the new new product development game - nonaka and takeuchi

The New New ProductDevelopment Game

by Hirotaka Takeuchi and Ikujiro Nonaka

Reprint 86116

Harvard Business Review

HBRJANUARY–FEBRUARY 1986

The New New ProductDevelopment Game

Hirotaka Takeuchi and Ikujiro Nonaka

The rules of the game in new product develop-ment are changing. Many companies havediscovered that it takes more than the ac-

cepted basics of high quality, low cost, and differen-tiation to excel in today’s competitive market. It alsotakes speed and flexibility.

This change is reflected in the emphasis companiesare placing on new products as a source of new salesand profits. At 3M, for example, products less thanfive years old account for 25% of sales. A 1981 surveyof 700 U.S. companies indicated that new products

would account for one-third of all profits in the 1980s,an increase from one-fifth in the 1970s.1

This new emphasis on speed and flexibility callsfor a different approach for managing new productdevelopment. The traditional sequential or “relayrace” approach to product development—exempli-fied by the National Aeronautics and Space Admin-istration’s phased program planning (PPP) system—may conflict with the goals of maximum speed andflexibility. Instead, a holistic or “rugby” approach—where a team tries to go the distance as a unit, passing

Copyright © 1986 by the President and Fellows of Harvard College. All rights reserved.

In today’s fast-paced, fiercely competitive world of com-mercial new product development, speed and flexibilityare essential. Companies are increasingly realizing thatthe old, sequential approach to developing new productssimply won’t get the job done. Instead, companies in Japanand the United States are using a holistic method—as inrugby, the ball gets passed within the team as it moves asa unit up the field.

This holistic approach has six characteristics: built-ininstability, self-organizing project teams, overlapping de-velopment phases, “multilearning,” subtle control, andorganizational transfer of learning. The six pieces fit to-gether like a jigsaw puzzle, forming a fast flexible processfor new product development. Just as important, the newapproach can act as a change agent: it is a vehicle forintroducing creative, market-driven ideas and processesinto an old, riged organization.

Mr. Takeuchi is an associate professor and Mr. Nonaka,a professor at Hitotsubashi University in Japan. Mr.Takeuchi’s research has focused on marketing and globalcompetition. Mr. Nonaka has published widely in Japanon organizations, strategy, and marketing.

Authors’ note: We acknowledge the contribution of Ken-ichi Imai in the development of this article. An earlierversion of this article was coauthored by Ken-ichi Imai,Ikujiro Nonaka, and Hirotaka Takeuchi. It was entitled“Managing the New Product Development Process: HowJapanese Companies Learn and Unlearn” and was pre-sented at the seventy-fifth anniversary.

Colloquium on Productivity and Technology, HarvardBusiness School, March 28 and 29, 1984.

the ball back and forth—may better serve today’scompetitive requirements.

Under the old approach, a product developmentprocess moved like a relay race, with one group offunctional specialists passing the baton to the nextgroup. The project went sequentially from phase tophase: concept development, feasibility testing, prod-uct design, development process, pilot production,and final production. Under this method, functionswere specialized and segmented: the marketing peo-ple examined customer needs and perceptions indeveloping product concepts; the R&D engineers se-lected the appropriate design; the production engi-neers put it into shape; and other functional special-ists carried the baton at different stages of the race.

Under the rugby approach, the product develop-ment process emerges from the constant interactionof a hand-picked, multidisciplinary team whosemembers work together from start to finish. Ratherthan moving in defined, highly structured stages, theprocess is born out of the team members’ interplay(see Exhibit 1). A group of engineers, for example,may start to design the product (phase three) beforeall the results of the feasibility tests (phase two) arein. Or the team may be forced to reconsider a decisionas a result of later information. The team does notstop then, but engages in iterative experimentation.This goes on in even the latest phases of the develop-ment process.

Exhibit 1 illustrates the difference between thetraditional, linear approach to product developmentand the rugby approach. The sequential approach,labeled type A, is typified by the NASA-type PPPsystem. The overlap approach is represented by typeB, where the overlapping occurs only at the border ofadjacent phases, and type C, where the overlap ex-tends across several phases. We observed a type B

overlap at Fuji-Xerox and a type C overlap at Hondaand Canon.

This approach is essential for companies seeking todevelop new products quickly and flexibly. The shiftfrom a linear to an integrated approach encouragestrial and error and challenges the status quo. It stimu-lates new kinds of learning and thinking within theorganization at different levels and functions. Just asimportant, this strategy for product development canact as an agent of change for the larger organization.The energy and motivation the effort produces canspread throughout the big company and begin tobreak down some of the rigidities that have set in overtime.

