and Harrington, 1981). Standardized tests to measure one's
intelligence quotient (IQ) are not reliable predictors of creative ability. They
suffer from cultural bias and many other shortcomings . IQ tests may be the best
thing we have for screening high school graduates for admission to colleges,
but they are not very effective atpredicting on-the-job success . The best predic-
tor of future creative behavior in a domain is past creative behavior in that same
domain (Barron and Harrington, 1981) .An accomplished professional acquain-
tance of one of the `authors is a medical doctor with undergraduate and
postdoctoral studies in mathematics . Over decades of professional work in
mathematics and modeling, he has observed that the best predictor of one's
ability to publish seminal work is that person's past history of such publication
(Albanese, 1997) .
It is humbling to consider Wiener's (1993) contention that a truly creative
idea "is, to a large extent a lucky and unpredictable accident." If this is true, it
has tremendous implications for organizational creativity. If senior leaders do
not create an environment in which serendipitous events can be pursued by the
R&D staff and other creative agents, then many potentially great ideas will be
lost possibly to the competition.
We hasten to interject the concept of emotional intelligence (EQ, as con-
trasted with the traditional measure of intelligence, IQ. EQ is a relatively new
concept to the business community (Coleman, 1997). In the future, a robust
economy will provide job opportunities for individuals with IQ as well as those
with EQ (Cox and Alm, 1999) . Fortunately, both IQ and EQ skills can be en-
hanced by education and training.
Knowledge
In a knowledge-based society an organization's most important asset is the
knowledge worker. There are two types ofknowledge, explicit and tacit. Nonaka
and Takeuchi (1995), in their book The Knowledge-Creating Company, offer
insights that go a long way toward creating an environment that is conducive to
an S4 strategy.
In short, the Western world views knowledge as primarily formal and sys-
tematic, the stuff of books and journals. By contrast, the Japanese and many
other Eastern societies view words and numbers as the tip of the knowledge
iceberg . They view knowledge as primarily tacit in nature. And tacit knowledge
23 4 NEED FOR CREATIVITY, INVENTION, AND INNOVATION
has two major components : A brain that contains know-how guides the skilled
hands of the craftsman. This knowledge is difficult to explain, or document in
writing. The second component consists of the internalized mental models and
belief systems that shape the way we view the world. To a large extent our
worldview colors our vision of what is possible and not possible. One of the
challenges facing any serious S4 implementation is learning how to extract this
tacit knowledge from the minds of one's knowledge workers. It is easy to inter-
rogate a process using control charts and designed experiments . It is another
thing to extract and share the organizational wisdom that is so essential to the
measurement and analysis phases of S4 implementation :
"What raises living standards over time isn't companies vying for customers
in existing industries . It's competition from new goods and services and new
production techniques, bringing about new industries, new jobs, and higher
living standards" (Cox and Alm, 1999). Only the accessible knowledge (and
wisdom) of knowledge workers can deliver on this promise .
Personality
It is difficult to decide where to begin on such a nebulous concept as personal-
ity, and yet we all know that personality exists, that personality differences
exist, and that these differences can be striking. A number of authors define
personality as a preferred way of interacting with the environment. The idea
that personality is all genetically based and hence unchangeable is flawed. Per-
sonality is at least partially under the control of the individual. A person who
wants to develop any specific personality trait(s) can be successful if he or she
is committed to the task . On the other hand, anyone who has been around an
obsessive-compulsive personality knows how "stuck" these individuals seem
to be in that mode of behavior. Some people undergo psychotherapy for years
in an attempt to deal with what their environment defines as a personality disor-
der. To the extent that certain mental states are chemically (genetically) based,
they may require medical attention . To the extent that they represent personal
preferences based on learned behavior, they can be modified with sufficient
motivation.
We will begin by presenting the list of "creative" personality traits posited
by Sternberg and Lubart (1995) ; see Table 14.3 . We will not discuss this list.
We present it so the reader might compare it with Csikszentmihalyi's (1996)
list discussed below. One should also consider Margaret Boden's (1991) per-
spective on creativity: she argues convincingly that any theory of creativity
must include a model of conceptual computation by the mind.
ENCOURAGING CREATIVITY 235
Our hope is that this overview will help the reader more readily identify
truly creative people within organizations . More important, we hope the reader
will come to appreciate and value the perspective that these sometimes misun-
derstood and mislabeled individuals can contribute to an S 4 work environment .
