STRATEGY • MANAGEMENT • COMPETITION

Many of the problems associated with quality

programs are the result of increased complex-
ity. The half-life concept, a new tool, strives to
make complexity more manageable. By doing

Are There Limits to

so, learning is accelerated and improvement
becomes continuous.

TOTAL QUALITY MANAGEMENT?

B Y A R T H U R M . S C H N E I D E R M A N

FTER NEARLY FOUR years were unable to explain their opposi- United States. During the final ques-

A of rapid quality improve-

ment, progress at Analog
Devices Inc. began to
plateau. A diagnosis carried out by the
21-member executive group singled
tion, the group members expressed
their instinctive rejection of the idea.
“My take is that that’s exactly the
wrong thing to do,” one of them said.
“I don’t have the time to waste on
tion-and-answer period, I asked the in-
structor whether he felt there were
any problems that were not solvable
using T.Q.M.’s seven management and
planning tools (the subject of the
out the root cause as lack of demon- something that I know will not work,” training course). His answer was un-
strated commitment to Total Quality another said. Such views were nearly expected. He assured us that there
Management (T.Q.M.) by this very unanimous. I had given them a stan- were no limits to these tools. “In fact,”
same group. As vice president of qual- dard T.Q.M. response, and they had he went on to say, “I believe that we
ity and productivity improvement, I emphatically turned it down. I knew could quickly cure cancer if we could
was expected to come up with possi- that understanding what was behind only get the research community to
ble corrective action. My solution was their instincts to reject T.Q.M. was a use these tools.”
that each member of the group take key to returning us to our path of con- This answer was in sharp con-
the most critical issue facing him and tinuous improvement. trast to a response I had received a
apply T.Q.M. methods and tools for its Shortly after that, I was a guest at month earlier to a similar question.
resolution. The resulting success sto- a two-day T.Q.M. training session held That time I was participating in a meet-
ries would dramatically demonstrate at a company that had won a Malcolm ing of senior executives who were
to the entire organization that top man- Baldrige National Quality Award. The studying and applying Hoshin Kanri
agement practices what it preaches. instructor was an outside consultant (another T.Q.M. tool) in planning and
My suggestion was met with uni- from one of the largest and most pres- implementing breakthroughs in their
versal skepticism. Although they all tigious T.Q.M. organizations in the organizations. Our instructor was a

...............................
Arthur M. Schneiderman is an independent consultant headquartered in Boxford, Mass. He specializes in helping senior man-
agers identify, measure and improve their most critical processes. Mr. Schneiderman was formerly vice president of quality and pro-
ductivity improvement at Analog Devices Inc. Before that he was a consultant with Bain & Company. Mr. Schneiderman is a gradu-
ate of Massachusetts Institute of Technology, holding B.S. and M.S. degrees in mechanical engineering and an M.S. in management
from the Sloan School.

Copyright ©1992-1998 Arthur M. Schneiderman. All rights reserved.

35
Issue 11
 STRATEGY • MANAGEMENT • COMPETITION

months

14 18 22
xity

hi
al Comple

ence with the Analog De- question of whether there exists a

7 9 11 vices executive group. complexity level above which im-
med
tion

provement fails and the half-life be-

Organiza

THE COMPLEXITY MAP comes infinite.

low 1 3 5 In 1988, I introduced the I have continued studying the re-
concept of the “half-life” as- lationship between T.Q.M. and
low med hi sociated with continuous process complexity. I have presented
Technical Complexity
Source: Arthur M. Schneiderman
process improvement.1 Af- the half-life concept at dozens of con-
ter benchmarking nearly ferences and numerous business
EXHIBIT I
100 examples of exemplary school classes. It is the subject of a
TARGET HALF-LIVES
use of the P.D.C.A. cycle Harvard Business School case study2
................... (Plan-Do-Check-Act), I ob- and has been described in several re-
Japanese expert in T.Q.M. The subject served that for each process being im- cent books on the subject of perfor-
of discussion was the applicability of proved, the rate of improvement was mance measurement.3, 4, 5 I have also
the seven Quality Control (Q.C.) tools constant. The (constant) time it took benefited from discussion with many
to some of the more difficult problems for each 50 percent reduction in defect T.Q.M. practitioners at companies that
that the participants were experienc- level (I called this the half-life) de- include Allied Signal, Bell Labs, Cum-
ing. I asked, “We have experienced pended on the complexity of the mins Engine, Cypress Semiconductor,
many serious recessions; could we process. I developed the matrix in Ex- GEC Ferranti, Hewlett-Packard and
make a Pareto diagram (a graphical hibit I, which summarizes these data. MEMC Electronic Materials, who have
form of rank ordering) of the root These data clearly showed that used the half-life method for goal set-
causes of these recessions?” The pro- as process complexity increased, the ting and improvement process diagno-
fessor’s terse answer ended the dis- rate of continuous improvement de- sis. My observations in this article are
cussion. “T.Q.M. doesn’t work on clined (the half-life became larger). a synthesis of our mutual experiences.
those kinds of problems,” he said. Furthermore, organizational com- Let me now make an assertion,
How can there be such differ- plexity was far more important than though it is a generalization and sub-
ences in the views of two T.Q.M. ex- technical complexity in slowing the ject to exceptions, that will form the
perts? Are there indeed limits to rate of improvement. In fact, it had basis for what follows: The complexity
T.Q.M. as practiced today? More im- nearly three times the effect that in- of processes most in need of improve-
portant, can we distinguish between creased technical complexity pro- ment in an organization increases with
situations where T.Q.M. is appropriate duced. Since the observed rate of im- the organizational level of the process
and those where it is not? The an- provement slows with increasing executors.
swers might help explain my experi- complexity, we are led to the obvious Exhibit II is a graphical represen-