In this article, we highlight companies both inJapan and in the United States that have taken a newapproach to managing the product development proc-ess. Our research examined such multinational com-panies as Fuji-Xerox, Canon, Honda, NEC, Epson,Brother, 3M, Xerox, and Hewlett-Packard. We thenanalyzed the development process of six specificproducts:

M FX-3500 medium-sized copier (introduced byFuji-Xerox in 1978)

M PC-10 personal-use copier (Canon, 1982)M City car with 1200 cc engine (Honda, 1981)M PC 8000 personal computer (NEC, 1979)M AE-1 single-lens reflex camera (Canon, 1976)M Auto Boy, known as the Sure Shot in the United

States, lens shutter camera, (Canon, 1979)

We selected each product on the basis of its impact,its visibility within the company as part of a “break-through” development process, the novelty of theproduct features at the time, the market success ofthe product, and the access to and availability of dataon each product.

EXHIBIT 1Sequential (A) vs. overlapping (B and C) phases of development

Phase 1 2 3 4 5 6

Phase

Phase

Type A

Type B

Type C

1 2 3 4 5 6

1 2 3 4 5 6

HARVARD BUSINESS REVIEW January–February 1986 3

MOVING THE SCRUM DOWNFIELD

From interviews with organization members fromthe CEO to young engineers, we learned that leadingcompanies show six characteristics in managing theirnew product development processes:

1. Built-in instability2. Self-organizing project teams3. Overlapping development phases4. “Multilearning”5. Subtle control6. Organizational transfer of learning

These characteristics are like pieces of a jigsawpuzzle. Each element, by itself, does not bring aboutspeed and flexibility. But taken as a whole, the char-acteristics can produce a powerful new set of dynam-ics that will make a difference.

Built-in InstabilityTop management kicks off the development processby signaling a broad goal or a general strategic direc-tion. It rarely hands out a clear-cut new productconcept or a specific work plan. But it offers a projectteam a wide measure of freedom and also establishesextremely challenging goals. For example, Fuji-Xer-ox’s top management asked for a radically differentcopier and gave the FX-3500 project team two yearsto come up with a machine that could be produced athalf the cost of its high-end line and still perform aswell.

Top management creates an element of tension inthe project team by giving it great freedom to carryout a project of strategic importance to the companyand by setting very challenging requirements. Anexecutive in charge of development at Honda re-marked, “It’s like putting the team members on thesecond floor, removing the ladder, and telling themto jump or else. I believe creativity is born by pushingpeople against the wall and pressuring them almostto the extreme.”

Self-organizing Project TeamsA project team takes on a self-organizing character asit is driven to a state of “zero information”—whereprior knowledge does not apply. Ambiguity and fluc-tuation abound in this state. Left to stew, the processbegins to create its own dynamic order.2 The projectteam begins to operate like a start-up company—ittakes initiatives and risks, and develops an inde-pendent agenda. At some point, the team begins tocreate its own concept.

A group possesses a self-organizing capability whenit exhibits three conditions: autonomy, self-transcen-dence, and cross-fertilization. In our study of the

various new product development teams, we foundall three conditions.

Autonomy. Headquarters’ involvement is limitedto providing guidance, money, and moral support atthe outset. On a day-to-day basis, top managementseldom intervenes; the team is free to set its owndirection. In a way, top management acts as a venturecapitalist. Or as one executive said, “We open up ourpurse but keep our mouth closed.”

This kind of autonomy was evident when IBMdeveloped its personal computer. A small group ofengineers began working on the machine in a con-verted warehouse in remote Boca Raton, Florida.Except for quarterly corporate reviews, headquartersin Armonk, New York allowed the Boca Raton groupto operate on its own. The group got the go-ahead totake unconventional steps such as selecting outsidesuppliers for its microprocessor and software pack-age.

We observed other examples of autonomy in ourcase studies:

M The Honda City project team, whose members’average age was 27, had these instructions from man-agement: to develop “the kind of car that the youthsegment would like to drive.” An engineer said,“It’s incredible how the company called in youngengineers like ourselves to design a car with a totallynew concept and gave us the freedom to do it ourway.”

M A small group of sales engineers who origi-nally sold microprocessors built the PC 8000 at NEC.The group started with no knowledge about per-sonal computers. “We were given the go-ahead fromtop management to proceed with the project, pro-vided we would develop the product by ourselvesand also be responsible for manufacturing, selling,and servicing it on our own,” remarked the project’shead.