Mihaly Csikszentmihalyi is a university professor who has made important
contributions in the area of creativity. He views creativity as one of many prop-
erties of a complex system the mind. Complex systems cannot be explained
by observing individual components of the system . Csikszentmihalyi (1996)
believes that creative individuals can be identified by the complexity that they
exhibit. Less-creative individuals presumably have less of this complexity, pos-
sibly none at all. The complexity that creative individuals exhibit shows up in
the form of dichotomous behaviors or traits . Whereas most people have person-
alities that are strongly unipolar, creative people tend to be bipolar in a number
of traits . This bipolarity expresses itself in the form of dichotomies . Creative
people have personality traits that are complex and confusing to the rest of us
because their personalities seem to be dual in nature .
For example, creative people tend to display a great deal of physical energy,
and yet at other times are found to be quiet and relaxed. A number of scientists
have reported "spontaneously" discovering answers to difficult questions after
periods of reflection and contemplation. Some people postulate that these spon-
taneous insights are a result of parallel brain processing that is occurring during
rest as well as work . Most creative people, including nonscientists, have work
patterns that allow for periods of physical inactivity and mental contemplation.
So one should not make any rash judgments upon entering a laboratory and
finding a senior scientist with her feet on the desk, looking out the window with
an inquisitive grin . High levels of physical activity at work often impress man-
agement . Are these people demonstrating creativity? Or are they producing a
lot of undesirable product very quickly? Long-term patterns of performance
can provide the answer. Do not be surprised if certain members of S 4 project
236 NEED FOR CREATIVITY, INVENTION, AND INNOVATION
teams vacillate between frenzied activity and immobility. Some of the prob-
lems these S4teams undertake demand as much creativity as a Picasso painting.
Creative people tend to be smart and naive at the same time. They seem to be
smart in a specific domain but naive in many other areas involving common
sense. In the world of Six Sigma, naivete may be more important than mastery
of facts and figures. Frequently a "dumb" question yields the insight needed to
solve a problem. S4 practitioners must be careful in their role as team leaders
not to embarrass team members who ask naive questions about statistical mat-
ters. The chances are very good that if one team member has a question, the
others have similar doubts, concerns, or confusion. Why do we use rational
subgroups of size four or five when creating control charts? Why don't factor
effects get confused when using full factorial designs involving all possible
combinations of factors? Smart people also have the ability to `propose novel
solutions using knowledge rather than naivete. This skill of envisioning solu-
tions or representations of problems is found in many individuals who have the
ability to think both divergently and convergently. Convergent thinking is the
traditional mode taught in school, focusing on finding the one right (best?)
answer. Divergent thinking looks for multiple solutions to a single problem
before winnowing the field to just one answer. There is considerable evidence
suggesting that S4 project teams will come up with better (more creative) solu-
tions if they are asked to provide multiple solutions (divergent thinking) rather
than a single optimum solution (convergent thinking).
Creative people possess the wonderful duality of playfulness and discipline.
The playfulness is a reflection of the childlike innocence these people possess.
Creative people withhold judgment, because in. their innocence there is no sin.
It is this playful, nonjudgmental behavior that gives creative people the capac-
ity to see possibilities that have been lobotomized in others. Do not be surprised
if the fun-loving jokester on an S4 project team is also the one who quietly
works alone until midnight trying to figure out what' the data really mean. Give
team members the freedom to be occasionally sarcastic or emphatically goal
oriented. Sarcasm, after all, can be an ironic form of humor. And effective goal
setting is demanded by most S4 activity.
Creative individuals can be highly imaginative or deeply grounded in real-
ity. A distinguishing characteristic between creative and noncreative people is
that creative people are good at coming up with highly original ideas that are
not bizarre . Normal people's ideas are rarely original but often bizarre . And as
with all ofthe dichotomies we are discussing, creative people have to be able to
switch between modes with relative ease.
Creative people seem to be both extroverted and introverted at the same
time. If one desires to stand apart from the pack in any field of endeavor, one
has to study, emulate, and converse with the current giants in that field. Con-
sider how you will work with both introverts and extroverts on an S4 project
team. Also consider how to effectively harness the creativity of the rare intro-
ENCOURAGING CREATIVITY 237
Motivation
Motivation -is defined as the driving force that leads someone to action. Pelz
and Andrews (1976) interviewed scientists in an attempt to identify their pri-
mary sources of motivation for engaging in the pursuit of science and discov-
ery. One interesting finding was that high performers were inclined toward a
scientific discipline, not lifelong employment as an "organization person ." Those
scientists who relied on their own ideas for motivation were highly effective .
