Beyond Toyota: How to Root Out Waste and Pursue Perfection - Harvard Business Review
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Beyond Toyota: How to Root Out Waste and Pursue Perfection by James P. Womack and Daniel T. Jones Harvard Business Review Reprint 96511
Harvard Business Review
HBR Subscriptions Harvard Business Review
U.S. and Canada Outside U.S. and Canada
Subscription Service Tower House
P.O. Box 52623 Sovereign Park
Boulder, CO 80322-2623 Lathkill Street
Market Harborough
Leicestershire LE16 9EF
Telephone: (800) 274-3214 Telephone: 44-85-846-8888
Fax: (617) 496-8145 Fax: 44-85-843-4958
American Express, MasterCard, VISA accepted. Billing available.
HBR Article Reprints Harvard Business School Publishing
HBR Index and Other Catalogs Customer Service - Box 230-5
HBS Cases 60 Harvard Way
HBS Press Books Boston, MA 02163
Telephone: U.S. and Canada (800) 545-7685
Outside U.S. and Canada: (617) 495-6117 or 495-6192
Fax: (617) 495-6985
Internet address: custserv@cchbspub.harvard.edu
HBS Management Productions Harvard Business School Management Productions videos are
Videos produced by award winning documentary filmmakers. You’ll find
them lively, engaging, and informative.
HBR Custom Reprints Please inquire about HBR’s custom service and quantity discounts.
We will print your company’s logo on the cover of reprints, collec-
tions, or books in black and white, two color, or four color. The pro-
cess is easy, cost effective, and quick.
Telephone: (617) 495-6198 or Fax: (617) 496-8866
Permissions For permission to quote or reprint on a one-time basis:
Telephone: (800) 545-7685 or Fax: (617) 495-6985
For permission to re-publish please write or call:
Permissions Editor
Harvard Business School Publishing
Box 230-5
60 Harvard Way
Boston, MA 02163
Telephone: (617) 495-6849HarvardBusinessReview
SEPTEMBER-OCTOBER 1996
Reprint Number
JAMES C. COLLINS AND JERRY I. PORRAS BUILDING YOUR COMPANY’S VISION 96501
DAVID A. THOMAS MAKING DIFFERENCES MATTER: A NEW PARADIGM 96510
AND ROBIN J. ELY FOR MANAGING DIVERSITY
ANN MAJCHRZAK BREAKING THE FUNCTIONAL MIND-SET IN 96505
AND QIANWEI WANG PROCESS ORGANIZATIONS
N. CRAIG SMITH, ROBERT J. THOMAS, A STRATEGIC APPROACH TO MANAGING PRODUCT RECALLS 96506
AND JOHN A. QUELCH
RICHARD B. FREEMAN TOWARD AN APARTHEID ECONOMY? 96503
WITH COMMENTARIES BY: ROBERT B. REICH, JOSH S. WESTON,
JOHN SWEENEY, WILLIAM J. MCDONOUGH, AND JOHN MUELLER
GEORGE STALK, JR., DAVID K. PECAUT, BREAKING COMPROMISES, BREAKAWAY GROWTH 96507
AND BENJAMIN BURNETT
JOHN STRAHINICH HBR CASE STUDY
THE PITFALLS OF PARENTING MATURE COMPANIES 96508
BARBARA E. TAYLOR, RICHARD P. CHAIT, SOCIAL ENTERPRISE
AND THOMAS P. HOLLAND THE NEW WORK OF THE NONPROFIT BOARD 96509
THOMAS DONALDSON WORLD VIEW
VALUES IN TENSION: ETHICS AWAY FROM HOME 96502
JAMES P. WOMACK IDEAS AT WORK
AND DANIEL T. JONES BEYOND TOYOTA: HOW TO ROOT OUT WASTE AND 96511
PURSUE PERFECTION
MARC LEVINSON BOOKS IN REVIEW
CAPITALISM WITH A SAFETY NET? 96504I D E A S A T W O R K
Lantech achieved tomers showed no evidence of lean-
ness when adjusted for the ups and
unimaginable results by downs in the business cycle.
We concluded that the problems
applying lean thinking to were twofold. Although many man-
every aspect of its business. agers had grasped the power of indi-
vidual lean techniques – quality
function deployment for product de-
velopment, simple pull systems to
Beyond Toyota: replace complex computer systems
for scheduling, and the creation of
work cells for operations ranging
How to Root Out Waste from credit checking and order entry
in the office to parts fabrication in
the plant – they had stumbled when
and Pursue Perfection it came to putting them all together
into a coherent business system.
That is, they could hit individual
notes (and loved how they sounded)
but still couldn’t play a tune. And
even those managers who could car-
ry a tune found it very hard to intro-
duce comprehensive change in those
mature organizations that make up
the great bulk of every national
economy at any point in time.
We therefore set out in 1992 to
by James P. Womack and Daniel T. Jones identify and articulate a comprehen-
sive lean business logic, which we
now call lean thinking. We studied
Six years ago, we wrote, with that they were adopting lean tech- 50 companies throughout the world
Daniel Roos, The Machine That niques – techniques for relentlessly in a wide variety of industries – from
Changed the World. The book sum- and continuously eliminating waste the company that had pioneered
marized the Massachusetts Institute from an operation. And in that the approach, Toyota, to such re-
of Technology’s study of the global heartland of global manufacturing, cent initiates as Japan’s Showa Manu-
automobile industry, which docu- the automobile industry, it was soon facturing, Germany’s Porsche, and
mented the great performance ad- impossible to find a manager any- U.S. companies ranging from giant
vantages that a best-in-class lean where who did not profess to be Pratt & Whitney to relatively small
manufacturer such as Toyota had “getting lean.” Lantech, a manufacturer of wrap-
over typical mass producers in West- Those claims were mostly wishful ping machines. We believe that enu-
ern countries. When we presented thinking. When we looked more merating the five steps those lean
our evidence, we feared the indus- closely, we found plenty of just-in- companies have taken will be useful
trial equivalent of an immune reac- time delivery systems that involved to managers everywhere.