Self-transcendence. The project teams appear to beabsorbed in a never-ending quest for “the limit.”Starting with the guidelines set forth by top manage-ment, they begin to establish their own goals andkeep on elevating them throughout the developmentprocess. By pursuing what appear at first to be con-tradictory goals, they devise ways to override thestatus quo and make the big discovery.

We observed many examples of self-transcendencein our field work. The Canon AE-1 project team cameup with new ideas to meet the challenging parame-ters set forth by top management. The companyasked the team to develop a high-quality, automaticexposure camera that had to be compact, lightweight,easy to use, and priced 30% lower than the prevailingprice of single-lens cameras. To reach this ambitioustarget, the project team achieved several firsts in

4 HARVARD BUSINESS REVIEW January–February 1986

camera design and production: an electronic brainconsisting of integrated circuits custom-made byTexas Instruments; modularized production, whichmade automation and mass production possible; andreduction in the number of parts by 30% to 40%. “Itwas a struggle because we had to deny our traditionalway of thinking,” recalled the head of the AE-1 team.“But we do that every day in the ongoing parts of ourbusiness,” responded another Canon executive. Theentire organization makes daily, incremental im-provements to strengthen what the president calls“the fundamentals”: R&D, production technology,selling prowess, and corporate culture.

The Honda City project team also achieved a break-through by transcending the status quo. The teamwas asked to develop a car with two competitivefeatures for the youth segment: efficiency in re-sources and fuel, and uncompromising quality at alow price. The team’s natural instinct was to developa scaled-down version of Honda’s best-selling Civicmodel. But after much debate, the team decided todevelop a car with a totally new concept. It chal-lenged the prevailing idea that a car should be longand low and designed a “short and tall” car. Con-vinced that an evolution toward a “machine mini-mum, human maximum” concept was inevitable,the team was willing to risk going against the indus-try norm.

Cross-fertilization. A project team consisting ofmembers with varying functional specializations,thought processes, and behavior patterns carries outnew product development. The Honda team, for ex-ample, consisted of hand-picked members from R&D,production, and sales. The company went a step furtherby placing a wide variety of personalities on the team.Such diversity fostered new ideas and concepts.

While selecting a diverse team is crucial, it isn’tuntil the members start to interact that cross-fertili-zation actually takes place. Fuji-Xerox located themultifunctional team building the FX-3500—con-sisting of members from the planning, design, pro-duction, sales, distribution, and evaluation depart-ments—in one large room. A project member gavethe following rationale for this step: “When all theteam members are located in one large room,someone’s information becomes yours, without eventrying. You then start thinking in terms of what’s bestor second best for the group at large and not onlyabout where you stand. If everyone understands theother person’s position, then each of us is more will-ing to give in, or at least to try to talk to each other.Initiatives emerge as a result.”

Overlapping Development PhasesThe self-organizing character of the team produces aunique dynamic or rhythm. Although the team mem-

bers start the project with different time hori-zons—with R&D people having the longest timehorizon and production people the shortest—they allmust work toward synchronizing their pace to meetdeadlines. Also, while the project team starts from“zero information,” each member soon begins toshare knowledge about the marketplace and the tech-nical community. As a result, the team begins towork as a unit. At some point, the individual and thewhole become inseparable. The individual’s rhythmand the group’s rhythm begin to overlap, creating awhole new pulse. This pulse serves as the drivingforce and moves the team forward.

But the quickness of the pulse varies in differentphases of development. The beat seems to be mostvigorous in the early phases and tapers off toward theend. A member of Canon’s PC-10 development teamdescribed this rhythm as follows: “When we aredebating about what kind of concept to create, ourminds go off in different directions and list alterna-tives. But when we are trying to come to grips withachieving both low cost and high reliability, ourminds work to integrate the various points of view.Conflict tends to occur when some are trying todifferentiate and others are trying to integrate. Theknack lies in creating this rhythm and knowing whento move from one state to the other.”

Under the sequential or relay race approach, aproject goes through several phases in a step-by-stepfashion, moving from one phase to the next only afterall the requirements of the preceding phase are satis-fied. These checkpoints control risk. But at the sametime, this approach leaves little room for integration.A bottleneck in one phase can slow or even halt theentire development process.

