HBRiANUAtY-riHUAir 1995

Disruptive Technologies:Catching the Wave

by Joseph L. Bower and Clayton M. Christensen

ne of the most consistent patterns inbusiness is the failure of leading compa-nies to stay at the top of their industrieswhen technologies or markets change.

Goodyear and Firestone entered the radial-tire mar-ket quite late. Xerox let Canon create the small-copier market. Bucyrus-Erie allowed Caterpillarand Deere to take over the mechanical excavatormarket. Sears gave way to Wal-Mart.

The pattern of failure has been especially strik-ing in the computer industry. IBM dominated themainframe market but missed by years the emer-gence of minicomputers, which were technologi-cally much simpler than mainframes. DigitalEquipment dominated the minicomputer marketwith innovations like its VAX architecture butmissed the personal-computer market almost com-pletely. Apple Computer led the world of personalcomputing and established the standard for user-friendly computing but lagged five years behind thcleaders in bringing its portable computer to market.

Why is it that companies like these invest aggres-sively-and successfully-in the technologies neces-sary to retain their current customers but then fail

DRAWING BV CHRISTOPHER BING

to make certain other technological investmentsthat customers of the future will demand* Un-doubtedly, bureaucracy, arrogance, tired executiveblood, poor planning, and short-term investmenthorizons have all played a role. But a more funda-mental reason lies at the heart of the paradox: lead-ing companies succumb to one of the most popular,and valuable, management dogmas. They stay closeto their customers.

Altbough most managers like to think tbey are incontrol, customers wield extraordinary power in di-recting a company's investments. Before managersdecide to launch a technology, develop a product,build a plant, or establish new channels of distribu-tion, they must look to their customers first: Dotheir customers want it? How big will the marketbe? Will tbe investment be profitable? The more as-tutely managers ask and answer these questions,

loseph L. Bower is the Donald Kirk David Professor ofBusiness Administration at the Harvard BusinessSchool in Boston. Massachusetts. Clayton M. Chris-tensen. an assistant professor at the Harvard BusinessSchcx}}. speciahzes in managing the commercializationof advanced technology.

43

DISRUPTIVE TECHNOLOGIES

the more completely their investments will bealigned with the needs of their customers.

This is the way a well-managed company shouldoperate. Right? But what happens when customersreject a new technology, product concept, or way ofdoing business because it does not address theirneeds as effectively as a company's current ap-proach? The large photocopying centers that repre-sented the core of Xerox's customer base at first hadno use for small, slow tabletop copiers. Tbe excava-tion contractors that had relied on Bucyrus-Erie'sbig-bucket steam- and diesel-powered cable sbovelsdidn't want hydraulic excavators because initiallythey v^ere small and weak. IBM's large commercial,government, and industrial customers saw no im-mediate use for minicomputers. In each instance,companies listened to their customers, gave themthe product performance they were looking for,and, in the end, were hurt by the very technologiestheir customers led them to ignore.

We bave seen this pattern repeatedly in an on-going study of leading companies in a variety of in-dustries that have confronted technological change.The research shows that most wcil-managed, estab-lished companies are consistently abead of theirindustries in developing and commercializing newtechnologies-from incremental improvements toradically new ;ipproachcs - as long as those tech-

Managers must beware ofignoring new technologies thatd^n't initially meet the needsof their mainstream i

nologics address the next-generation performanceneeds of their customers. However, tbese samecompanies are rarely in the forefront of commer-cializing new technologies that don't initially meetthe needs of mainstream customers and appeal onlyto small or emerging markets.

Using the rational, analytical investment pro-cesses that most well-managed companies have de-veloped, it is nearly impossible to build a cogentcase for diverting resources from known customerneeds in established markets to markets and cus-tomers that seem insignificant or do not yet ex^t.After all, meeting the needs of established cus-tomers and fending off competitors takes all the re-sources a company has, and then some. In well-managed companies, the processes used to identify

customers' needs, forecast technological trends,assess profitability, allocate resources across com-peting proposals for investment, and take newproducts to market are focused - for all the rightreasons-on current customers and markets. Theseprocesses are designed to weed out proposed prod-ucts and technologies that do nor address cus-tomers' needs.