...............................
1
Schneiderman, A., “Setting Quality Goals,” Quality Progress, April 1988, pp. 51-57.
2
Cooper, R., and Kaplan, R., “The Design of Cost Management Systems” (Englewood Cliffs, N.J.: Prentice Hall, 1991), pp. 226-239; Harvard
Business School Case 9-190-061; and Harvard Business School Teaching Note 05-191-103.
3
Dixon, J.,Nanni, A., and Vollmann,T., “The New Performance Challenge, Measuring Operations for World-Class Competition” (Home-
wood, Ill.: Dow Jones-Irwin, 1991), pp. 140-150.
4
Howell,R., Shank, J., Soucy, S., and Fischer, J., “Cost Management for Tomorrow, Seeking the Competitive Edge” (Morristown, N.J.: Finan-
cial Executives Research Foundation, 1992), pp. 127-149.
5
Lynch, R., and Cross,K., “Measure Up!, Yardsticks for Continuous Improvement” (Cambridge, Mass.: Basil Blackwell, Inc., 1991), p. 137.

36
Second Quarter ’98
 STRATEGY • MANAGEMENT • COMPETITION

Executive

tation of this assertion. The x- skipped, solutions often result

al Level
Upper Mgmt. s
axis is complexity, ranging from se in unintended consequences or
ces
uncomplex to very complex. I ro fail to have an impact on root

tion
Mid. Mgmt. l P
have avoided the word “simple” ca causes within the subprocess.

Organiza
r iti
since an uncomplex process may Supervisor ofC When there are interac-
not appear to be simple to its on tions outside of the team, they
gi
owners. The y-axis is organiza- Re tend to be limited in scope and
Worker
tional level. It ranges from nonex- un very highly visible. Usually, they in-
Complexity
empt workers to the entity leader Source: Arthur M. Schneiderman volve the previous process step
or C.E.O. The diagonal band rep- EXHIBIT II (supplier) and the next process
CONCEPTUAL FRAMEWORK
resents the region into which step (customer). With cellular
most processes tend to fall.
.................... manufacturing, both are in very
extent of which is far more wide- close physical proximity and real-time
PROBLEM SOLVING FOR spread than in the West. communication is easy. One system,
UNCOMPLEX PROCESSES A Q.C. circle consists of a first line kanban, that is used extensively in
The construct provides a useful supervisor and all of those who report Japan assures speedy communica-
road map for the historic evolution directly to him or her. They constitute tions of problems both upstream and
of T.Q.M. and a possible indicator a team that has complete ownership downstream of the subprocess, before
of its future direction. T.Q.M. had its of a subprocess. In general, the im- a large inventory of defective output
origins in Japan in the 1950’s with provement activities of Q.C. circles are accumulates and while the trail is still
the landmark visits of W. Edwards limited to the team’s local work area. warm enough to identify a root cause.
Deming and J.M. Juran. It was also The subprocess is usually indepen- By the nature of these subprocesses,
during that decade that the Quality dent of external processes. they are uncomplex. The level of com-
Control circle had its roots. Under Each Q.C. circle then follows a plexity, as we will see, is central to the
the leadership of the famed Kaoru seven-step process (see article, page current limits to T.Q.M.
Ishikawa, Q.C. circle activities have 38). This process — collecting data, In terms of the complexity map, it
grown to become the foundation of discovering root causes, implement- seems fair to say that the first 20 years
the Japanese T.Q.M. movement, the ing corrective action — has been uti- of T.Q.M. activity in Japan were focused
................... lized successfully to im- in the lower left hand corner. The teams

EXHIBIT III prove subprocesses in were Q.C. circles, the method was sev-
QUALITY CONTROL CIRCLES manufacturing and other en steps and the tools were the seven
operations for more than 40 Q.C. tools. The problems that they
Executive
years. For successful Q.C. solved were additive, noninteractive
Upper Mgmt.
circles, adherence to the and static in nature. I call them Type I
vel

process — not just finding a problems. (See Exhibit III.)