Under the holistic or rugby approach, the phasesoverlap considerably, which enables the group toabsorb the vibration or “noise” generated throughoutthe development process. When a bottleneck appears,the level of noise obviously increases. But the processdoes not come to a sudden halt; the team manages topush itself forward.

Fuji-Xerox inherited the PPP system (see type A inExhibit 1) from its parent company, but revised it intwo ways. First, it reduced the number of phases fromsix to four by redefining some of the phases andaggregating them differently. Second, it changed thelinear, sequential system into the so-called“sashimi” system. Sashimi is slices of raw fish ar-ranged on a plate, one slice overlapping the other (seeExhibit 2.)

The sashimi system requires extensive interactionnot only among project members but also with sup-pliers. The FX-3500 team invited them to join theproject at the very start (they eventually produced90% of the parts for the model). Each side regularly


visited the other’s plants and kept the informationchannel open at all times. This kind of exchange andopenness—both within the project team and withsuppliers—increases speed and flexibility. Fuji-Xeroxshortened the development time from 38 months foran earlier model to 29 months for the FX-3500.

If sashimi defines the Fuji-Xerox approach, thenrugby describes the overlapping at Honda. Like arugby team, the core project members at Honda stayintact from beginning to end and are responsible forcombining all of the phases.

In the relay-like PPP system, the crucial problemstend to occur at the points where one group passesthe project to the next. The rugby approach smoothsout this problem by maintaining continuity acrossphases.

The Auto Boy project proceeded with much over-lapping across phases as well. Canon’s design engi-neers stayed alert throughout the process to makesure their design was being converted into what theyhad in mind. The production people intruded ontothe design engineers’ turf to make sure that the de-sign was in accord with production scale economies.

The overlapping approach has both merits and de-merits. Greater speed and increased flexibility are the

“hard” merits. But the approach also has a set of“soft” merits relating to human resource manage-ment. The overlap approach enhances shared respon-sibility and cooperation, stimulates involvement andcommitment, sharpens a problem-solving focus, en-courages initiative taking, develops diversified skills,and heightens sensitivity toward market conditions.

The more obvious demerits result from having tomanage an intensive process. Problems include com-municating with the entire project team, maintain-ing close contact with suppliers, preparing severalcontingency plans, and handling surprises. This ap-proach also creates more tension and conflict in thegroup. As one project member aptly put it, “If some-one from development thinks that 1 out of 100 isgood, that’s a clear sign for going ahead. But if some-one from production thinks that 1 out of 100 is notgood, we’ve got to start all over. This gap in percep-tion creates conflict.”

The overlapping of phases also does away withtraditional notions about division of labor. Divisionof labor works well in a type A system, where man-agement clearly delineates tasks, expects all projectmembers to know their responsibilities, and evalu-ates each on an individual basis. Under a type B or C

EXHIBIT 2Fuji-Xerox’s product development schedule

ModelsUnitsproduced

Time

Bench3

Engineering13

Protoype19

Production

DWG FAB ASSY TEST

MOD

SOFTDWG DVT

DWG FAB ASSY DEBUG QAT

RCP

ASSYFABDWG

DVT

REG

DWG

PROD INS SIM DEBUG

QAT

DVT

PROD:

INS:

SIM:

DEBUG:

QAT:

REG:

MOD:

SOFT:

DVT:

RCP:

DWG:

FAB:

ASSY:

TEST:

Design anddrawing

Modification Debugging Production

Softwaredevelopment

Qualityassurancetest

InspectionFabrication

SimulatedmassproductionPassing

governmentregulation

Design verification testAssembly

Testing Reliability ofcritical parts


system, the company accomplishes the tasks throughwhat we call “shared division of labor,” where eachteam member feels responsible for—and is able towork on—any aspect of the project.

MultilearningBecause members of the project team stay in closetouch with outside sources of information, they canrespond quickly to changing market conditions.Team members engage in a continual process of trialand error to narrow down the number of alternativesthat they must consider. They also acquire broadknowledge and diverse skills, which help them createa versatile team capable of solving an array of prob-lems fast.

Such learning by doing manifests itself along twodimensions: across multiple levels (individual, group,and corporate) and across multiple functions. Werefer to these two dimensions of learning as “mul-tilearning.”

Multilevel learning. Learning at the individuallevel takes place in a number of ways. 3M, for exam-ple, encourages engineers to devote 15% of theircompany time to pursuing their “dream.” Canonutilizes peer pressure to foster individual learning. Adesign engineer for the PC-10 project explained, “Mysenior managers and some of my colleagues reallystudy hard. There is no way I can compete with themin the number of books they read. So whenever I havetime, I go to a department store and spend severalhours in the toy department. I observe what’s sellingand check out the new gadgets being used in the toys.They may give me a hint or two later on.”