Those who relied on their supervisors for stimulation were below-average
performers . And lastly, these scientists reported much higher motivation when
engaged in "broad mapping ofnew areas" as opposed to probing "deeply into a
narrow area."
238 NEED FOR CREATIVITY, INVENTION, AND INNOVATION
Koestler (1964) claims that man not only responds, to the environment but
also interacts with it by asking questions . He claims man's primary motivations
for prodding and probing the environment are novelty, surprise, conflict, and
uncertainty.
Shekerjian (1990) makes the interesting observation that the greater one's
mastery of a discipline, the less motivation there is to generate new approaches
to problem solving. This suggests that regular, if not frequent, transitions into
new disciplines should be highly motivational . It is certainly consistent with
Koestler's point that man loves to tinker in areas rich in novelty and surprise .
Could it be that world-class "familiarity" breeds confidence, expertise, status,
and laziness?
Gretz and Drozdeck (1992) assert that most people are motivated by needs,
and that the three needs that most commonly drive human behavior are the
needs for power, association with others, and achievement.
So what should motivate S4 project teams to work diligently to increase cus-
tomer satisfaction and improve organizational profitability? First we must make
the point that an S4strategy is, aboveall else, a major organizational investment
in people. The infrastructure requirements to enable and maintain an effective
S4 strategy are driven by investments in human capital . Cox and Alm (1999)
claim that "the rewards of investing in human capital go to the workers them-
selves in the form of fatter paychecks and to companies in the form of higher
productivity and greater profits ." As global competition increases and the rate
of technological change accelerates, there will be tremendous social turmoil as
the uneducated and unprepared secede from the expanding economy. Those
with the capability and desire to keep pace with radical change will find virtu-
ally limitless opportunity for variety, growth, and prosperity.
LIST OF SYMBOLS
Control chart: A statistical tool used to track an important condition over time
and watch for changes in both the average and the variation of that condi-
tion. There are two general types of control charting to use, depending on
the type of data being plotted . The two types of data are attribute (count)
data and variables (continuous) data. Attribute data are typically associated
with a p chart, np chart, c chart, and u chart. Variables data are associated
with XmR charts, x and R charts, and x and S charts . See Chapter 10 of
Breyfogle (1999), Implementing Six Sigma, for more details on how to se-
lect the proper control chart for a given situation and type of data.
Control plan: A description ofthe inputs to a process that should be monitored
or error-proofed for the purpose of maintaining satisfactory output for a
KPOV.
Correlation: The determination of the effect of one variable upon another in a
dependent situation .
Correlation coefficient (r): A statistic that describes the strength of a linear
relationship between two variables is the sample correlation coefficient . A
correlation coefficient can take values between -1 and +1 . A -1 indicates
perfect negative correlation, while a +1 indicates perfect positive correla-
tion. A zero indicates no correlation.
Cost of poor quality : Cost-of-quality issues often are given the broad catego-
ries of internal failure costs, external failure costs, appraisal costs, and pre-
vention costs .
Critical to quality (CTQ) : A term widely used by General Electric in its Six
Sigma activities, which describes an element of a design, characteristic of a
part, or attribute of a service that is critical to quality in the eyes of the
customer. Formerly known as key quality characteristics . Similar to the KPOV
term used within this book. Within GE, sometimes referred to as "Y."
Customer : Someone for whom work or a service is performed . The end user of
a product is a customer of the employees within a company that manufac-
tures the product. There are also internal customers in a company. When an
employee does work or performs a service for someone else in the company,
the person who receives this work is a customer of this employee.
Deductive reasoning: The act of reasoning
_Working that starts from the general and
moves toward the more specific. from general knowledge, it is of-
ten possible to hypothesize more detailed or specific actions, outcomes, ef-
fects, or consequences that would flow from this general knowledge . The
act of engaging in this kind of thinking is called deductive reasoning . See
also Inductive reasoning .
Defect: Afailure to meet an imposed requirement on a single quality character-
istic, or a single instance of nonconformance to the specification .