tional Le

Mid. Mgmt. eps solution — is seen as criti-

St MOVING UP THE DIAGONAL:
7 cal. Data collection and root
Organiza

Supervisor cause analysis break ADDING COMPLEXITY TO

Q.C. Circles
processes down into com- PROCESSES
7 Q.C. Tools
Worker ponents and causal rela- The clear success of the Q.C. circle
un very
Complexity tionships that describe the movement in Japan led to the natural
type I
subprocess. desire to move improvement activity
Source: Arthur M. Schneiderman When these steps are to the ranks of management. But Q.C.
continued on page 39

37
Issue 11
 STRATEGY • MANAGEMENT • COMPETITION

THE SEVEN-STEP QUALITY PROCESS AND THE SEVEN Q.C. TOOLS

The seven-step method followed of the problem. Steps 2 and 3 are do it. Often, proposed solutions
by the Q.C. circle is a special case there to improve the odds of com- are implemented without a test
of the P.D.C.A. Cycle (see Exhibit). ing up with the right solution. of their effectiveness. Step 5 re-
The rigorous adherence to this It is not my job to implement. quires the completion of a pilot or
problem-solving process avoids Have you ever received one of test of the proposed solution. If
several common difficulties en- those memos that say “I’ve stud- improvement results, then the so-
countered by teams. ied your problem and here is what lution is implemented on a full-
Solving the wrong problem. you should do”? Steps 4 and 5 are scale basis. One of the major parts
Step 1 and the associated prework there to assure that the objective of process redesign and re-engi-
assures that the team has an im- is not a solution. The objective is neering is the elimination of steps
portant and clearly defined prob- an implemented improvement. All that do not add value. Often,
lem and that the other six steps
can be completed by the group. I 1 Identification of Theme PLAN
have found that in well over half What
of the cases in which teams fail to
Data Collection
solve a problem, it is because 2 and Analysis
they chose the wrong problem to
solve or because there was a lack
of agreement among the team 3 Causal Analysis Why
members on what the problem
was. For this reason, when I am Who
Solution Planning
doing seven-step training, I
4 and Implementation
When
Where
DO
spend one-third of the time on How
step 1.
Jumping from problem state-
5 Evaluation of Results CHECK
ment to solution. If you ask suc-
cessful people for the keys to
6 Standardization ACTION
their success, they almost always
have problem-solving skills near
the top of the list. But it is re- Reflection and
7 Next Problem
markable how often even good
problem solvers are wrong. That Source: Arthur M. Schneiderman

is because we have a natural ten- THE P.D.C.A. CYCLE AND THE 7 STEPS
dency to jump from the problem
............................
statement (step 1) to a solution members of the team own the en- these steps were created to im-
(step 4) without bothering to col- tire improvement process. prove the process, but without
lect data and analyze the causes I know this will work; let’s just the completion of step 5, they had
continued on page 40