In fact, the processes and incentives that compa-nies use to keep focused on their main customerswork so well that they blind those companies toimportant new technologies in emerging markets.Many companies have learned the hard way theperils of ignoring new technologies that do not ini-tially meet the needs of mainstream customers. Forexample, although personal computers did notmeet the requirements of mainstream minicomputer users in the early \9H0s, the computing powerot the desktop machines improved at a much fasterrate than minicomputer users' demands for com-puting power did. As a result, personal computerscaught up with the computing needs of many of thecustomers of Wang, Prime, Nixdorf, Data General,and Digital Equipment, Today they are perfor-mance-competitive with minicomputers in manyapplications. For the minicomputer makers, keep-ing close to mainstream customers and ignoringwhat were initially low-performance desktop tech-

nologies used by seemingly insignifi-cant customers in emerging marketswas a rational decision-hut one thatproved disastrous.

The technological changes thatdamage established companies areusually not radically new or difficultfrom a technological point of view.They do, however, have two impor-tant characteristics: First, they typi-cally present a different package of

performance attributes - ones that, at least at theoutset, are not valued by existing customers. Sec-ond, the performance attributes that existing cus-tomers do value improve at such a rapid rate thatthe new technology can later invade those estab-lished markets. Only at this point will mainstreamcustomers want the technology. Unfortunately forthe established suppliers, by then it is olten toolate: the pioneers of the new technology dominatethe market.

It follows, then, that senior executives must firsthe able to spot the technologies that seem to fall in-to this category. Next, to commercialize and devel-op the new technologies, managers must protectthem from the processes and incentives that aregeared to serving established customers. And the

44 HARVARD BUSINESS REVIEW January-February 1995

only way to protect them is to create organizationsthat are completely independent from the main-stream business.

o industry demonstrates the danger ofstaying too close to customers moredramatically than the hard-disk-driveindustry. Between 1976 and 1992, disk-

drive performance improved at a stunning rate:the physical size of a 100-mcgahyte (MB) systemshrank from S,400 to 8 cubic inches, and the costper MB fell from $560 to $5. Technological change,of course, drove these breathtaking achievements.Ahout half of the improvement came from a hostof radical advances tbat were critical to continuedimprovements in disk-drive performance,- the otherbalf came from incremental advances.

The pattern in the disk-drive industry has beenrepeated in many other industries: the leading, es-tablished companies have consistently led the in-dustry in developing and adopting new technolo-gies that their customers demanded - even whenthose technologies required completely differenttechnological competencies and manufacturing ca-pahilities from the ones the companies had. In spiteof this aggressive technological posture, no singledisk-drive manufacturer has been able to dominatethe industry for more than a few years. A series ofcompanies have entered the business and risen toprominence, only to be toppled by newcomers whopursued technologies that at first did not meet theneeds of mainstream customers. As a result, notone of the independent disk-drive companies thatexisted in 1976 survives today.

To explain the differences in the impact of cer-tain kinds of technological innovations on a givenindustry, the concept oi performance trajectories-the rate at which the performance of a product hasimproved, and is expected to improve, over t ime-can be helpful. Almost every industry has a criticalperformance trajectory. In mechanical excavators,the critical traicctory is the annual improvement incubic yards of earth moved per minute. In photo-copiers, an important performance trajectory is im-provement in number of copies per minute. In diskdrives, one crucial measure of performance is stor-age capacity, which has advanced S0% each year onaverage for a given size of drive.

Different types of technological innovations af-fect performance trajectories in different ways. Onthe one hand, sustaining technologies tend tomaintain a rate of improvement; that is, they givecustomers something more or hetter in the at-tributes they already value. For example, thin-filmcomponents in disk drives, which replaced conven-

tional ferrite heads and oxide disks hetween 1982and 1990, enabled information to be recorded moredensely on disks. Engineers had been pushing thelimits of the performance they could wring fromferrite heads and oxide disks, but the drives em-ploying these technologies seemed to have reachedthe natural limits of an S curve. At that point, newthin-film technologies emerged that restored - orsustained- the historical trajectory of performanceimprovement.

On the other hand, disruptive technologies intro-duce a very different package of attributes from theone mainstream customers historically value, andthey often perform far worse along one or two di-mensions that are particularly important to thosecustomers. As a rule, mainstream customers areunwilling to use a disruptive product in applica-tions they know and understand. At first, then, dis-ruptive technologies tend to be used and valued on-ly in new markets or new applications; in fact, theygenerally make possihle the emergence of new mar-kets. For example, Sony's early transistor radiossacrificed sound fidelity but created a market forportable radios by offering a new and differentpackage of attributes-small size, light weight, andportability.