Defective : A unit of product containing one or more defects .
24 6 GLOSSARY
Defects per million (DPM): A defect rate expressed as the number of defect
occurrences observed per million units produced. Also referred to as the
parts-per-million (ppm) defect rate.
Defects per million opportunities (DPMO): Rather than calculating defect
rates based on the number of final units that pass or fail testing, some orga-
nizations estimate how many opportunities there are for failure in a given
final product. When this approach is used, DPMO is defined as the number
of defects counted, divided by the actual number of opportunities to make a
defect, then multiplied by one million. This measure can be converted di-
rectly into a sigma quality level.
Defects per unit (DPU) : The number of defects counted, divided by the num-
ber of products or characteristics produced.
Degrees of freedom (df or v): Number of measurements that are indepen-
dently available for estimating a population parameter. For a random sample
from a population, the number of degrees of freedom is equal to the sample
size minus one .
Design of experiments (DOE) : A method of experimentation in which pur-
poseful changes are made to the inputs to a process (factors) in order to
observe the corresponding changes in the outputs (responses).
Design for manufacturing : A set of design techniques for creating products
that can be produced and assembled easily, thus improving quality while
reducing cost.
Design for Six Sigma (DFSS): Designing and creating a component, system,
or process with the intent of meeting and/or exceeding all the needs of cus-
tomers and KPOV requirements upon initial release . The goal of DFSS is
that there be no manufacturing issues with the design upon the initial release
of the design to manufacturing .
Distribution : A pattern that randomly collected numbers from a population
follow. The normal, Weibull, Poisson, binomial, and log-normal distribu-
tions discussed in this book are applicable to the modeling of various indus-
trial situations.
DOE : See Design of experiments .
Effect : That which is produced by a cause .
Entitlement: The expected performance level of a product or process when its
major sources of variation are identified and controlled.
Error (experimental) : Ambiguities during data analysis from such sources as
measurement bias, random measurement error, and mistake .
Evolutionary operations (EVOP) : An analytical approach whereby process
conditions are changed structurally in amanufacturing process (e.g., using a
fractional factorial experiment design matrix) for the purpose ofanalytically
determining changes to make for product improvement .
GLOSSARY 24 7
(see Shewhart Control Charts) observed empirically that three standard de-
viations below the process mean or median is the most appropriate value for
the LCL.
Lower specification limit (LSL): The lowest value of a product dimension or
measurement that is acceptable to the customer.
Main effect: An estimate of the effect of a factor measured independently of
other factors .
Management by fact: Management that bases decisions and actions on verifi-
able data rather than by instinct, opinion, or hunch.
Mean : The mean of a sample (x) is the sum of all the responses divided by the
sample size. The mean of a population (g) is the sum of all responses of the
population divided by the population size. In a random sample of a popula-
tion, x is an estimate of g for the population .
Measurement systems : The complete process ofobtaining measurements. This
includes the collection of equipment, operations, procedures, software, and
personnel that affects the assignment of a number to a measurement
characteristic .
Measurement systems analysis : See Gage repeatability and reproducibility
(R&R).
Minitab : Statistical analysis software that is often used within Six Sigma train-
ing.
Multi-vari chart: A chart that is constructed to display the variance within
units, between units, between samples, and between lots.
Nonconformity : A condition observed within a product or process that does
not conform to one or more customer specifications . Often referred to as a
defect. A nonconforming part or process may contain more than one non-
conformity.
Normal distribution : Abell-shaped distribution that is often useful to describe
various physical, mechanical, electrical, and chemical properties .
Normal variation : See Common cause variation .
np chart: See Control chart.
Null hypothesis (Ho) See Hypothesis testing .
One-at-a-time experiment: An individual tries to fix a problem by making a
change and then executing a test. Depending on the findings, something else
may need to be tried. This cycle is repeated indefinitely.
Outlier: A data point that does not fit a model because of an erroneous reading
or same other abnormal situation.
Pareto chart: A graphical technique used to quantify problems so that effort
can be expended in fixing the "vital few" causes, as opposed to the "trivial
many." Named after wufredo Pareto, a European economist.
250 GLOSSARY
Pareto principle: Eighty percent (the trivial many) of the trouble results from
20% (the vital few) of the problems.
Passive data analysis : Data are collected and analyzed as the process is cur-
rently performing. Process alterations are not assessed .
p chart: See Control chart.