38
Second Quarter ’98
 STRATEGY • MANAGEMENT • COMPETITION

continued from page 37

circles were trained to avoid, if possi- helped identify the interactions be- come old. On the contrary, the 7 Q.C.
ble, problems that involved human tween the subprocesses that make up tools are still the most important set of
behavior, particularly when the be- the more complex processes encoun- techniques in Japan. Rather, the 7M
havior was that of others outside of tered as we move up the diagonal. tools should perhaps be called ‘sup-
the team, and to focus instead on ma- The Japanese Union of Scientists plemental seven Q.C. tools,’ and in this
chines, materials and methods. The and Engineers (J.U.S.E.) embarked on sense, it is recommended that these
reason was that the seven Q.C. tools a process for developing a set of tools methods should be learned after the 7
did not prove useful in uncovering that would be more useful in solving Q.C. tools have been well mastered.”
root causes when the problems had a increasingly complex problems. The Real understanding and knowl-
significant human element (behavior) results of these efforts were first pub- edge of the seven management tools is
to them. lished in 1977. Because they proved to very limited outside of Japan. Efforts
Also, the fundamental assumption be very successful, they were pub- to teach them are often confounded by
in asking “why” five times is that a large lished in English in 1986.6 Kaoru their apparently overwhelming level of
problem is the sum of many indepen- Ishikawa’s foreword to the 1986 publi- detail. Quality Function Deployment
dent small problems. Many of us have cation describes the relationship of (Q.F.D.), for example, is based on the
seen the failure of this assumption the “old” and “new” tools: structured use of four of these tools.
when an apparently significant im- “This issue features the seven Many organizations that have tried the
provement in one subprocess pro- management tools for Q.C. The seven formal version of Q.F.D.7 have found
duced a more than offsetting degrada- management tools for Q.C. (7M tools) themselves bogged down in a level of
tion of another subprocess somewhere were proposed in 1977 by the Com- detail that tries the patience of most
else in the system. I would speculate mittee for Developing Q.C. Tools by se- Western managers.
that much of the hundreds of millions lecting various conventional tech- I have been a student of these
of dollars, even billions of dollars, of niques for creativity and management. tools for more than a decade. I have
cost reductions attributed to “cost of These 7M tools are used in Japan to- had the opportunity to apply some of
quality” improvements that mysteri- day to implement T.Q.M. them as a member of a 13-person team
ously fail to appear at the bottom line, “Now the seven quality control that spent full time over a five-week
has been absorbed elsewhere in the tools (7 Q.C. tools) — i.e., Pareto dia- period designing the Center for Qual-
system as a result in large part of the in- gram, cause and effect diagram, strat- ity Management, a network of organi-
terdependency of processes. ification, check sheet, histogram scat- zations implementing T.Q.M. In this ef-
In many systems, the whole is not ter diagram, graphs and control fort, we were guided by Prof. Shoji
the sum of its parts. The system takes charts — were proposed about 30 Shiba of J.U.S.E., the University of
on its nature by the interactions of its years ago and are widely used to solve Tsukuba and the Massachusetts Insti-
parts. This is particularly true of the more than 90 percent of the problems tute of Technology.8 I offer some ob-
problems encountered by manage- in Japanese enterprises. servations about these tools based on
ment. It was believed that the P.D.C.A. “Recently, in Japan the 7M tools my experience:
cycle and the seven-step problem- are sometimes quoted as the ‘new 7 ➢ Unlike the usual analytic tools
solving process were still valid. What Q.C. tools.’ But this is not to be misun- that deal with a series of numbers, the
was needed was a new set of tools that derstood that the 7 Q.C. tools have be- seven management tools deal with
............................... continued on page 41

6
Reports of Statistical Application Research, Japanese Union of Scientists and Engineers, Vol. 33, No. 2, June 1986.
7
See for example: Akoa, Y., editor, “Quality Function Deployment, Integrating Customer Requirements into Product Design” (Cambridge,
Mass.: Productivity Press, 1990).
8
Shiba, S.,Graham, A., and Walden, D.,“A New American T.Q.M.” (Cambridge, Mass.: Productivity Press, 1993).

39
Issue 11
 STRATEGY • MANAGEMENT • COMPETITION

continued from page 38

no effect on the problem they Methodology.” It taught us the tomers exceed their credit limits?
were intended to eliminate. evolution of the modern method Because 3: Because 80 percent
We’re too busy to waste time of inquiry that we would be using of the time when they placed the
on paperwork. Without step 6, so- as engineers: the continual jour- order, they didn’t know that the
lutions reside in volatile human ney, back and forth, between the order would put them over their
memory. Organizational learning level of thought (theory) and the credit limits.
is quickly followed by organiza- level of reality (observation). Tools Why 4: Why didn’t they know
tional forgetting. Processes drift like calculus or microscopes were that the order put them over their
back to their pre-improvement the means used in carrying out credit limits?
performance levels. the process. Because 4: Because 98 percent
Next problem, please. Step 7 Steps 2 and 3 — data collection of the time we didn’t know that
has a vital element to it that focus- and root cause analysis — usually the order would put them over
es on the improvement process it- utilize an activity called “ask why their credit limits, so we couldn’t
self. Reflection on the team’s im- five times.” In effect, the problem tell them.
provement experience often is broken down into smaller and Why 5: Why didn’t we know
uncovers correctable weaknesses smaller pieces. The tools used in that the order would put them
in skills, methods or team dynam- this process are called the seven over their credit limits?
ics. Step 7 helps teams become Q.C. tools. One, the Pareto analy- Because 5: Because the order-
more effective at problem solving, sis, is a bar chart that ranks the entry screen does not show current
thus reducing the half-life. The reasons for each level of “why.” available credit for each customer.
subprocess of step 7 can also be The largest bar is the one that the By the time a team answers the
1
described as deuterolearning: next “why” is targeted to. fifth “why,” both the root cause
learning how to improve the way As an example, consider the or- and corrective action become very
the team improves in order to der-fulfillment process. One of its de- clear. Fifth “why” solutions usually
accelerate its resulting rate of fects is late shipment to customers. involve straightforward reversal of
improvement. Why 1: Why were 30 percent the root cause. In this example,
It is important to distinguish of our shipments made late to our adding “available credit” to the
the difference between a prob- customers? order-entry screen was a simple
lem-solving process and problem- Because 1: Because 20 percent cure for this part of the problem.
solving tools. The process repre- of the late lines were held up by You may have calculated from
sents a paradigm for discovery. It the credit department. this example that the solution im-
is an idealized series of activities Why 2: Why were lines held proved on-time delivery by only 1
that should produce the desired up by the credit department? percent. To make a major dent on
outcome. When I was an engi- Because 2: Because for 25 per- these kinds of problems, you need
neering undergraduate in the cent of those held up, the cus- to have many teams working on
late 1950’s, a required course was tomers exceeded their credit limits. the many elements of the prob-
“Philosophy and Scientific Why 3: Why did these cus- lem. Hence, the use of the term
...............................
1
Argyris, C., and Schön, D., “Organizational Learning: A Theory of Action Perspective” (Reading, Mass.: Addison-Wesley Publishing
Company, 1978).