In the history of the hard-disk-drive industry, theleaders stumbled at each point of disruptive tecb-nological change: when the diameter of disk drivesshrank from theoriginal 14 inches to H inches, thento S.25 inches, and fmally to 3.5 inches. Each ofthese new arcbitectures^initially offered the marketsubstantially less storage capacity than the typicaluser in the established market required. For exam-ple, the 8-incb drive offered 20 MB when it was in-troduced, wbile the primary market for disk drives 'at that time-mainframes-required 200 MB on av- 'erage. Not surprisingly, the leading computer man-ufacturers rejected the 8-inch architecture at first.As a result, their suppliers, whose mainstreamproducts consisted of 14-inch drives with morethan 200 MB of capacity, did not pursue the disrup-tive products aggressively. The pattern was repeat-ed when the 5.25-inch and 3.5-inch drives emerged:established computer makers rejected the drives asinadequate, and, in turn, their disk-drive suppliersignored them as well.

But while they offered less storage capacity, thedisruptive architectures created other important at-tributes-internal power supplies and smaller sizel8-inch drives); still smaller size and low-cost step-per motors (5.25-incb drives); and ruggedness, lightweight, and low-power consumption 13.5-inch drives).From the late 1970s to the mid-1980s, the avail-ability of the three drives made possible the devel-

HARVARD BUSINESS REVIEW lanuary-February 1995 45

How Disk-Drive Performance Mei Market Needs

3000

2000

1000

700600500400

300

200

100

'GoaoU

d-D

isI

5X

706050

40

30

20

10

765

4

3

Point at which hard-diskdrives invademinicomputer market

Point at which hard-diskdrives invade partable-computer market

Point at which hard-diskdrives invade personal-computer market

75 76 -77 '78 '79 '80 '81 '82 '83 '84 '85 '86 '87 '88 '89 '90

Year

opment of new markets for minicomputers, desk-top PCs, and portable computers, respectively.

Although thc smaller drives represented disrup-tive technological ehange, eaeh was technological-ly straightforward. In fact, there were engineers atmany leading companies who championed the newteehnologies and built working prototypes withbootlegged resources before management gave

a formal go-ahead. Still, tbe lending companieseould not move the products tbrougb their organi-zations and into tbe market in a timely way. Eacbtime a disruptive technology emerged, betweenone-balf and two-thirds of tbe establisbed manu-facturers failed to introduce models employing tbenew arcbitecture-in stark contrast to tbeir timelylaunches of eritieal sustaining tecbnologies. Tbose

46 HARVARD BUSINESS REVIEW lanuary-February

DISRUPTIVE TECHNOLOGIES

companies that finally did launch new models typi-cally lagged behind entrant companies by twoyears-eons in an industry whose products' life cy-cles are often two years. Three waves of entrantcompanies led these revolutions; they first capturedthe new markets and then dethroned the leadingcompanies in the mainstream markets.

How could technologies that were initially infe-rior and usetul only to new markets eventually

None ol the established leadersin the disk-drive industiylearned from the experiences ofthose that fell before them.

threaten loading companies in established mar-kets? Once the dismptive architectures became es-tablished in theirnew markets, sustaining innova-tions raised each architecture's performance alongsteep trajectories - so steep that the performanceavailable from each architecture soon satisfied theneeds of customers in the established markets. Forexample, the S.25-inch drive, whose initial S MB ofcapacity in lySO was only a fraction of tbc capacitythat tbe minicomputer market needed, became ful-ly performance-competitive in the minicomputermarket by 1986 and in the mainframe market by1991. |See tbc graph "How Disk-Drive PerformanceMet Market Needs.")