tion, in which managers in other re- nothing more than the relocation of
gions and industries would reject inventories from the company we James P. Womack advises compa-
were visiting to the next nies on how to apply lean thinking
company upstream. In of- to their operations and maintains a
Managers are struggling to fices and plants, we found research affiliation with the Japan
unlinked islands of lean Program at the Massachusetts Insti-
combine lean techniques operating techniques. tute of Technology in Cambridge,
into a coherent system. And we found many al-
legedly lean product-
Massachusetts. Daniel T. Jones is a
professor of management at the
development groups that Cardiff Business School of the Uni-
lean techniques as irrelevant to their were nothing more than compart- versity of Wales and the director of
circumstances or impossible to im- mentalized organizations with new its Lean Enterprise Research Centre.
plement. Instead, we discovered that labels. One statistic in particular ex- This article is adapted from their
we were battering down an open posed the truth: the inventories that book Lean Thinking: Banish Waste
door. We encountered scores of man- the North American, European, and and Create Wealth in Your Corpora-
agers in industries as diverse as aero- Japanese economies need to support tion, to be published by Simon &
space and construction who told us a given level of sales to end cus- Schuster in September 1996.
Copyright © 1996 by James P. Womack and Daniel T. Jones. All rights reserved. HARVARD BUSINESS REVIEW September-October 1996I D E A S A T W O R K
1. Define value precisely from the stream almost always exposes enor- 4. Design and provide what the
perspective of the end customer in mous – indeed, staggering – amounts customer wants only when the cus-
terms of a specific product with spe- of muda in the form of unnecessary tomer wants it. Letting the end
cific capabilities offered at a specific steps, backtracking, and scrap as the customer pull the product from the
price and time. As the late Taiichi product proceeds from department value stream in this fashion elimi-
to department and from nates muda in the form of designs
company to company. that are obsolete before the product
All industrial thinking must The organizational can be introduced, finished-goods
mechanism for defining inventories, elaborate inventory-
begin by differentiating value and identifying the tracking systems, and remaindered
value for the customer value stream from con-
cept to launch, order to
goods no one wants.
5. Pursue perfection. As lean tech-
from muda. delivery, and raw materi- niques begin to be applied up and
al to finished product is down the value stream, something
the lean enterprise – a very odd starts to happen. It dawns
Ohno, one of the creators of the leg- continuing conference of all con- on those involved that there is no
endary Toyota Production System, cerned parties to create a channel for end to the process of reducing effort,
put it, all industrial thinking must the stream, dredging up all the time, space, cost, and mistakes
begin by differentiating value for the muda. For a full explanation of this while offering a product that is ever
customer from muda – the Japanese concept, see our article “From Lean more nearly what the customer ac-
term for waste. Production to the Lean Enterprise” tually wants.
While seemingly straightforward, (HBR March-April 1994). Why should that be? Because the
this step is actually hard to carry out 3. Make the remaining value- four initial steps interact with one
and for a very simple reason: for any creating steps flow. Making steps another in a virtuous circle. A more
product more complex than a tooth- flow means working on each design, precise definition of value always
pick and for any service more com- order, and product continuously challenges the steps in the value
plicated than a haircut, value must from beginning to end so that there stream to reveal waste, and getting
flow across many companies and is no waiting, downtime, or scrap value to flow faster always exposes
through many departments within within or between steps. This usu- hidden muda. Then, the harder cus-
each company. Although each entity ally requires introducing new types tomers pull, the more the impedi-
along the route may or may not de- of organizations or technologies ments to flow are revealed, permit-
fine value for the end customer, it and getting rid of “monuments” – ting them to be removed.
certainly will define value for itself – machines whose large scale or com-
to turn a profit, to advance the ca- plex technology necessitates operat- The Lean Revolution
reers of those in each department, to ing in a batch mode. at Lantech
utilize existing production assets Many Western managers mistak- Applying these five concepts re-
fully, and so forth. When all those enly believe that flow is something quires a complete organizational
definitions of value are added up, one can achieve only gradually transformation, and it’s difficult for
they often conflict with or cancel through kaizen, or continuous incre- the uninitiated to know where to
out one another. Consequently, fail- mental improvement. However, by start. Lantech of Louisville, Ken-
ure to specify value correctly before first practicing kaikaku, or radical tucky, provides an excellent exam-
applying lean techniques can easily improvement, lean thinkers at com- ple of how to make the leap in an ex-
result in providing the wrong prod- panies we have studied were often isting operation.
uct or service in a highly efficient able to transform in a single day Lantech’s founder, Pat Lancaster,
way – pure muda. the production activities required to is a heroic American type. He grew
2. Identify the entire value stream make one product from a batch-and- up tinkering in the family work-
for each product or product fam- queue system to a continuous flow. shop, convinced from an early age
ily and eliminate waste. The value As a result, they doubled productiv- that he could be an inventor. In
stream is all the specific actions re- ity and dramatically reduced errors 1972, when Lancaster was 29, he had
quired to bring a specific product and scrap. A similar rearrangement his big idea: a new way for manufac-
through three critical activities of of product-development and order- turers to wrap their products for
any business: product definition scheduling activities produced gains shipment. He and his brother invest-
(from concept through detailed de- of comparable magnitude. When pro- ed $300 in simple metalworking
sign and engineering to production cesses truly flow, products that re- tools to build their first wrapping
launch), information management quired years to design take months, machine, rented a small warehouse,
(from order taking through detailed orders that required days to process and went to work under the corpo-
scheduling to delivery), and physical are completed in hours, and the rate name of Lantech.