Point estimate : An estimate calculated from sample data without a confidence
interval.
Population : The totality ofitems under consideration . Often referred to as the
"universe ."
Probability : The chance of something's happening .
Problem solving : The process of determining the cause from a symptom and
then choosing an action to improve a process or product.
Problem statement: A statement that describes in specific and measurable terms
what is wrong; and the impact potential if the problem is not fixed.
Process : A method to make or do something that involves a number of steps.
Process capability: A normal range of performance of a process when it is
operating in a state of statistical process control.
Process capability indices (Cp and C k): C is a measurement of the allowable
tolerance spread divided by the actual 6 6 data spread . C k has a ratio similar
to that of C except that this ratio considers the shift of the mean relative to
the central specification target .
Process flow diagram (chart) : Path of steps of work used to produce or do
something .
Process spread : The range of values that a given product or process character-
istic displays. This term most often applies to the range, but it could also be
used in relation to variance .
Qualitative factor: Afactor thathas discrete levels . For example, product origi-
nation with factor levels of supplier A, supplier B, and supplier C.
Quantitative factor: Afactor that is continuous. For example, a product can be
manufactured with a process temperature factor between 50°C and 80°C.
Quality function deployment (QFD): A technique that is used, for example,
to get the "voice of the customer" in the design of a product .
Random effects (or components of variance) model : An experiment where
the variance of factors is investigated (as opposed to a fixed effects model).
Random causes: See Common cause variation .
Range : For a set of numbers, the absolute difference between the largest and
smallest value.
Regression analysis : A statistical technique for determining the relationship
between one response and one or more independent variables .
GLOSSARY 25 1
Weibull distribution : This distribution has a density function that has many
possible shapes . The shape parameter (b) and the location parameter (k) de-
scribe the two-parameter distribution. This distribution has an x-intercept
value at the low end of the distribution that approaches zero, which yields a
zero probability of a lower value. The three-parameter has, in addition to the
other parameters, the location parameter (xo), which is the lowest x-intercept
value.
XmR charts : Also known as Individuals Moving Range (ImR charts) . See Con-
trol chart.
x & R charts : See Control chart.
x & S charts : See Control chart.
"X's": The designation for those variables that are the root causes of effects
observed in products or processes ("explanatory variables" in regression
analysis) . S4 focuses on measuring and improving X's so as to promote sub-
sequent improvements in Y's . Also calledindependent variables, orfactor(s)
in a designed experiment.
"Y's": The designation for those variables that are the observed effects or out-
comes resulting from changes to the input variables affecting a product or
process . We generally assume that these Y's are the result of one or more X's
expressing themselves in a product or process . Also called dependent vari-
ables, or response(s) in a designed experiment.
Yield, Final: The ratio of acceptable product at the end of a process to the
initial number product that was started initially with the process . Often this
ratio is multiplied by 100 and expressed as a percentage value. For example,
a yield of 0.857 equates to an 85.71/1a yield.
Yield,.Rolled Throughput : See Rolled throughput yield.
REFERENCES
AIAG (1995), Statistical Process Control (SPC) Reference Manual, Chrysler Corpora-
tion, Ford Motor Company, General Motors Corporation .
Albanese, Richard A. (1997), personal communication .
Allen, Thomas J. (1977), Managing the Flow of Technology, MIT Press, Cambridge,
MA.
Amabile, T.M . (1988), A Model of Creativity and Innovation in Organizations, Re-
search in Organizational Behavior 10 : 123-167.
American Productivity & Quality Center (1988), Measuring White Collar Work, Hous-
ton, TX .
Aviation (1998), "Success with Six Sigma Often an Elusive Goal," Aviation Week and
Space Technology, vol. 139, no . 20, p. 53 .
Barron, F. and Harrington, D.M. (1981), Creativity, Intelligence, and Personality. An-
nual Review of Psychology 32: 439-476.
Bergquist, Timothy M. and Ramsing, Kenneth D. (September 1999), Measuring Perfor-
mance After Meeting Award Criteria : Study Compares Perceived Success to Finan-
cial Data of Award Winners and Applicants . Quality Progress 32(9): 66-72.
Bisgaard, Soren (1991), Teaching Statistics to Engineers. TheAmerican Statistician 45(4):
pp . 274-283 .