Learning is pursued emphatically at the group levelas well. Honda, for example, dispatched several mem-bers of the City project team to Europe for threeweeks when the project reached a dead end at theconcept development phase. They were told simplyto “look around at what’s happening in Europe.”There they encountered the Mini-Cooper—a smallcar developed decades ago in the United Kingdom—which had a big impact on their design philosophy.

While it was developing the PC-10 copier, Canonteam members left the project offices to hold anumber of meetings in nearby hotels. In one of theearly meetings, the entire project team broke up intosubgroups, each with a representative from the designteam and the production team. Each subgroup wastold to calculate the cost of a key part and figure outways of reducing that cost by one-third. “Since everysubgroup faced the same mandate and the same dead-line, we had no choice,” recalled one project member.Learning took place in a hurry.

Learning at the corporate level is best achieved byestablishing a company-wide movement or program.Fuji-Xerox, for example, used the total quality con-

trol (TQC) movement as a basis for changing thecorporate mentality. TQC was designed to heightenthe entire organization’s sensitivity toward simulta-neous quality and productivity improvement, mar-ket orientation, cost reduction, and work simpli-fication. To achieve these goals, everyone in theorganization had to learn the basics of techniques likestatistical quality control and value engineering.

Hewlett-Packard embarked on a four-phased train-ing program in marketing as part of the corporation’saim to become more market-oriented. The companynow brings in top academics and business consult-ants to spread the marketing message. It also appliestechniques borrowed from the consumer packagedgoods industry, such as focus group interviews, quan-titative market research, and test marketing. Further,the company has created a corporate marketing divi-sion to accelerate what one insider calls “the transi-tion from a company run by engineers for engineersto one with a stronger marketing focus.”

Multifunctional learning. Experts are encouragedto accumulate experience in areas other than theirown. For instance:

M All the project members who developed Epson’sfirst miniprinter were mechanical engineers whoknew little about electronics at the start. So theleader of the project team, also a mechanical engi-neer, returned to his alma mater as a researcher andstudied electrical engineering for two years. He didthis while the project was under way. By the timethey had completed the miniprinter project, all theengineers were knowledgeable about electronics. “Itell my people to be well-versed in two technologicalfields and in two functional areas, like design andmarketing,” the leader said. “Even in an engineering-oriented company like ours, you can’t get ahead with-out the ability to foresee developments in the mar-ket.”

M The team working on NEC’s PC 8000 consistedof sales engineers from the Electronic Devices Divi-sion. They acquired much of the know-how to de-velop the company’s first personal computer by put-ting together TK 80, a computer kit, and introducingit on the market two years in advance of the PC 8000;and by stationing themselves for about a year, evenon weekends, at BIT-IN, an NEC service center in themiddle of Akihabara, talking with hobbyists andlearning the user’s viewpoint.

These examples show the important role that mul-tilearning plays in the company’s overall human re-source management program. It fosters initiative andlearning by doing on the part of the employees andhelps keep them up to date with the latest develop-ments. It also serves as a basis for creating a climatethat can bring about organizational transition.


Subtle ControlAlthough project teams are largely on their own, theyare not uncontrolled. Management establishesenough checkpoints to prevent instability, ambigu-ity, and tension from turning into chaos. At the sametime, management avoids the kind of rigid controlthat impairs creativity and spontaneity. Instead, theemphasis is on “self-control,” “control through peerpressure,” and “control by love,” which collectivelywe call “subtle control.”

Subtle control is exercised in the new productdevelopment process in seven ways:

1. Selecting the right people for the project teamwhile monitoring shifts in group dynamics andadding or dropping members when necessary.“We would add an older and more conservativemember to the team should the balance shifttoo much toward radicalism,” said a Hondaexecutive. “We carefully pick the project mem-bers after long deliberation. We analyze thedifferent personalities to see if they would getalong. Most people do get along, thanks to ourcommon set of values.”

2. Creating an open work environment, as in thecase of Fuji-Xerox.

3. Encouraging engineers to go out into the fieldand listen to what customers and dealers haveto say. “A design engineer may be tempted totake the easy way out at times, but may reflecton what the customer had to say and try to findsome way of meeting that requirement,” notedan engineer from Fuji-Xerox.