transformation (from raw materials throughput time for physical pro- Lancaster’s idea was a device that
to a finished product in the hands of duction shrinks from months or would stretch-wrap pallets of goods
the customer). Identifying the value weeks to days or minutes. with plastic film so that they could
HARVARD BUSINESS REVIEW September-October 1996 5I D E A S A T W O R K
How Lantech Makes Its Stretch Wrappers
Old Batch-and-Queue System New Continuous-Flow System
Storage Sawing Machining Welding Subassembly Incoming
of raw materials Cell for Q model
materials
Storage Storage of parts in Storage of
Frame painting
of process incoming
finished components
goods
Final assembly
Crating Storage of Touch-up Ship
painted finished
frames goods
Frame painting
be shipped easily from plant to plant years. Soaring energy prices created corrosion-inhibiting base coat and a
within a manufacturing system and an overwhelming advantage for cosmetic finish coat to the complet-
then onward, as finished products, stretch-wrapping. By 1979, Lantech ed frame. The subassembly depart-
to wholesalers and retailers. In con- had sales of $13.4 million and em- ment built component systems from
trast to traditional shrink-wrapping, ployed 158 people. parts purchased from suppliers. And
in which plastic bags are placed Lancaster had created his initial the final-assembly department at-
loosely around palletloads of goods design and his first machine in a tached the component systems to
and then shrunk in an oven until continuous flow of activities. So the frame.
they fit tightly, stretch-wrapping Lantech was born lean, like most In pursuit of efficiency, Lantech
would pull plastic wrap tightly start-up businesses. However, when built its four basic types of machines
around the palletload as it rotated on he began to make his product in vol- in batches; it fabricated and assem-
a turntable, eliminating the energy, ume, in the late 1970s, it didn’t seem bled 10 to 15 machines of a type at
equipment, effort, and time required practical to run an established busi- one go. However, because customers
for heat-treating. ness that way. Lancaster hired an ex- usually bought only one machine at
Lancaster soon discovered that a perienced operations manager to run a time, the company had to store
complex set of precision rollers his new plant, an engineering direc- most of the machines in each batch
could exert a smooth force on the tor to create a variety of configura- in a finished-goods area until they
plastic to stretch it before it was tions of the basic concept, and a were purchased. A stretch wrapper
sales director to manage thus had to take quite a circuitous
a sales force of inde- route during its creation. (See the
In pursuit of efficiency, pendent distributors. It exhibit “How Lantech Makes Its
seemed natural for the Stretch Wrappers.”)
Lantech built its four operations, sales, and en- Complexity increased exponen-
basic types of stretch gineering managers to or-
ganize Lantech into a se-
tially as Lantech tried to move the
orders gathered by the independent
wrappers in batches. ries of departments, each sales force through the office and
with a specialized task. the plant. Because the $10,000-to-
In the plant, the sawing $150,000 machines were usually
wound around the pallet. Eventu- department used metal saws to fash- customized, the sales force had to
ally, his system could wrap with ion frame members from steel contact Lantech for authorization
a given amount of plastic an area beams. The machining department before quoting a price.
7.5 times the size that a shrink sys- drilled and punched holes in the Proposals were sent for cost analy-
tem could wrap. steel to create points for attaching sis to the engineering applications
When Lancaster obtained patents component systems. The welding department, which then sent the ac-
for his concepts in the early 1970s, department welded together the ceptable price back to the sales force.
they were so general and broad that parts for the machine’s frame. The Once the customer accepted the
he could fend off competitors for painting department applied both a price, the order traveled from the
6 HARVARD BUSINESS REVIEW September-October 1996I D E A S A T W O R K
morning, work- stretch wrapper or delayed the deliv-
ers in every de- ery date and built a properly config-
partment – saw- ured machine from scratch.
One Cell’s Production Flow: The Q Model ing, machining, Soon the master schedule devel-
welding, paint- oped in the scheduling department
painting
Frame
ing, subassem- and the ever changing demands from
bly, final assem- the sales group were pulling the
Sawing Machining Welding
bly, touch-up, and plant in opposite directions. Expe-
crating – picked diters from the order management
up a printout department moved through the
with their tasks plant with a “hot list.” They visited
Final-assembly track
for the day. At departments in sequence and or-
the end of the dered the workforce to make just
day, every depart- one item of a batch so that they
Testing Subassembly Subassembly ment reported could take that part immediately to
and shipping of roll of control its progress to the next department and move it to
carriage module
the scheduling the head of its queue. In an extreme
department. situation, it was possible to get a
This system stretch wrapper built in less than
was fine in the- four weeks. However, doing so
ory but always a caused the schedule for other ma-
mess in practice chines to slip and created the neces-
because of the sity for more expediting.
sales staff through engineering appli- conflict between customers’ chang- The system sounds chaotic, and it
cations, design, and credit checking ing desires and the logic driving the was. But in most of the industrial
before returning to design, which production system. Lancaster and world, such an approach was and
generated a bill of materials – a list of his operations manager directed is the standard method for mak-
every part needed to manufacture each department to do its work in ing products when there are many
that specific machine. The order batches. They wanted to minimize possible versions, when the produc-
with the bill of materials then went the time Lantech’s
to the production operation’s sched- machinery was idle
uling department, where a comput- during the changeover Salespeople tried to beat
erized material-requirements-plan- to making a new part,
ning (MRP) system assigned it a as well as to mini- the system to get machines
place in the master schedule. Be-
cause every department had a queue
mize opportunities to
misset machines. But
for customers faster.
of orders, there usually were delays. their approach inevi-
As a result, orders generally took 12 tably produced a typical batch-and- tion process is complex, and when
to 14 workdays to travel from the queue environment, in which each throughput times are long.