Blakeslee, Jerome A. Jr. (July 1999), Implementing the Six Sigma Solution: How to
Achieve Quantum Leaps in Quality and Competitiveness. Quality Progress 32(7):
77-85.
Boden, Margaret A. (1991), The Creative Mind, Basic Books, New York, NY
Botkin, James, Dimancescu, Dan, and Stata, Ray (1984), The Innovators, Harper &
Row, New York, NY.
25 6 REFERENCES
Box, George (1991), Teaching Engineers Experimental Design With a Paper Helicop-
ter, CQPI Report 76, Center for Quality and Productivity Improvement, University
. of Wisconsin at Madison, Madison, WI.
Box, G.E .P., Bisgaard, S. and Fung, C. (1988), An Explanation andCritique of Taguchi's
Contributions to Quality Engineering, Quality and Reliability Engineering Interna-
tional, 4(2) : 123-131 .
Box, G.E .P, Hunter, W.G., and Hunter, S.J . (1978), Statistics for Experimenters, John
Wiley, New York, NY
Bouckaert, Geert, and Balk, Walter (Winter 1991), Public Productivity Measurement :
Disease and Cures, Public Productivity & Management Review, pp. 229-235.
Boyett, Joseph H., Kearney, A. T., and Conn, Henry P. (1992), What's Wrong with Total
Quality Management, Tapping the Network Journal, pp . 10-14.
Bradstreet, Thomas E. (1996), Teaching Introductory Statistics Courses So That
Nonstatisticians Experience Statistical Reasoning, TheAmerican Statistician 50(1):
69-78.
Breyfogle, Forrest W. III (1999), Implementing Six Sigma : Smarter Solutions Using
Statistical Methods, John Wiley & Sons, New York, NY
British Deming Association (1991),A System ofProfound Knowledge (Number9), SPC
Press, Knoxville, TN.
British Deming Association (1990), Profound Knowledge (Number6), SPC Press, Knox-
ville, TN.
Brooks, Frederick P Jr. (1975), The Mythical Mon-Month: Essays on Software Engi-
neering, Addison-Wesley, Reading, MA .
Brown, Mark G., and Svenson, Raynold A. (July/August 1988), Measuring R&D Pro-
ductivity, Research Technology Management, pp . 39-43.
Bryce, G. Rex (1992), Data Driven Experiences in an Introductory Statistics Course for
Engineers Using Student Collected Data, 1992 Proceedings of the Section on Statis-
tical Education, pp. 155-160, American Statistical Association, Alexandria, VA .
Buderi, Robert (Jan . 18, 1993), American Inventors are Reinventing Themselves, Busi-
ness Week, pp . 78-82.
Carnegie Commission on Science, Technology, and Government (December 1992),
Environmental Research and Development: Strengthening the Federal Infrastruc-
ture, p. 42 .
Chang, Richard Y (1994), Creating High-Impact Training, Richard Chang Associates,
Irvine, CA. .
Chappell, Tom (1999), Managing Upside Down, The Seven Intentions of Values-Cen-
tered Leadership, William Morrow and Company, Inc., New York.
Cheek, Tom (1992), Six Sigma Black Belt Certification, Signed by A. W. Wiggenhorn,
President, Motorola University, andMikel Harry, Ph.D ., Corporate Director Six Sigma
Research Institute, September 25, 1992.
Cobb, George (1991), Teaching Statistics : More Data, Less Lecturing, Amstat News,
vol. 182, pp . 1, 4.
Collier, D.W. and Gee, R.E . (May 1973), A Simple Approach to Post-Evaluation of
Research, Research Management, pp. 12-17.
REFERENCES 25 7
Folkman, Joe (1998), Employee Surveys That Make a Difference, Using Customized
Feedback Tools to Transform Your Organization, Executive Excellence Publishing,
Provo, UT.
Franklin, LeRoy A., Cooley, Belva J., and Elrod, Gary (October 1999), Comparing the
Importance of Variationand Mean of aDistribution, Quality Progress 32(10) : 90-94.
Gabel, Stanley H. and Callanan, Terry (May 19, 1999), Six Sigma at Eastman. Kodak
Company, 1999 ASA Quality and Productivity Conference, American Statistical
Association, Schenectady, NY.
General Electric (1997), General Electric Company 1997 Annual Report.
General Electric (1996), General Electric Company 1996 Annual Report.