40
Second Quarter ’98
 STRATEGY • MANAGEMENT • COMPETITION

continued from page 39

language-based data. They are tools tarian and democratic: everyone has
“mass movement” to describe for fact processing, rather than nu- equal access to inputs and the “one per-
Q.C. circle activities. merical data processing. A typical son, one vote” rule is employed when-
9
This process, often illustrat- piece of language data is a statement ever votes are taken. The consensus na-
ed with successive Pareto dia- of a fact, for example: “25 percent of ture of the seven management tools
grams, used to be referred to as the policy committee meetings in 1995 attempts to subjugate individual in-
started more than 15 minutes late.” stinct to the view of the team. Many
“peeling the onion.” Others call
Because of their dependence on lan- managers have found that the consen-
it “drill down.” It is also related
guage, a shared understanding and sus view is in conflict with their own in-
to the solution to the proverbial
use of the science of semantics is a tuition, particularly when the team is
question “How do you eat an
prerequisite. made up of their subordinates and the
elephant?” The answer: “One ➢ The tools attempt to identify problem is complex. Yet one set of
bite at a time.” Because each the interrelationship among the vari- ground rules establishes as a require-
bite may represent a fraction of ous elements of a problem. They iden- ment “integrity — in that people com-
a percent of the total problem, tify causality and assess the relation- mit to use the outcome of a long
many teams must be gnawing ship between all of the “whats” and process rather than just go through the
away at all of the parts of the “hows” of a problem. In one case, I was exercise.” I know several senior execu-
problem. At Analog Devices, we part of a team that spent the better tives, who reached their position
mobilized more than 50 teams part of a day determining the strength through good judgment and instincts,
throughout the corporation to of the relationship for each element of who have told me that they will never
work on improving the order- a 4,000-cell matrix. subject themselves to the seven man-
➢ They focus on detailed plan- agement tools again. The destructive ef-
fulfillment process. In the peri-
ning as a precursor to flawless execu- fect of overruling a team that has spent
od 1986-1990, they succeeded
tion. It is said that the Japanese spend days reaching a consensus view was
in reducing late shipments from
five times as much time in planning as unacceptably disempowering to them.
more than 30 percent to less
we do, yet complete the whole ➢ They force a solution to the
than 3 percent.
process (plan + do) in less than half problem. Under the assumption that
Occasionally, by accident or the time.10 doing something is better than doing
intent, a genius comes along ➢ They make formal use of right- nothing, they will lead a team to the
and solves a problem at the brained (that is, creative or intuitive) best collective guess as to what to do.
first or second “why” level. processes as part of problem solving. ➢ They are incredibly labor in-
These solutions constitute the ➢ They force consensus through tensive.11 The use of the tools cannot
breakthroughs that most their structure. They are totally egali- be effectively delegated since they
Western companies have re- ...............................
lied upon as the basis of their
9
It is useful to distinguish three types of data. Type I data are numerical data that are continuous-
improvement activities. How- ly generated by a process. This type of information is used to control a process. Type II data are
ever, the mass movement na- numerical data collected specifically to test a hypothesis; for example, does machine A produce
more defects than machine B? Type III data are language data. This type consists of statements of
ture of Q.C. circles can em- fact. They are data in that they can be confirmed by others.
power the entire work force 10
King, B., “Better Designs in Half the Time” (Methuen, Mass.: QOAL/QPC, 1992).
toward problem resolution 11
There are an increasing number of software packages available to ease the “drafting” require-
and form a more powerful col- ments of these tools. However, the need for the team to look at every possible interaction remains.
Even in Japan, the need for T.Q.M. CAD tools is well recognized. Of the 260 employees of J.U.S.E.
lective genius. and its subsidiary organizations, more than half are working on T.Q.M. software development.

41
Issue 11
 STRATEGY • MANAGEMENT • COMPETITION

Executive

In terms of our construct (see Ex- ➢ A modification of the P.D.C.A.