A company's revenue and cost structures playa critical role in the way it evaluates proposedtechnological innovations. Generally, disruptivetechnologies look financially unattractive to estab-lished companies. Tbc potential revenues fromthe discernible markets are small, and it is oftendifficult to project how big the markets for tbetechnology will be over the long term. As a result,managers typically conclude that the technology can-not make a meaningful contribution to corporategrowtb and, therefore, that it is not wortb the man-agement effort required to develop it. In addition.established companies bave often installed highercost structures to serve sustaining technologiestban tbose required by disruptive technologies. Asa result, managers typically see tbemselves as hav-ing two choices when deeiding whether to pursuedisruptive technologies. One is to go downmarketand accept the lower profit margins of tbe emergingmarkets that the disruptive technologies will ini-tially serve. The other is to go upmarket witb sus-

taining technologies and enter market segmentswbose profit margins are alluringly high. [For ex-ample, the margins ot IBM's mainframes are stillhigher tban those of PCs). Any rational resource-allocation process in companies serving establishedmarkets will cboose going upmarket ratber thangoing down.

Managers of companies that have championeddisruptive technologies in emerging markets look

at tbe world quite differently. Witb-out tbe high cost structures of tbeirestablished counterparts, these com-panies find tbe emerging markets ap-pealing. Once the companies bavesecured a foothold in tbe marketsand improved tbe performance oftheir technologies, tbe establishedmarkets above them, served byhigh-cost suppliers, look appetizing.When tbey do attack, the entrant

companies find the established players to be easyand unprepared opponents because the opponentshave been looking upmarket themselves, discount-ing the threat Irom below.

It is tempting to stop at this point and concludethat a valuable lesson has been learned: managerscan avoid missing tbe next wave by paying carefulattention to potentially disruptive technologiesthat do not meet current customers' needs. But rec-ognizing the pattern and figuring out how to breakit are two different tbings. Although entrants in-vaded established markets with new technologiesthree times in succession, none of the establishedleaders in tbe disk-drive industry seemed to learnfrom the experiences of those that fell before them.Management myopia or lack of foresight cannot ex-plain these failures. Tbe problem is that managerskeep doing what has worked in tbe past: serving therapidly growing needs of tbeir current customers.The processes tbat successful, well-managed com-panies bave developed to allocate resources amongproposed investments are incapable of funnelingresources into programs tbat current customers ex-plicitly don't want and wbose profit margins seemunattractive.

Managing tbe development of new technologyis tightly linked to a company's investment pro-cesses, Most strategic proposals-to add capacity orto develop new products or processes - take shapeat the lower levels of organizations in engineeringgroups or project teams. Companies then use ana-lytical planning and budgeting systems to selectfrom among the candidates competing for funds.Proposals to create new businesses in emergingmarkets art particularly challenging to assess be-

HARVARD BUSINESS REVIEW January-Fcbnury 1995 47

DISRUPTIVE TECHNOLOGIES

cause they depend on notoriously utireliablc esti-mates of market size. Beeause managers are evalu-ated on their ability to place the right bets, it is notsurprising that in well-managed companies, mid-and top-level managers back projects in which themarket seems assured. By staying close to lead cus-tomers, as they have been trained to do, managersfocus resources on fulfilling the requirements ofthose reliable customers that can be served prof-itably. Risk is reduced-and careers are safeguard-ed-by giving known customers what they want.

eagate Technology's experience illus-trates the consequences of relying onsuch resource-allocation processes toevaluate disruptive technologies. By al-

most any measure, Seagate, based in Scotts Valley,California, was one of the most successful and ag-gressively managed companies in the history of tbemicroelectronics industry: from its inception in1980, Seagate's revenues had grown to more thanS700 million by 1986. It had pioneered 5.25-inchhard-disk drives and was the main supplier of themto IBM and IBM-compatible personal-computermanufacturers. Tbe company was the leading man-ufacturer of 5.25-inch drives at the time the disrup-tive 3.5-inch drives emerged in the mid-1980s.

Engineers at Seagate were the second in the in-dustry to develop working prototypes of 3.5-incb

Seagate paid the price forallowing start-ups to leadway into emerging market

drives. By early 1985, they had made more than 80such models with a low level of company funding.The engineers forwarded the new models to keymarketing executives, and tbe trade press reportedtbat Seagate was actively developing 3.5-incbdrives. But Seagate's principal customers - IBMand other manufacturers of AT-class personalcomputers - showed no interest in the new drives.They wanted to incorporate 40-MB and 60-MBdrives in their next-generation models, and Sea-gate's early 3.5-incIi prototypes packed only 10 MB.In response, Seagate's marketing executives low-ered their sales forecasts for the new disk drives.