Gilman, John J. (July 1993), personal communication .
Gilman, John J. (1992), Inventivity: The Art and Science of Research Management, Van
Nostrand Reinhold, New York, NY
Gitlow, Howard S. and Gitlow, Shelly J. (1987), The Deming Guide to Quality and
Competitive Position, Prentice-Hall, Englewood Cliffs, NJ.
Godfrey, A. Blanton (May 19, 1999), Six Sigma Quality: From Strategic Deployment to
Bottom-Line Results, 1999 ASA Quality and Productivity Conference, American
Statistical Association, Schenectady, NY
Godfrey, A. Blanton (December 1999), Building a Scorecard, Quality Digest 19(12):
16 .
Goleman, Daniel P (1997), Emotional Intelligence : Why It Can Matter More Than IQ,
Bantam Books, New York, NY
Gomory, Ralph (June 1990), Essays: Of Ladders, Cycles and Economic Growth, Scien-
tific American, p. 140.
Gretz, Karl F. and Drozdeck, Steven R. (1992), Empowering Innovative People, Probus
Publishing, Chicago, IL.
Gunter, Bert (1993), Through a Funnel Slowly with Ball Bearing and Insight to Teach
Experiment Design, The American Statistician 47(4): 265-269.
Hahn, Gerald J., Hill, William J., Hoerl, Roger W., and Zinkgraf, Stephen A. (August
1999), The Impact of Six Sigma-A Glimpse Into the Future of Statistics, American
Statistician 53(3): 208-215.
Harry, Mikel J. (February 2000a), The Quality Twilight Zone, Quality Progress 33(2):
68-71 .
Harry, Mikel J. (January 2000b), A New Definition Aims to Connect Quality Perfor-
mance with Financial Performance, Quality Progress 33(1): 64-66.
Harry, Mikel and Schroeder, Richard (2000), Six Sigma: The Breakthrough Manage-
ment Strategy Revolutionizing the World's Top Corporations, Currency, New York,
NY
Hendrix, C.D. (April 1991), Signal-to-Noise Ratios: AWasted Effort, Quality Progress,
pp. 75-76.
Hiebeler, Robert, Kelly, Thomas B., Ketteman, Charles (1998), Best Practices, Building
Your Business with Customer Focused Solutions, Arthur & Anderson, Simon &
Schuster, New York.
REFERENCES 259
Hogg, Robert V (1991), Statistical Education : Improvements Are Badly Needed, The
American Statistician 45(4): 342-343.
Hughes, Richard L., Ginnett, Robert C., and Curphy, Gordon J. (1993), Leadership:
Enhancing the Lessons ofExperience, Richard D. Irwin, Homewood, IL.
Hughes, Richard L., Ginnett, Robert C., and Curphy, Gordon J. (1993), Leadership-
Enhancing the Lessons of Experience, Richard D. Irwin, Boston, MA .
Jonash, Ronald S. and Sommerlatte, Tom (1999), The Innovation Premium, Arthur D.
Little, Inc., Perseus Books, Reading, MA .
Jones, D. (1998), FirmsAir for Six Sigma Efficiency, USA Today, July 21, 1998 Money
Section.
Juran, Joseph - M. (1999a), personal communications : www.juran .com/drjuran/
bio-jmj.html.
Juran, Joseph M. and Godfrey, A. Blanton (1999b), Juran's Quality Handbook (Fifth
Edition), McGraw-Hill, New York, NY
Juran, J.M. and Gryna, Frank M. (1988), Juran's Quality Control Handbook (Fourth
Edition), McGraw-Hill, New York, NY.
Juran, J.M . (1992), Juran on Quality by Design: The New Stepsfor Planning Quality
into Goods and Services, Free Press, New York, NY.
Juran, J.M . (1989), Juran on Leadership for Quality: An Executive Handbook, Free
Press, New York, NY.
Juran, JAM. (1988), Juran on Planning for Quality, Free Press, New York, NY.
Juran, J.M . (1964), Managerial Breakthrough, McGraw-Hill, New York, NY.
Jusko, Jill (6 Dec 1999), A Look at Lean, Industry Week.
Kaplan, Robert S. and Norton, DavidP. (1996), The Balanced Scorecard, Harvard Busi-
ness School Press, Boston, MA.
Kaplan, Robert S. and Norton, DavidP (Sep/Oct 1993), Putting the Balanced Scorecard
to Work, Harvard Business Review 71(5): 138-140.