Upper Mgmt. . hibit IV), we can describe the seven cycle called the “C.A.P.Do” (Check-
al Level

.A
D.C
P. management tools as a tool set for Act-Plan-Do) that reviews quarterly
s
ep
tion

Mid. Mgmt. cross-functional teams dealing with in- progress against goals and takes nec-
St Q.I.P. Teams
7
Organiza

7 Q.C. Tools creased problem complexity. Several essary corrective action to put the
Supervisor
Q.C. Circles
7 Mgmt. Tools kj
tree
matrix } Q.F.D.
V.O.C. of the tools (e.g., kj diagram, tree dia- project back on track.
7 Q.C. Tools gram, matrix data analysis) form the The kind of breakthrough proj-
Worker
basis of Q.F.D., the front end of which ects for which Hoshin Kanri is used
un very
Complexity is called the Voice of the Customer tend to be high in organizational com-
type I type II
(V.O.C.). Intermediate complexity prob- plexity, but low in technical complex-
Source: Arthur M. Schneiderman
lems have static interactions and very ity. The need is more for synchroniza-
EXHIBIT IV visible causal relationships. I refer to tion and resourcing than for technical
MOVING TO MIDDLE them as Type II processes. innovation. Also, the goals often call
MANAGEMENT for 10-times or 100-times annual im-
HOSHIN KANRI provement rather than 2-times (12
...............
One management area in which the month half-life) improvements that re-
rely heavily on the personal knowl- Japanese have made significant sult from the continuous incremental
edge of all of the team members. progress in recent years is in synchro- improvement activities described ear-
Therefore, they are very expensive to nizing or aligning corporatewide activ- lier. This accelerated improvement is
use. Proponents believe that the cost ity in the accomplishment of a few, more the result of increased resourc-
is justifiable in terms of the quality of clearly defined, organizationally com- ing of the P.D.C.A. process than extra-
the results compared with other plex breakthrough objectives. The tool ordinary breakthroughs. I understand
processes for making decisions. that they have developed for doing this that today any Deming Prize winner
There is widespread acceptance is called Hoshin Kanri (or policy de- must have a working Hoshin Kanri
and use of these tools in Japan. Judg- ployment, as it is known as in the West). system in place.
ing from the results, in terms of cycle Its most important elements are:
time reduction (particularly in new ➢ The deployment of both goals LIMITS TO T.Q.M.
product development), product inte- and the means for their achievement The Type II processes are moderately
gration and other factors, they appear throughout the organization. complex and have static interactions.
to be very effective, particularly when ➢ A process known as “catchball” They are the current frontier of T.Q.M.
applied to cross-functional processes. in which the required means are bal- and the seven management tools. Type
Experience in the West is more limit- anced against the available resources I processes that are independent and
ed. Companies like Milliken, Procter & and know-how. static are the proven world of Q.C. cir-
Gamble, Xerox and Ford speak very ➢ An integration of top-down cles. This leads us to the final version
enthusiastically about their success- (goals) and bottom-up (means) plan- of our complexity map, Exhibit V. The
es with these tools. Other companies ning processes. top circle represents the complex in-
have had disappointing results as ➢ The use of detailed implementa- teractive and dynamic processes that
their managers reject outright their tion plans that focus on the 5W’s + 1H: occupy much of the time of top man-
mechanical detail and rigidity. what, why, when, how, where and who.12 agement. These are the processes that

...............................
12
Some of the readers may remember Kipling’s 1902 story called “The Elephant Song.” In it appeared the words: “I keep six honest serving
men (they taught me all I knew); their names are what and why and when and how and where and who.” This is another Western inven-
tion borrowed by the Japanese.

42
Second Quarter ’98
 STRATEGY • MANAGEMENT • COMPETITION

are often counterintuitive in their be- Excessive national debt? Lack of edu- agement processes is both unknown
havior. Should they be the realm of in- cation, training and skills? Not only do and unknowable.
dividual intuitive managers who rely we lack a rank-ordered list of root caus- We need not go as far up the
on mental models? Or can teams of es, we lack a consensus on the causal complexity diagonal as our national
people address these issues? What linkages themselves, including their economic processes. Even within our
tools should they use? These Type III direction. And, we certainly cannot organizations, top management is
processes currently lie outside of the blame this on a shortage of data. struggling with critical processes in
boundary of T.Q.M. Does it even make sense to talk which the causal linkages are at best
I mentioned earlier that the sev- about the P.D.C.A. cycle for these obscure.
en management tools look at the in- types of problems? If we lack the tools One example is the new-product
teractions of the various elements of to identify causality, how can we tell if generation process. In many indus-
the total process. In using these tools, our corrective action worked or not? tries (automotive, semiconductor, in-
we ask such questions as: Maybe some other factor made things dustrial equipment, for instance) cy-
➢ Does A cause B to happen? look better or worse. Some might ar- cle time or time to market is of the
➢ Will the effect of A on B be very gue that if you get enough data and an-
strong, moderate, weak or nonexistent? alyze them, you can find the causal
➢ Which has more of an effect on relationships.
Executive
C — A or B? That argument fails in two ways. ?
➢ What is the relative importance First, the scientific methodology that . ?
Upper Mgmt. .A
.C
vel
of A, B and C on D? we have used for 500 years states that tional Le P.D Hoshin Kanri
(org. complexity)
➢ What could go wrong, and what before you can design an experiment, s
synchronization
resourcing
Mid. Mgmt. ep
is the best countermeasure? you need to have a hypothesis that St Q.I.P. Teams
7
Organiza