sales staff to the scheduling depart- part waited its turn at the entrance Lantech’s departmentalized engi-
ment, even though the actual pro- to each department and then re- neering process for developing new
cessing time was less than 2. tur ned to a central parts ware- models employed a similar batch-
Because the movement of prod- house to await its next processing and-queue approach. To create a new
ucts through the plant was so errat- step. Incoming steel usually spent design, it was necessary to have the
ic, the company created a separate 16 weeks at Lantech before reaching marketing staff, engineers from sev-
order-management department the shipping dock as a completed eral specialties, the purchasing staff,
within the sales group to enable the machine, even though the total time and operations planners working to-
independent sales force to commu- required to perform all the fabrica- gether. The marketing group ascer-
nicate with the plant about where tion and assembly steps was just tained what the customer wanted,
the machine was in the production three days. and the chief engineer translated
process and to expedite the order if Confronted with long throughput those desires into engineering speci-
the customer was getting restless. times, the sales force frequently fications. One mechanical engineer
(See the exhibit “How Lantech tried to beat the system in order to then designed the moving mechani-
Processes Orders.”) secure machines for customers cal parts, and another designed the
The MRP system melded a long- faster. A favorite approach was to frame. An electrical engineer de-
term forecast for orders with ac- order machines on speculation and signed the control system, and a
tual orders as they were received to then, when a real customer was manufacturing engineer designed
create a daily production sched- found, to alter the options very the fabrication tools. Once the de-
ule, which assigned tasks to each late in the production process. The signs of the product and the tools
department in the plant. Each factory then either reworked the were finalized, an industrial engi-
HARVARD BUSINESS REVIEW September-October 1996 7I D E A S A T W O R K
neer from the production depart- or her desk. To get rush projects fered in only very minor ways from
ment designed the route the model through the system, Lantech again previous models. And we made tons
would have to take through the had to turn to expediters. In practice, of money.”
plant to be built, and the list of nec- it usually took a year to introduce a Then, on June 26, 1989, Lantech
essary parts was placed in the sched- minor improvement and three or lost a patent-infringement suit
uling computer. Meanwhile, the four years to introduce a new family against a competitor that was offer-
purchasing staff lined up suppliers of machines, although the time it ing lower-priced clones of Lantech
would have taken if proj- machines. The verdict threw open
ects spent no time in the market. By the end of 1989,
“The bottom fell out of my queues or backtracking clones with roughly comparable per-
was only a few weeks for formance started to appear every-
pricing, and worse was minor improvements and where. “The bottom fell out of my
coming. I knew Lantech six months for a new
family of machines.
pricing, and I knew worse was com-
ing as soon as the business cycle
was walking dead.” In summary, Lantech turned down,” Lancaster says. “In
conducted its three major my heart, I knew that Lantech was
activities – creating new walking dead.”
for parts not made at Lantech. (See designs, managing information on No quitter, Lancaster tried most
the exhibit “How Lantech Develops what to make, and physically pro- of the remedies popular in the U.S.
New Products.”) ducing its machines – in a classic business community at the time.
In its infancy, Lantech had only a batch-and-queue manner. Many His first approach was to reorganize
half dozen engineers, but even then steps added no value, nothing flowed, the company into separate profit
communication barriers were sub- customers couldn’t pull, and manag- centers for “standard products” and
stantial as a design moved from the ers focused on minimizing variations “specials” (those requiring exten-
marketing group to the chief engi- in operations rather than on pursu- sive customization) in order to in-
neer to the mechanical engineers to ing perfection. crease accountability and to move
the electrical engineer to the manu- Until 1989, Lantech was able to the highly customized products out
facturing engineer to the industrial tolerate those deficiencies. “We of the path of machines sold in high-
engineer. Getting from the initial were selling a top-priced product er volumes. After a visit to Milliken,
concept to a complete production- that had major performance advan- the South Carolina textile giant, he
ready design required rework and tages over competitors’ products be- also introduced total quality man-
backtracking, and as the company cause of my patent position,” Lan- agement in order to put the voice of
grew and more engineers were caster recalls. “We offered so-so the customer first and foremost.
added, the communication prob- quality in terms of manufacturing Lantech’s “good enough” standard
lems worsened. defects in machines delivered to cus- for the acceptable level of delivered
What’s more, each engineer typi- tomers. We took more than a year to defects and customer service was
cally had a stack of projects on his develop ‘new’ machines, which dif- replaced with talk about perfection.
How Lantech Processes Orders
Old Batch-and-Queue System New Continuous-Flow System
Sales staff Regional Purchasing
sales
coordinators Sales Order entry/ Welding
Scheduling Engineering
Credit by product applications
Quoting checking by product
Order Production
entry/Scheduling work
orders Purchasing
Engineering
(MRP master by product
applications Quick
schedule) response
team for
Design and price
BOM* quotations Manufacturing
Production expediters
Credit
checking
* bill of materials
8 HARVARD BUSINESS REVIEW September-October 1996I D E A S A T W O R K
How Lantech Develops New Products
Old Batch-and-Queue System New Continuous-Flow System
Product Dedicated product teams
definition work together in one location
Engineering Marketing Industrial Purchasing
specifications engineering
Mechanical
engineering Team Design in concurrent
leader development Launch
Electrical
engineering
Manufacturing Mechanical Electrical Manufacturing
engineering engineering engineering engineering
Industrial
engineering
Over the next few years, Lantech The third approach to the crisis scheduling system that gave every
initiated a second approach: an ef- was a new production method called worker direct access to the status of
fort to create an empowered orga- Max-Flex. The idea was to slash lead every machine in production. “It
nization and to build trust between times by building inventories of ma- seemed to be a wonderful marriage
management and the workforce and jor components far in advance, then of technology and democracy,” Lan-
among the different departments. assembling machines to customers’ caster recalls. “Everyone could look
Lancaster replaced those senior specifications very quickly when an into the computer to see what was
order was confirmed. The going on all over the plant and get
objective was to over- their work orders immediately. Our
“We wanted to work come Lantech’s pricing slogan was Data to the People.”
disadvantage by promis- The new system required a new
together in teams, but we ing more rapid delivery of computer, a new management-infor-
were all revved up with machines with customer-
specified features.
mation-system department, and
workers on the floor to update the
nowhere to go.” Lead times fell from 16 computer when they completed each
weeks to 4. But the costs task. As José Zabaneh, the manu-
were enor mous. Engi- facturing director, notes, “Pretty
managers who had a command- neering changes were frequent. As soon, workers were fully ‘in control,’
and-control style with managers a result, it often was necessary to yet the system was wildly inaccurate
willing to work in a team-based or- retrofit the mountain of components because many items simply never
ganization, and the company con- that had been built in advance. In ad- got entered. The old MRP system
ducted extensive training in team dition, the cost of carrying that was slow but 99% accurate. Our
processes, team leadership, and indi- mountain was substantial. But, new ‘democratic’ system was a com-
vidual interaction. most exasperating, despite Lantech’s plete catastrophe; instead of infor-
Those programs were an essential best efforts at planning production, mation, we had given muda to the
start, but they lacked a direct con- cases quickly arose in which one people.” Making matters worse, the
nection to the company’s core activ- critical component needed to com- magnitude of inputs and changes
ities. As Bob Underwood, a longtime plete a machine was lacking. (Tai- made the computer run very slowly.