Kaplan, Robert S . and Norton, David P. (Jan/Feb 1992), The Balanced Scorecard-
Measures That Drive Performance, Harvard Business Review 70(1): 71-79.
Kiemele, M.J . (1998), Information presented in this paragraph was contributed by M.J .
Kiemele, Ph .D ., of Air AcademyAssociates .
King, B. (1987), Better Designs in Half the Time, Implementing QFD in America, Goal/
QPC, Methuen, MA.
Koestler, Arthur (1964), The Actof Creation, Viking Penguin, New York, NY
Kuhn, Thomas S. (1970), The Structure of Scientific Revolutions (Second Edition, En-
larged), University of Chicago Press, Chicago, IL .
Latzko, William J. andSaunders, DavidM. (1995), Four Days with Dr. Deming, Addison-
Wesley, Reading, MA .
Levitt, Theodore (1991), Thinking About Management, Free Press, New York, NY
Lobbestael, Wayne and Vasquez, Bud (May-June 1991), Measure to Improve: Ideas on
Implementing Measurement, Program Manager, pp. 39-44.
Lowe, J. (1998), Jack Welch Speaks, John Wiley & Sons, New York, NY
260 REFERENCES
Pyzdek, Thomas (November 1999b), Why Six Sigma Is Not Enough, Quality Digest,
p. 26.
Pyzdek, Thomas (June 1999c), Six Sigma Is Primarily a Management Program, Quality
Digest, p. 26 .
Quality Sourcebook (January 2000), Quality Digest 20(1): 109-127.
Robbins, Anthony (1991), Awaken the Giant Within, Fireside, NY
Ross, P.J . (1988), Taguchi Techniques forQuality Engineering, McGraw-Hill, NewYork,
NY
Roussel, Philip A. (1991), Third Generation RED, HarvardBusiness School Press, Bos-
ton, MA.
Scherkenbach, William W. (1988), The Deming Route to Quality and Productivity : Road
Maps and Roadblocks, ASQC Quality Press, Milwaukee, WI.
Schiemann, WilliamA. and Lingle, John H. (1999), BULLSEYE!: Hitting Your Strategic
Targets,Through High-Impact Measurement, Free Press, New York, NY
Schmidt, Stephen R. (1993), Personal Communication, Design of Experiments Work-
shop, University of Texas at Austin, Austin, TX .
Schmidt, S.R., Kiemele, M.J. andBerdine, R.J . (1997), Knowledge BasedManagement,
Air Academy Press & Associates, Colorado Springs, CO .
Schmidt, Warren H. and Finnegan, Jerome P. (1992), The Race Without a Finish Line,
Jossey-Bass, San Francisco, CA, p. 309.
Schrage; Michael (2000), Serious Play, Harvard Business School Press, Boston, MA.
Schrage, Michael (1990), Shared Minds, Random House, New York, NY
Senge, P. M. (1990), The Fifth Discipline : TheArt and Practice of the Learning Organi-
zation, Doubleday/Current, New York, NY
Shekerjian, Denise (1990), Uncommon Genius: How Great Ideas Are Born, Viking,
New York, NY
Slater, Robert (1999), The GE Way Fieldbook, McGraw-Hill Companies, Inc., New York,
Slater, Robert (2000), The GE Way Fieldbook: Jack Welch's Battle Plan for Corporate
Revolution, McGraw-Hill, New York, NY
Snee, Ronald D. (September 1999a), Why Should Statisticians Pay Attention to Six
Sigma? : An Examination for Their Role in the Six Sigma Methodology, Quality
Progress 32(9): 100-103.
Snee, Ronald D. (May 19, 1999b), Why Should We Pay Attention to Six Sigma, 1999
ASA Quality and Productivity Conference, American Statistical Association,
Schenectady, NY
Stahl, Michael J. and Steger, Joseph A. (January 1977), Measuring Innovation and Pro-
ductivity: APeer Rating Approach, Research Management, pp. 35-38.
Sternberg, Robert J. and Lubart, Todd I. (1995), Defying the Crowd: Cultivating Cre-
ativity In a Culture of Conformity, Free Press, New York, NY
Stookey, S.D . (January 1980), ThePioneering Researcher and theCorporation, Research
Management, pp . 15-18.
262 REFERENCES