7 Q.C. Tools
The more complex a problem, the the experiment will either confirm or
more difficult it is to answer these ques- deny. Only then will you know which
Supervisor
Q.C. Circles
7 Mgmt. Tools kj
tree
matrix } Q.F.D.
V.O.C.

tions. Often, the answer depends on data to collect. The world of data is in- Worker
7 Q.C. Tools

other aspects of the process: A will finite. And, unlike quality, data are not un very
Complexity
cause B to happen if C is close to D, and free. It is easy to overlook the cost in type I type II type III
Source: Arthur M. Schneiderman
E happens before F. Furthermore, in dy- both time and money of collecting and
namic systems the answers to these analyzing data, particularly data
EXHIBIT V
questions are time dependent. No in- about complex systems. THE COMPLEXITY MAP
dividual or team can assess these in- Second, to understand dynamic
teractions in their heads or even on processes, you need several cycles
...............
dozens of sheets of paper. worth of data with the process held order of three to five years. Often, the
At the beginning of this paper, I relatively fixed. We hardly ever have success of the product in the market-
mentioned one such problem, period- that situation. I recently heard of a place cannot be assessed for another
ic recessions and the associated un- study that showed that the average couple of years. Let us apply the
employment. No one would deny that United States corporation reorganizes P.D.C.A. cycle to that process. We plan
excess unemployment is a defect as- every four years. It spends two years the product, we do it and then we are
sociated with our economic process. planning and two years implementing ready to check what we did by com-
What are the root causes of unemploy- each reorganization. Therefore, the paring actual results with the plan. But
ment? What, even, are causal factors? average United States company is re- more than five years have gone by.
Do high taxes cause unemployment? organizing all of the time. It may be What can we learn from the check?
Unfavorable balance of payments? that causality in many important man- Much of the disparity is caused by

43
Issue 11
 STRATEGY • MANAGEMENT • COMPETITION

wrong assumptions and forecasts views into the process to formalize many of our most advanced products.
made half a decade ago. Even if we milestones and to foresee future prob- Simulation modeling allows us to
learn what we did wrong back then, lems and their potential solutions. characterize the dynamic relation-
how applicable would it be to today’s ➢ Introduce parallelism into the ship between the various elements of
rapidly changing world? process wherever feasible. a process. It demonstrates that the
P.D.C.A. cannot be applied to Long-cycle-time processes inher- changes in a system are driven not on-
processes that have a cycle time that is ently rely upon good instincts for their ly by these relationships themselves,
long compared with the time in which design. They also require the use of but also by the state of the various
the environment (that is, the external process metrics rather than results parts of the system relative to their
14
processes to which it links) changes. In metrics. The number of new products desired state. It captures the inherent
fact, the way to improve the new prod- introduced, time to market and new time delay between cause and effect
uct generation process is to focus on product revenue are ineffective met- that masks the underlying leverage
the planning process itself, not on the rics for a long new-product generation points in the process. Most impor-
results of the process. Design a stan- process. Process metrics — like per- tant, it holds the promise of reducing
dardized process that is most likely to cent of scheduled milestones missed, the complexity of the seven manage-
produce good results. It will probably number of engineering change orders ment tools by pointing to the vital few
include subprocesses that: or number of schedule changes this interrelationships, often hidden, that
➢ Require proponents to be ex- month — are much more effective - really drive the behavior of complex
plicit about their assumptions. This drivers of improvement. Follow the systems. As a complement to the sev-
will guarantee that they have thought process and you will maximize the en management tools, it provides a
about such critical issues as cus- probability of product success. more palatable approach for senior
tomers, costs and competitors. managers.
Hewlett-Packard’s Break Even Time SIMULATION MODELING Simulation models provide a lab-
13
(B.E.T.) is a superb vehicle for assur- In the semiconductor industry, we oratory in which P.D.C.A. can be per-
ing that those closest to the process faced complexity issues in designing formed on long-cycle-time processes.
and the related subprocesses (manu- VLSI circuits. On a single tiny chip, we Since simulation models are often
facturing, marketing and sales) have can now produce the functionality based on managers’ perceptions of
thought through the issues and as- that required a room or even a build- the structure of a problem, they can
sessed the sensitivity of the resourc- ing worth of equipment only a few incorporate mental models with a
ing decision to the assumptions. The decades ago. The dense packing of the high right-brain or intuitive content.
B.E.T. system — that is, the process of various elements produces interac- It is time that we re-examine sim-
calculating a value using all sub- tions that could be identified and elim- ulation modeling as a tool to be added
processes — is powerful. However, inated only by costly trial and error. to the T.Q.M. tool set in order to move
the resulting B.E.T. as a performance Today our designers rely on simula- the boundary of its applicability up the
measure is of questionable value, be- tion modeling to reduce the time and complexity diagonal. I say re-examine
cause it may foster less risk-taking or cost to market. In fact, the need for because simulation modeling of com-
innovation. better simulation tools is a current plex, interactive, dynamic systems
➢ Introduce frequent design re- bottleneck in the development of was introduced by Jay Forrester of