production worker, puts it, “We ichi Ohno noted long ago that the A consultant recommended buying
learned to respect one another and more inventory you have, the less a much more powerful computer.
wanted to work together in teams, likely you are to have the one part By the end of 1991, Lantech’s or-
but we were all revved up with no- you actually need.) The solution was ders were falling despite price reduc-
where to go.” The factory was still a new team of expediters to get the tions, and the factory was unable to
a mess. Product development was missing components built. accommodate the continual shifts in
still too slow. And the sales force Yet a fourth approach to the crisis demand. “We began losing money,
was still playing games to beat the was better information technology. and our fundamental ideas on how
lead-time problem. In 1991, Lantech installed a new to run the business were in a melt-
HARVARD BUSINESS REVIEW September-October 1996 9I D E A S A T W O R K down,” Lancaster says. Then he dis- and waste – muda of all sorts – would thought the value stream and flow. covered lean thinking by accident: be banished. The value stream would In less than a week, all the equip- when he advertised an opening for flow smoothly, continuously, and rap- ment was moved into a new configu- the position of vice president for op- idly. Hicks got the job. ration. Only the painting booth, a erations, one person responded with classic monument, survived as a de- some highly unusual ideas. That Eliminating Wasteful partment. But once parts had gone man was Ron Hicks. Activities and Creating Flow through the painting booth, they re- Although Ron Hicks does not look As it happened, the lean trans- turned to the individual cells for like a revolutionary, he started a rev- formation at Lantech was easy in subassembly, final assembly, test- olution when he went to work for one important respect: customers ing, and crating. (See, in the exhibit Lantech in March 1992. He had were satisfied with the company’s “How Lantech Makes Its Stretch learned how to be a revolutionary stretch-wrapping equipment. Be- Wrappers,” the chart “One Cell’s while serving as operations vice cause its value to them was not in Production Flow: The Q Model.”) president of Hennessy Industries, a question and because Lantech un- Each morning, every hour, the saw manufacturer of automotive repair derstood that value, the company operator would start production of a tools in Tennessee that had become could safely skip the first step in ap- new machine. A kit of all the frame a lean organization. plying lean thinking. parts required for the machine was To transform itself into a lean or- Upon joining Lantech, Hicks im- ready by the end of the hour and was ganization, a company needs three mediately went to work with a sim- rolled approximately three feet to types of leaders: someone who is ple plan to untangle the flow of the machining station. From the committed to the business for the value by establishing a dedicated machining station, it proceeded an long term and can be the anchor that production process for each of the hour later about four feet to weld- provides stability and continuity; four product families. His plan ing. Fourteen hours after the start of someone with deep knowledge of called for disbanding the production parts fabrication, the completed ma- lean techniques; and someone who departments and regrouping the ma- chine was ready to ship. can smash the organizational barri- chinery so that all the equipment To make this simple system suc- ers that inevitably arise when dra- needed to make each of the four ceed, Lantech needed to change a matic change is proposed. Lancaster models was located together in four generation of thinking about work filled the first role, Hicks the second, separate production cells. Lantech and how people work together. First, and Zabaneh the third. also would have to “right-size” because all the jobs were directly In the newly empowered spirit many of its tools–get rid of the huge, linked, with no buffers of invento- of Lantech, Hicks was invited to overly complex machines (or monu- ried parts, it was essential that all Louisville and interviewed by the ments) and install smaller saws and employees think about standardiz- people he would manage. His simple machining tools – so that they could ing their work so that a given task proposal to them came as a revela- fit in the work cells. This step was would take the same amount of time tion: Lantech would immediately the kaikaku phase – the time to tear every time and also would be done form teams to rethink the value things apart and recombine them in correctly on the first attempt. By de- stream and the flow of value for a totally different way. Not only sign, either the whole cell was work- every product in the plant and for would products flow continuously ing smoothly at the same pace or ev- every step in order taking and prod- from start to finish, but the waste- erything came to a halt. For that uct development. Lantech would fulness of moving parts back and reason, every task needed to be care- identify all the activities required at forth from central storage to each fully described in a posted diagram the time to design, order, and manu- department, the long waits, and the so that everyone in the production facture a stretch-wrapping machine, delayed discovery of quality prob- cell could understand what every- would eliminate those that were not lems would be eliminated. one else was doing. truly needed, and then would per- The first production cell, which Second, because machines were to form in a rapid sequence those that would make the company’s new- be made only when ordered, it was did create value – processing one est product line, the Q model, was important to introduce the concept machine, one design, one order at the acid test. A kaikaku team of that Toyota calls takt time. Takt a time. Batches, queues, backflows, Lantech’s best workers quickly re- time is determined by dividing the 10 HARVARD BUSINESS REVIEW September-October 1996
I D E A S A T W O R K
Toppling the Monuments at Pratt & Whitney
Lean thinkers call any machine shaped cast blades so that they of part to processing another, and
whose technology and scale re- would snap firmly into a turbine it was also difficult to remove
quire operating in a batch mode disc. The machines were techni- the encasing metal. Twenty-two
a “monument.” Because lean cal marvels: they could grind a skilled technicians were needed
thinking calls for continuous- blade in as little as three minutes, to maintain the elaborate com-
flow operations, monuments are and they were fed and unloaded puterized control system. (See the
evil – another form of muda, or robotically without any human exhibit “How Pratt & Whitney
waste. The recent lean conver- intervention. Grinds Blades.”)