Manufacturing and financial executives at theeompany pointed out another drawback to tbe 3.5-inch drives. According to their analysis, the newdrives would never be competitive with rhe 5.25-

incb architecture on a cost-per-megabyte basis-animportant metric that Seagate's customers used toevaluate disk drives. Given Seagate's cost structure,margin.s on tbe bigher-capacity 5.25-incb modelstberefore promised to be much higher than tbose onthe smaller products.

Senior managers quite rationally decided that the3.5-inch drive would not provide the sales volumeand profit margins tbat Seagate needed from a newproduct. A former Seagate marketing executive re-called, "We needed a new model that could becometbe next ST412 |a 5.25-inch drive generating morethan S300 million in annual sales, which was near-ing tbe end of its life cycle|. At the time, tbe entiremarket for 3.S-inch drives was less than $50 mil-lion. Tbe 3.5-inch drive just didn't fit tbe hill - forsales or profits.'

The shelving of tbe 3.5-inch drive was not a sig-nal that Seagate was complacent about innovation.Seagate subsequently introduced new models of5.25-inch drives at an accelerated rate and, in so do-ing, introduced an impressive array of sustainingtechnological improvements, even though intro-ducing them rendered a significant portion of itsmanufacturing capaci ty obsolete.

Wbile Seagate's attention was glued to the per-sonal-computer market, former employees of Sea-gate and other 5,25-incb drive makers, who hadbecome frustrated by tbeir employers' delays in

launching 3.5-inch drives, founded anew company, Conner Peripherals.Conner focused on selling its 3.5-inch drives to companies in emerg-ing markets for portable computersand small-footprint desktop prod-ucts iPCs tbat take up a smalleramount of space on a desk). Conner'sprimary customer was Compaq

Computer, a customer that Seagate had neverserved. Seagate's own prosperity, coupled withConner's focus on customers who valued differentdisk-drive attributes [ruggedness, physical volume,and weight!, minimized the threat Seagate saw inConner and its 3.5-inch drives.

From its beachhead in the emerging market forportable computers, however, Conner improvedthe storage capacity of its drives by 50% per year.By tbe end oi 1987, 3.5-inch drives packed thecapacity demanded in tbe mainstream personal-computer market. At this point, Seagate executivestook their company's 3.5-inch drive off the shelf,introducing it to the market as a defensive responseto the attack of entrant companies like Conner andQuantum Corporation, the other pioneer of 3.5-inch drives. But it was too late.

48 HARVARD BUSINESS REVIEW laniuiy Fchruarv

By then, Seagate faced strongcompetition. For a while, thecompany was able to defend itsexisting market by selling ,?..S-inch drives to its established cus-tomer base - manufacturers andresellers of full-size personal c<,>m-putcrs. In fact, a large proportionof its 3.5-inch products continuedto he shipped in frames that en-abled its customers to mount thcdrives in computers designed toaccommodate 5.25-inch drives.But, in the end, Seagate could onlystruggle to become a second-tiersupplier in the new portable-com-puter market.

In contrast, Cormer and Quan-tum built a dominant position itithe new portable-computer mar-ket and then used their scale andexperience base in designing andmanufacturing 3.5-inch products to drive Seagatefrom the personal-computer market. In their 1994fiscal years, the combined revenues of Conner andQuantum exceeded $5 billion.

Seagate's poor timing typifies tbe responses ofmany establisbed companies to thc emergenceof disruptive technologies. Seagate was willing toenter thc market for 3.5-inch drives only when ithad become large enough to satisfy the company'sfinancial requircments-that is, only when existingcustomers wanted the new technology. Seagate hassurvived through its savvy acquisition of ControlData Corporation's disk-drive business in 1990.With CDC's technology base and Seagate's volume-manufacturing expertise, the company has becomea powerful player in the business of supplying large-capacity drives for high-end computers. Nonethe-less, Seagate has been reduced to a shadow of its for-mer self in the personal-computer market.

How to Assess Disruptive Technologies

PerFormance improvementrequired by moinslreoin market

Expected trajectory olperformance improvement

Current perbrmonce of potentially disruptive technology

Time

t should come as no surprise that fewcompanies, when confronted with dis-rtiptive technologies, bave been able toovercome thc handicaps of size or suc-

cess. But it can be done. There is a method to spot-ting and cultivating disruptive technologies.