4. Establishing an evaluation and reward systembased on group performance. Canon, for exam-ple, applied for patents for products from thePC-10 project on a group basis.

5. Managing the differences in rhythm throughoutthe development process. As mentioned earlier,the rhythm is most vigorous in the early phasesand tapers off toward the end.

6. Tolerating and anticipating mistakes. Engineersat Honda are fond of saying that “a 1% successrate is supported by mistakes made 99% of thetime.” A Brother executive in charge of R&Dsaid, “It’s natural for young engineers to makea lot of mistakes. The key lies in finding themistakes early and taking steps to correct themimmediately. We’ve taken steps to expedite thetrial production cycle for that reason.” A 3Mexecutive noted, “I believe we learn more frommistakes than from successes. That’s not to saywe should make mistakes easily. But if we domake mistakes, we ought to make them crea-tively.”

7. Encouraging suppliers to become self-organiz-ing. Involving them early during design is a stepin the right direction. But the project teamshould refrain from telling suppliers what to do.As Xerox found out, suppliers produce betterresults when they have the problem explainedto them and are allowed to decide how to fur-nish the parts.

Transfer of LearningThe drive to accumulate knowledge across levels andfunctions is only one aspect of learning. We observedan equally strong drive on the part of the projectmembers to transfer their learning to others outsidethe group.

Transfer of learning to subsequent new productdevelopment projectsor tootherdivisions intheorgani-zation takes place regularly. In several of the compa-nies we studied, the transfer took place through “os-mosis”—by assigning key individuals to subsequentprojects. A Honda executive explained, “If the factoryis up and running and the early-period claims areresolved, we dismantle the project team, leaving onlya few people to follow through. Since we have only a

Corporate rugby scores

Some companies are already making headway inspeeding up new product development:

A new copier—the 9900—took Xerox three years todevelop, whereas the company spent more than fiveyears developing a comparable earlier model.

A portable Brother printer—the EP-20—was devel-oped in less than two years. It took the company morethan four years to develop an earlier model.

One of John Sculley’s top priorities, when appointedpresident of Apple in 1984, was to cut the company’sproduct development time from 3.5 years down to oneyear.

Other organizations are beginning to add flexibilityto product development:

Black & Decker recently unveiled 50 new power toolproducts at the National Hardware Show in Chicagoto compete more effectively with Japanese power toolmakers.

When Yamaha threatened its leadership position inthe Japanese market in 1982, Honda unleashed some30 new motorcycle models within a six-month period.

IBM broke from its tradition of designing everythinginternally and used a microprocessor designed by IntelCorporation and a basic operating system designedby Microsoft Corporation to develop its personal com-puter.


limited number of unusually able people, we turnthem loose on another key project immediately.”

Knowledge is also transmitted in the organizationby converting project activities to standard practice.At Canon, for example, the Auto Boy project pro-duced a format for conducting reviews that was usedin later projects. One team member recalled, “Weused to meet once a month or so to exchange noteson individual subprojects in progress and once inthree months or so to discuss the project from a largerperspective. This pattern later became institutional-ized into the monthly and quarterly progress reviewsadopted from the PC-10 minicopier project.”

Naturally, companies try to institutionalize thelessons derived from their successes. IBM is trying toemulate the personal computer development pro-ject—which was completed in 13 months with out-side help—throughout the company.

At Hewlett-Packard, the personal computer groupis reprogramming the way the entire company devel-ops and sells new products. In the past, the companywas famous for designing a machine for a particularcustomer and charging a premium price. But it re-cently engineered its ThinkJet—a quiet inkjetprinter—for low-cost mass production and priced itlow. Within six months of its introduction, theprinter captured 10% of the low-end market.Hewlett-Packard began to apply what it had learnedfrom designing and pricing ThinkJet to its minicom-puter line. Within months of putting ThinkJet on themarket, the company introduced a minicomputersystem for a broad corporate audience at a modestprice.

But institutionalization, when carried too far, cancreate its own danger. Passing down words of wisdomfrom the past or establishing standard practices basedon success stories works well when the externalenvironment is stable. Changes in the environment,however, can quickly make such lessons impractical.

Several companies have tried to unlearn old les-sons. Unlearning helps keep the development teamin tune with the realities of the outside environment.It also acts as a springboard for making more incre-mental improvements.

Much of the unlearning is triggered by changes inthe environment. But some companies consciouslypursue unlearning. Consider these examples:

M Epson’s target is to have the next-generationmodel in development stages as a new model is beingintroduced on the market. The company tells itsproject teams that the next-generation model mustbe at least 40% better than the existing one.