sion at Pratt & Whitney, the jet But there were several prob- In fact, eight of the nine pro-
engine manufacturer, provides an lems. It took eight hours to shift cessing steps in this completely
excellent example of the need for the machines from making one automated system, plus the auto-
lean technologies. type of blade to making another. matic movement of parts in the
The primary monument in the Before being placed in the ma- automated storage system, added
company’s turbine-blade plant in chine, the blade had to be en- no value whatsoever – muda that
North Haven, Connecticut, was cased in a special metal alloy so is typical of such high-tech mon-
an $80 million complex of ten that it would not shatter during uments. Because of the long
computer-controlled, 12-axis the grinding. It was difficult to changeover times and the ten-day
blade-grinding machines that switch from processing one type journey through the grinding sys-
HowPratt & Whitney Grinds Blades
Old Automated System New System of Cells
Encase Gauge Grind Gauge Remove X-ray Grind Grind Grind EDM EDM Weld
blade in metal
metal casing
alloy
Automated storage-and- Inspect Acid
retrieval system clean
Grind Grind Grind Grind Grind Wash
Cell with eight three-axis grinding machines and two electrostatic discharge
AGVs – automated guide vehicles (EDM) machines (drawn to larger scale than the previous figure)
HARVARD BUSINESS REVIEW September-October 1996 11I D E A S A T W O R K
tem, the company was forced to machines in a tight cell fed manu- duced the amount of required
make batches of parts. That ally by one worker who, with a work space by more than 60%.
meant it needed to base produc- part-time helper, advanced parts Finally, by utilizing a small
tion on a forecast rather than on from one machine to the next, amount of direct labor rather
actual orders. standardized the work, gauged than an enormous amount of
In 1994, a radical-improve- the parts to check quality, unnecessary capital, the new
ment, or kaikaku, team conclud- changed each machine for the approach halved the total manu-
ed that the North Haven plant next type of part in 100 seconds, facturing cost, and the capital
needed a totally different ap- and made only what was needed investment for each new cell
proach. The team proposed re- when it was needed. ($1.7 million) paid for itself in less
placing each automated grinding By increasing actual processing than a year. After the first of the
machine with eight simple three- time from 3 minutes to 75 min- new cells went into operation at
axis grinding machines that had a utes, the new approach reduced the beginning of 1996, the North
proprietary mechanical-fixturing total time through the system Haven plant achieved a cost and
system, which eliminated the from ten days to 75 minutes. By quality position no one in the
need for encasing the blades. simplifying the machines, it world could match and demon-
After getting the go-ahead, the slashed downtime for change- strated lean technology at its best.
team arranged all the three-axis overs by more than 99% and re- (See the table “Less Is More.”)
Less Is More
Automated Grinding Process Lean Cell
Space per product cell 6,430 square feet 2,480 square feet
Distance blade travels during grinding process 2,500 feet 80 feet
Average inventory per cell 1,640 blades 15 blades
Batch size 250 blades 1 blade
Throughput time (sum of cycle times) 10 days 75 minutes
Need for hazardous processes acid cleaning, X ray no acid, no X ray
Changeover downtime 480 minutes 100 seconds
Grinding cost per blade (indexed to 100)* 100 49
Tooling cost for new type of blade (indexed to 100)* 100 30
*The exact figures are proprietary information. The point is that the cost of blade grinding has been cut in half and the cost of tooling by 70%.
12 HARVARD BUSINESS REVIEW September-October 1996I D E A S A T W O R K
number of orders placed by cus- own department. As long as we met and gave unfaltering support to the
tomers in a given period into the our daily production quota, we were new approach despite the setbacks.
amount of available production time left alone. What’s more, the real kick Hicks and his technical consultant,
in that period. For example, if cus- in the work was ‘fire fighting,’ in Anand Sharma, had the skills to
tomers were asking for 16 Q models which the ‘Lantech Volunteer Fire work the bugs out, and Zabaneh pro-
a day and the plant was running one Department’ went into crisis mode vided the emotional energy – what
eight-hour shift, takt time would be to get an emergency order through Ohno once called “the defiant atti-
one half hour. the system or to elimi-
Establishing takt time was critical nate a sudden produc-
to avoid the natural tendency to pro- tion bottleneck. I was Although the workforce
duce too fast, building up wasteful one of the best fire-
inventories, and was the best way to fighters at Lantech remained constant, the
focus the work team on getting all
the work done in the available time.
and I loved it.”
Hicks was propos-
number of shipped
When demand slackened and takt ing a new system machines doubled.
time was increased, it would be pos- of standardized work
sible to move some workers to other and takt time, which
tasks, such as maintenance. Simi- sounded like the kind of regimen tude” – to keep the kaikaku team
larly, an increase in demand and a that has traditionally caused produc- moving ahead even when no one
shortening of takt time would pro- tion workers to chafe. Moreover, he knew how to solve the next prob-
vide an excellent opportunity for was proposing to make complete lem. Gradually, and then more and
applying kaizen to the activity – for stretch wrappers, one at a time, in more rapidly, value began to flow.
figuring out how the same number precise response to customers’ de- By the fall of 1992, Lantech had
of people could produce a complete mands, and that idea seemed both converted its entire production sys-
machine in less time by further re- illogical and impossible to a group tem from departmentalized batch
fining each task and eliminating of employees with 20 years’ experi- methods to continuous flow in cells.
more muda. ence as batch thinkers. Finally, he Even the production of the largest
Finally, because customers or- claimed that if the work was stan- machine, the $150,000 H model,
dered each of the four basic models dardized by the work team, the ma- flowed continuously with a takt
of stretch wrappers with a wide vari- chines were realigned to permit con- time of one week.
ety of options, Lantech also needed tinuous flow, and takt time was The impact of the new approach
to figure out how to perform equip- adhered to with no working ahead, on performance was remarkable.
ment changeovers quickly. That way, there would be no more fires to Although the plant’s workforce re-
all variants of a basic model could fight. “It didn’t sound like much fun, mained virtually constant at around
be made in a continuous flow with and I thought it would never work,” 300, the number of shipped ma-
no stoppage. Underwood says. chines doubled between 1991 and
When the kaikaku team organized When the new cell concept was 1995. Moreover, Lantech could pro-
by Hicks proposed the new cell con- ready to go, it didn’t work. All kinds duce a machine in about half the
cept and the elimination of produc- of problems suddenly emerged, and space previously required, the num-
tion departments, many production the widespread feeling was that ber of defects reported by customers
Hicks was pushing an fell from 8 per machine in 1991 to .8
impractical concept. At in 1995, and start-to-finish produc-
Making machines one at a that point, Zabaneh, the tion time for the Q model (which
manufacturing director, had the highest volume and the
time, in response to played the primary role. shortest takt time) fell from 16
customers’ demands, “I was so fed up with
our failures and so taken
weeks to 14 hours.