...............................
13
House, C., and Price, R., “The Return Map: Tracking Product Teams,” Harvard Business Review, vol. 69, no. 1, January-February 1991, p. 92.
14
Schneiderman, A.,“Metrics for the Order Fulfillment Process, Parts 1 and 2,” Journal of Cost Management (Summer 1996): pp. 30-42,
(Fall 1996): pp. 6-17.

44
Second Quarter ’98
 STRATEGY • MANAGEMENT • COMPETITION

M.I.T. in the early 1950’s.15 Mr. Forrester new T.Q.M. tool set, the seven man- crease in complexity. We have had to
recognized the role of feedback loops agement tools. They appear effective devise project planning systems to co-
in determining the dynamic behavior for dealing with the Type II medium- ordinate the many tasks associated
of complex systems. His landmark re- complexity cross-functional problems with complex logistical efforts. Build-
search appeared in the Harvard Busi- that are omnipresent in the middle of ing a modern skyscraper or putting a
ness Review in the late 1950’s. our organizations. We need to under- man on the moon requires coordina-
A team can build a simulation stand their region of usefulness and as- tion of the activities of many people
model and identify the leverage points sure that they are used thoughtfully, and organizations. The division of la-
(root cause equivalents) for correc- rather than mechanically. They are an bor involved in these types of projects
tive action intervention. They can val- extremely costly set of tools (in terms is made possible by the ability to spec-
idate the model by comparing the be- of time). Furthermore, we must always ify fully the timing and required inter-
havior of the model with their real-life remember that consensus does not al- actions among the various players.
experiences with the real system. ways lead to the best answer. Detailed “specs” exist that fully char-
They can apply the P.D.C.A. cycle by Finally, we have the Type III prob- acterize the interfaces among the var-
observing the results of their correc- lems. These are highly complex, in- ious parties involved in these proj-
tive actions and changing the simula- teractive and dynamic and seem to ects. These specs do not yet exist for
tion model as appropriate. create the greatest amount of pain Type III management processes.
both inside and outside of our organi- Type III problems are outside of
CONCLUSIONS zations. The T.Q.M. mandate to “man- the current boundary of T.Q.M. I have
Yes, there are there limits to T.Q.M., for age by fact,” interpreted literally, gives shown how T.Q.M. practitioners have
now. T.Q.M. is a means for achieving short shrift to innovation and insight augmented their tool set to allow
improved organizational perfor- as a legitimate problem-solving them to deal with increasingly com-
mance. It is a product and it must meet method. People who repeatedly make plex problems. I have also proposed
the same customer quality require- good decisions in highly complex sit- simulation modeling as a means for
ments as any product in today’s uations based on their experience or further extending the boundary of ap-
world: gut feelings have significant contribu- plicability of T.Q.M., by developing
➢ It must meet specifications. tions to make. Even if they cannot ex- specs that characterize the complexi-
➢ It must be fit for the use to plain the workings of their mental ty of Type III processes. Should that
which it is put by the customer. models of a situation, it does not mean happen, we will again be faced with
➢ It must be low in cost, relative that those models are incorrect. Often the question: are there limits to
to its benefits. they are better than the descriptions T.Q.M.? I will then be hard pressed to
➢ It must change to meet the la- that we force to fit into our analytical define what is left of management out-
tent (unstated) needs of its cus- frameworks. Holistic problem-solving side of T.Q.M. Perhaps it is time for us
tomers. finds the right balance between data to face up to the reality that there is a
We need to determine the “specs” collection and analysis and the unde- fading distinction between Total Qual-
of T.Q.M., to help the customer to use scribed mental models of effective ity Management, total management
T.Q.M. in accordance with its specs problem solvers. and management itself (i.e., T.Q.M. =
and to improve the tool set to make Our ability to form mental mod- T.M. ≡ M.) SB
&

the tools more user friendly. els of complex situations, however,

We are beginning to understand a has not kept pace with the rate of in- Reprint No. 98204

...............................
15
See for example: “Collected Papers of Jay W. Forrester” (Cambridge, Mass.: Wright-Allen Press, 1975).

45
Issue 11