M When Honda was building the third-generationCivic model, its project team opted to scrap all theold parts and start anew. When the car made its debut

before the public, all the new parts were displayedright next to the car at the request of the projectmembers. The car won the 1984 Car of the YearAward in Japan.

M Fuji-Xerox has refined its sashimi approach, firstadopted for the FX-3500. Compared with that effort,a new product today requires one-half of the originaltotal manpower. Fuji-Xerox has also reduced theproduct development cycle from 4 years to 24months.

SOME LIMITATIONS

Some words of caution are in order. The holisticapproach to product development may not work inall situations. It has some built-in limitations:

M It requires extraordinary effort on the part of allproject members throughout the span of the develop-ment process. Sometimes, team members recordmonthly overtime of 100 hours during the peak and60 hours during the rest of the project.

M It may not apply to breakthrough projects thatrequire a revolutionary innovation. This limitationmay be particularly true in biotechnology or chemis-try.

M It may not apply to mammoth projects like thosein the aerospace business, where the sheer projectscale limits extensive face-to-face discussions.

M It may not apply to organizations where productdevelopment is masterminded by a genius whomakes the invention and hands down a well-definedset of specifications for people below to follow.

Some limitations also stem from the scope of ourresearch. Our sample size was limited to a handful ofcompanies, and our findings were drawn, for the mostpart, from observing how the development processwas managed in Japan. General conclusions, there-fore, must be made with some caution. But as newapproaches to product development gain acceptancein the United States, the difference between the twocountries may not be so much a difference of kind asa difference of degree.

MANAGERIAL IMPLICATIONS

Changes in the environment—intensified competi-tion, a splintered mass market, shortened product lifecycles, and advanced technology and automat-ion—are forcing managements to reconsider the tra-ditional ways of creating products. A product thatarrives a few months late can easily lose severalmonths of payback. A product designed by an engi-neer afflicted with the “next bench” syndrome—the


habit of designing a product by asking the coworkeron the next bench what kind of a product he or shewould like—may not meet the flexible requirementsof the marketplace.

To achieve speed and flexibility, companies mustmanage the product development process differently.Three kinds of changes should be considered.

First, companies need to adopt a management stylethat can promote the process. Executives must rec-ognize at the outset that product development sel-dom proceeds in a linear and static manner. It in-volves an iterative and dynamic process of trial anderror. To manage such a process, companies mustmaintain a highly adaptive style.

Because projects do not proceed in a totally rationaland consistent manner, adaptability is particularlyimportant. Consider, for example, situations where:

M Top management encourages trial and error bypurposely keeping goals broad and by tolerating am-biguity. But at the same time, it sets challenging goalsand creates tension within the group and within theorganization.

M The process by which variety is amplified (dif-ferentiation) and reduced (integration) takes placethroughout the overlapping phases of the develop-ment cycle. Differentiation, however, tends to domi-nate the concept development phase of the cycle, andintegration begins to take over the subsequentphases.

M Operational decisions are made incrementally,but important strategic decisions are delayed asmuch as possible in order to allow a more flexible

response to last-minute feedback from the market-place.

Because management exercises subtle forms ofcontrol throughout the development process, theseseemingly contradictory goals do not create totalconfusion. Subtle control is also consistent with theself-organizing character of the project teams.

Second, a different kind of learning is required.Under the traditional approach, a highly competentgroup of specialists undertakes new product develop-ment. An elite group of technical experts does mostof the learning. Knowledge is accumulated on anindividual basis, within a narrow area of focus—whatwe call learning in depth.

In contrast, under the new approach (in its extremeform) nonexperts undertake product development.They are encouraged to acquire the necessary knowl-edge and skills on the job. Unlike the experts, whocannot tolerate mistakes even 1% of the time, thenonexperts are willing to challenge the status quo.But to do so, they must accumulate knowledge fromacross all areas of management, across different levelsof the organization, functional specializations, andeven organizational boundaries. Such learning inbreadth serves as the necessary condition for shareddivision of labor to function effectively.

Third, management should assign a different mis-sion to new product development. Most companieshave treated it primarily as a generator of future reve-nue streams. But in some companies, new productdevelopment also acts as a catalyst to bring aboutchange in the organization. The personal computerproject, for example, is said to have changed the wayIBM thinks. Projects coming out of Hewlett-Packard’spersonal computer group, including ThinkJet, havechanged its engineering-driven culture.