A promise that Lancaster had
seemed illogical to workers. with the logic of the new made to his workforce in 1992 clear-
system that I threw my ly helped Lantech make the transi-
heart into it,” he recalls. tion so quickly: he had announced
workers and managers were baffled “I called a meeting of the work- that no one would be let go as a
or dismayed. As Bob Underwood, force and announced that I would result of the conversion. Instead,
one of the most highly skilled work- stay all night and all weekend to Lantech assigned the freed-up work-
ers on the floor, notes, “We were work on the problems we were en- ers to a companywide kaikaku team
used to a system in which each of us countering but that I would not to plan the improvement of other ac-
had a set of hard-earned skills–in my spend even one second discussing tivities. (These eventually included
case, it was the ability to adjust non- the possibility of going back to the office activities and helping suppli-
conforming parts so they would fit. old batch-and-queue system.” ers deliver parts just in time.) Under-
We were used to doing our own work As the transformation got under wood, the original skeptic and chief
as we saw fit at our own pace in our way, Lancaster took the long view fireman, headed the team. After
HARVARD BUSINESS REVIEW September-October 1996 13I D E A S A T W O R K
every improvement, Lantech trans- time required to get an order into
ferred the best workers in the re- production from three weeks to only
vamped process to the kaikaku two days.
team, making it clear that the as-
signment was a promotion, not a Letting the Customer Pull
punishment. the Product
As the lean revolution gained Lantech also found that it no
momentum in the plant, the com- longer needed most of its computer-
panywide kaikaku team turned its ized scheduling system and retired
attention to Lantech’s office. As its mainframe. It retained its MRP
Lancaster puts it, “If we could make system to provide long-term produc-
a machine in 14 hours, how could tion forecasts to suppliers, which
we live with an order-taking and still needed to know the capacity
credit-checking process that re- they would require to serve Lantech
quired three weeks? And why did we for one or two years. However, the
need an elaborate product-tracking MRP system was no longer used to
system to keep customers informed order parts from suppliers. Most sup-
about the status of their product if pliers delivered parts right before
we could schedule it and make it in the production cells needed them, or
only a few days?” just in time. Under Lantech’s kanban
Lantech employed the same tech- system, when a cell used a small box
niques it had used in the plant to of parts, a card was sent immediate-
transform the office. The kaikaku ly to the supplier of the box, telling it
team, including all the workers in- that it had to deliver another.
volved in the process and one out- At Lantech, day-to-day produc-
side technical consultant (Sharma), tion scheduling could now be done
on a large white board
in the sales office, where
The time to get an order orders were written as
soon as they were con-
into production fell from firmed. During our visits
three weeks to two days. to Lantech, the slots on
the board were filled
from three days to two
collectively mapped the entire value weeks ahead of the current date,
flow and looked for wasted time and and the plant was only manufactur-
effort. As the team rethought the ing machines with firm orders.
steps and as orders began to flow The highly visible white board
continuously from one adjacent was a remarkable spur to the sales
worker to the next, with no depart- force, particularly during any time
mental barriers, the best of the freed- when the blank space was increas-
up workers once again were assigned ing. It is an excellent example of yet
to the kaikaku team to lay the another lean technique: visual con-
groundwork for tackling the next trol. One of the principles of lean
activity. They remained on the team thinking is that if every employee
until growth in output or new busi- can see the status of an activity, he
ness initiatives created a need for or she will be able to take appropri-
them elsewhere. ate action.
After Lantech had applied those The sales office sent the roster
techniques to its entire order-taking of machines to be made each day to
and plant-scheduling system, it un- the four production cells. The new
derstood its costs much better, streamlined order flow was a strik-
which enabled it to set prices for ing contrast to the old labyrinth.
each machine more scientifically. In Rethinking the product develop-
addition, the company now could ment process was the final step in
explain its prices to distributors and Lantech’s transformation. From the
customers more clearly, which elim- early days of the plant conversion,
inated time-consuming haggling (a Lancaster knew that he would need
major source of muda). Finally, the to grow his business dramatically in
changes in the systems slashed the order to keep everyone busy as pro-
14 HARVARD BUSINESS REVIEW September-October 1996I D E A S A T W O R K
ductivity zoomed. That meant turn- without much success. A few bet- work nonstop until they had com-
ing strategic thinking on its head. the-company projects were pushed pleted their projects. The company
“I didn’t have time to find a new through by a designated “dictator” simply dropped the lower-ranking
business to go into, and I didn’t have or a “heavyweight” project leader. projects – the kind that had formerly
the money to buy out a major com- But in general, Lantech relied on so- clogged the engineering department.
petitor. Instead, I needed to revital- called lightweight team leaders to Under the new product-develop-
ize and expand my product range so coordinate the activities of the nu- ment system, a design progressed in
that I could sell more in an estab- merous technical specialists, who, a streamlined fashion.
lished market I knew well,” he says. in reality, continued to pursue their The S Series, the first product to
Realizing that his batch-and- individual priorities. In no case was come through the new system, dem-
queue product-development system the team leader – dictator or light- onstrated the potential of the ap-
would take years to come up with weight coordinator – responsible for proach. Launched in mid-1995, the
market-expanding products, he de- ensuring that the product pleased S Series was developed in a year –
cided to make new-product designs the customer and made money for a quarter of the time it had taken to
flow continuously, like orders and Lantech during its production life. develop its predecessor. It took only
machines. “We needed a design to In 1993, Lantech went to a new about half the hours of engineering
move continuously from the initial system of dedicated teams led by a that it would have required in the
concept to the launch of production. “directly responsible individual” old days. And its defect rate was sub-
That meant no stopping because of clearly charged with the success of stantially lower than that of previ-
the bureaucratic needs of our organi- the product during its lifetime. The ous new products.