No company finds it easy to mobilize itself forchange, especially in noncrisis situations. But theself-transcendent nature of the project teams and thehectic pace at which the team members work help totrigger a sense of crisis or urgency throughout theorganization. A development project of strategic im-portance to the company, therefore, can create awartime working environment even during times ofpeace.

Changes affecting the entire organization are alsodifficult to carry out within highly structured com-panies, especially seniority-based companies like theones commonly found in Japan. But unconventionalmoves, which may be difficult to pull off during timesof peace, can be legitimized during times of war. Thusmanagement can uproot a competent manager orassign a very young engineer to the project withoutencountering much resistance.

Once the project team is formed, it begins to rise

The sport of rugby

One of the charms of the Rugby Union game is the infi-nite variety of its possible tactics. Whatever tactics ateam aims to adopt, the first essential is a strong andskilful [sic] pack of forwards capable of winning initialpossession from the set pieces. For, with the ball in itshands, a team is in a position to dictate tactics whichwill make the best use of its own particular talents, atthe same time probing for and exposing weaknessesin the opposing team. The ideal team has fast andclever half-backs and three-quarters who, with running,passing, and shrewd kicking, will make sure that thepossession won by the forwards is employed to themaximum embarrassment of the opposing team.

From The Oxford Companion to World Sports and Games ed.John Arlott (London: Oxford University Press) copyright © 1975 Re-printed with permission of the publisher.


in stature because of its visibility (“we’ve been hand-picked”), its legitimate power (“we have uncondi-tional support from the top to create somethingnew”), and its sense of mission (“we’re working tosolve a crisis”). It serves as a motor for corporatechange as project members from a variety of func-tional areas begin to take strategic initiatives thatsometimes go beyond the company’s conventionaldomain and as their knowledge gets transferred tosubsequent projects.

The environment in which any multinational com-pany—from the United States or Japan—operates haschanged dramatically in recent years. The rules of thegame for competing effectively in today’s world mar-ket have changed accordingly. Multinationals mustachieve speed and flexibility in developing products;

to do so requires the use of a dynamic process involv-ing much reliance on trial and error and learning bydoing. What we need today is constant innovation ina world of constant change.

1. Booz Allen & Hamilton survey reported in Susan Fraker,“High-Speed Management for the High-Tech Age,” Fortune,March 5, 1984, p. 38.

2. See, for example, Ilya Prigozine, From Being to Becoming (SanFrancisco, Calif.: Freeman, 1980); Eric Jantsch, “Unifying Princi-ples of Evolution,” in Eric Jantsch, ed., The Evolutionary Vision(Boulder, Colorado: Westview Press, 1981); and Devendra Sahal, “AUnified Theory of Self-Organization,” Journal of Cybernetics,April–June, 1979, p. 127. See also Todao Kagono, Ikujiro Nonaka,Kiyonari Sakakibara, and Akihiro Okumura, Strategic vs. Evolu-tionary Management: A U.S.-Japan Comparison of Strategy andOrganization (Amsterdam: North-Holland, 1985).


Harvard Business ReviewHBR Subscriptions Harvard Business Review

U.S. and Canada Outside U.S. and CanadaSubscription Service Tower HouseP.O. Box 52623 Sovereign ParkBoulder, CO 80322-2623 Lathkill Street

Market HarboroughLeicestershire LE16 9EF

Telephone: (800) 274-3214 Telephone: 44-85-846-8888Fax: (617) 496-8145 Fax: 44-85-843-4958American Express, MasterCard, Visa accepted. Billing available.

HBR Article ReprintsHBR Index and Other CatalogsHBS CasesHBS Press Books

Harvard Business School PublishingCustomer Service - Box 230-560 Harvard WayBoston, MA 02163Telephone: U.S. and Canada (800) 545-7685Outside U.S. and Canada: (617) 495-6117 or 495-6192Fax: (617) 495-6985Internet address: [email protected]

HBS Management ProductionsVideos

Harvard Business School Management Productions videos areproduced by award-winning documentary filmmakers. You’ll findthem lively, engaging, and informative.

HBR Custom Reprints Please inquire about HBR’s custom service for quantity discounts.We will print your company’s logo on the cover of reprints,collections, or books in black and white, two color, or four color. Theprocess 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 EditorHarvard Business School PublishingBox 230-560 Harvard WayBoston, MA 02163Telephone: (617) 495-6849

the new new product development game - nonaka and takeuchi

Documents