zation, no backflows to correct mis- annual corporate-planning process Any business must be measured
takes, and no hitches during produc- identified the major products to be by its ability to make enough profit
tion ramp-up,” he says. developed and ranked them. The to renew itself. If the transition
Lantech had experimented with company assigned a dedicated team at Lantech had cost a fortune in
various types of development teams of specialists to each of the top-rank- new investment or had disrupted
in the late 1980s and early 1990s ing products and told those teams to the company’s ability to satisfy
Lantech’s Performance Leap
Batch and Queue, 1991 Continuous Flow, 1995
Development time for a new product family 3 to 4 years 1 year
Employee hours required to make one machine 160 80
Manufacturing space per machine 100 square feet 55 square feet
Average number of defects per delivered machine 8 .8
Value of in-process and finished-goods inventory* $2.6 million $1.9 million
Production throughput time 16 weeks 14 hours to 5 days
Product-delivery lead time† 4 to 20 weeks 1 to 4 weeks
*Sales more than doubled between 1991 and 1995. If Lantech’s traditional sales-to-inventory ratio had held constant, at least $5.2 million in inventory
would have been needed to support the 1995 sales volume.
†Thisis the period between the placement of the customer’s order and the delivery of the machine. In 1991, most of this time was spent in the production
system. In 1995, when sales soared and demand outstripped Lantech’s production capacity, most of it was waiting time for a production slot.
HARVARD BUSINESS REVIEW September-October 1996 15I D E A S A T W O R K
customers, it would have been an What’s more, Lancaster sees no ness gurus would have people be-
interesting technical exercise but end in sight. He notes that as layers lieve that the coupling of low-cost,
hardly a revolution in business prac- of muda are stripped away, more easily accessible data with interac-
tice. In fact, the amount of invest- muda is always exposed. Despite the tive educational software for knowl-
ment required was virtually zero. performance leap that Lantech has edge workers will produce a great
For the most part, workers freed up made, it can identify as many oppor- leap in productivity and well-being.
by the elimination of inefficient tunities for improvement today as it We are skeptical.
could four years ago. In the past 20 years, we have seen
Based on four years of the robotics revolution, the materi-
Workers freed up by the studying organizations als revolution (remember the predic-
like Lantech, we have de- tion that cars would have ceramic
elimination of inefficient veloped the following engines and airplanes would be built
tasks reconfigured tools rules of thumb: Convert-
ing a classic batch-and-
entirely of plastic?), the micro-
processor and personal-computer
and rethought processes. queue production system revolution, and the biotechnology
to clearly specified value revolution. Yet domestic product per
streams that flow contin- capita – the average amount of value
tasks reconfigured the tools and re- uously as they are pulled by the cus- created per person – in all the devel-
thought the office and development tomer will double labor productivity oped countries has remained stuck.
processes. And the transformation throughout the system (for direct, For the most part, the problem is
reduced the amount of computers, managerial, and technical workers, not the new technologies them-
space, and expensive tooling that the from raw materials to delivered selves. The problem is that they
company required. (See the table product) while cutting production often are misapplied and initially
“Lantech’s Performance Leap.”) throughput times and inventories by affect only a small part of the
The effect on customers was 90%. Errors reaching the customer, economy. A few companies, such as
equally dramatic. Lantech’s share scrap within the production process, Microsoft, grow from infants to gi-
of the stretch-wrapping market job-related injuries, time to market, ants overnight, but the great majori-
zoomed from 38% in 1991 to 50% and the effort required to develop ty of economic activities – construc-
in 1995, when the company sold new products will usually be cut in tion and housing, transportation,
2,585 units and had revenues of half. And a wider variety of products the food supply system, manufactur-
$60 million. As a result, the com- will be able to be offered at modest ing, and personal services – are af-
pany, which had suffered a large oper- additional cost. fected only over a long period, if at
ating loss in 1991, was generating Moreover, those gains are just the all. New technologies and invest-
solid profits by 1993 and had be- beginning. They are
come the industry’s leading finan- the kaikaku bonus
cial performer by 1994. from the initial radical Lean thinking could be
realignment of the val-
Pursuing Perfection ue stream. By making the antidote to economic
In an ironic twist, Lantech has re-
vitalized itself by banishing batches
continuous incremen-
tal improvements in
stagnation.
and their associated muda from the their pursuit of perfec-
design and production of a product tion, companies can usually double ments in human capital may gener-
whose sole use is to wrap batches of productivity again within two to ate growth over the long term, but
products for shipment within com- three years and halve inventories, er- lean thinking has demonstrated the
plex production and distribution rors, and lead times. Because a com- power to produce green shoots of
chains. Lancaster, therefore, has em- pany can put operations through growth all across this landscape
barked on a new strategic exercise: kaikaku and kaizen scrutiny over within a few years.
to think through how the emerging and over again, indefinitely, it will Lean thinking always works when
world of small-lot production and never reach an end to the improve- applied in a comprehensive way.
continuous flow will affect his cus- ments it can make. The problem is a shortage of man-
tomers’ packaging needs. Results of this magnitude could be agers with the knowledge and ener-
Meanwhile, Lantech as an organi- the antidote to stagnation in the ad- gy to make the leap. What compa-
zation is steadily striving for perfec- vanced economies. Conventional nies and the whole world need now
tion–a state in which every action in thinking about economic growth fo- is more Pat Lancasters taking heroic
the organization creates value for cuses on new technologies and addi- measures to define value correctly,
the customer. The pace of Lantech’s tional training – a focus that helps to identify the value stream, and to
improvement activities has not explain the fascination with the make value flow more and more per-
slowed. Every major activity in the falling costs of computing power and fectly at the pull of the customer.
company undergoes a three-day with the growing ease of moving Reprint 96511
kaizen several times a year. data around the planet. Many busi- To place an order, call 1-800-545-7685.
16 HARVARD BUSINESS REVIEW September-October 1996You can also read