Determine whether the technology is disruptiveot sustaining. Thc first step is to decide which ofthe myriad technologies on the horizon are dis-ruptive and, of tbose, which arc real threats. Mostcompanies have well-conceived processes for iden-tifying and tracking the progress oi potentially sus-taining technologies, because they are important to

serving and protecting current customers. But fewhave systematic processes in place to identify andtrack potentially disruptive technologies.

One approach to identifying disruptive technolo-gies is to examine internal disagreements over thedevelopment of new products or technologies. Wbosupports the project and who doesn't? Marketingand financial managers, because of their managerialand financial incentives, will rarely support a dis-ruptive technology. On thc other hand, technicalpersonnel with outstanding track records will oftenpersist in arguing that a new market for the tech-nology will emerge-even in the face of oppositionfrom key customers and marketing and financialstaff. Disagreement between the two groups oftensignals a disruptive technology that top-level man-agers should explore.

DeEine the strategic significance of the disruptivetechnology. The next step is to ask the right peoplethe right questions ahout the strategic importanceof the disruptive technology. Disruptive technolo-gies tend to stall early in strategic reviews becausemanagers either ask the wrong questions or ask thewrong people thc right questions. For example, es-tablished companies have regular procedures forasking mainstream customers-especially the im-portant accounts where new ideas are actuallytested-to assess the value of innovative products.Generally, these customers are selected becausethey arc the ones striving the hardest to stay aheadof their competitors in pushing the performance oftheir products. Hence these eustomers are mostlikely to demand the highest performance from ,

HARVARD BUSINESS REVIEW lanuary-Fchruaiy 1995 49

DISRUPTIVE TECHNOLOGIES

their suppliers. For this reason, lead customers arereliably accurate when it comes to assessing the po-tential of sustaining technologies, but they are reli-ably inaccurate when it comes to assessing the po-tential of disruptive technologies. They are thewrong people to ask.

A simple graph plotting product performance asit is defined in mainstream markets on the verti-cal axis and time on the horizontal axis can helpmanagers identify both the right questions and theright people to ask. First, draw a linedepicting the level of performanceand the trajectory of performanceimprovement that customers havehistorically enjoyed and are likely toexpect in the future. Then locate theestimated initial performance levelof the new technology. If the tech-nology is disruptive, tbe point willlie far below the performance demanded by currentcustoiners. (See the graph "How to Assess Disrup-tive Technologies.")

V What is the likely slope of performance improve-ment of the disruptive technology compared withthe slope of performance improvement demandedby existing markets? If knowledgeable technolo-gists believe the new technology might progressfaster than the market's demand for performanceimprovement, then that technology, which doesnot meet customers' needs today, may very welladdress them tomorrow. The new technology, there-fore, is strategically critical.

Instead of taking this approach, most managersask tbe wrong questions. They compare the antici-pated rate of performance improvement of the newtechnology with that of the established technology.If the new technology has the potential to surpassthe established one, the reasoning goes, they shouldget busy developing it.

Pretty simple. But this sort of comparison, whilevalid for sustaining technologies, misses the cen-tral strategic issue in assessing potentially disrup-tive technologies. Many of the disruptive technolo-gies we studied never surpassed the capability ofthe old technology. It is the trajectory of the disrup-tive technology compared with that of the maikeithat is significant. For example, the reason themainframe-computer market is shrinking is notthat personal computers outperform mainframesbut because personal computers networked with afile server meet the computing and data-storageneeds of many organizations effectively. Main-frame-computer makers are reeling not because theperformance of personal-computing technologysurpassed the performance of mainframe technolo-

gy but because it intersected with the performancedemanded by the established i3](jr̂ (?(.

Consider the graph again. If technologists believethat the new technology will progress at the samerate as the market's demand for performance im-provement, the disruptive technology may be slow-er to invade established markets. Recall that Sea-gate had targeted personal computing, wheredemand for hard-disk capacity per computer wasgrowing at 30% per year. Because the capacity of

Small, hungry organizationsare good at agilely changing

product and market strategies.

3.5-inch drives improved at a much faster rate, lead-ing 3.5-inch-drive makers were able to force Seagateout of the market. However, two other 5.25-inch-drive makers, Maxtor and Micropolis, had targetedthe engineering-works tat ion market, in which de-mand for hard-disk capacity was insatiable. In thatmarket, the trajectory of capacity demanded wasessentially parallel to the trajectory of capacity im-provement that technologists could supply in the3.5-inch architecture. As a result, entering the 3.5-inch-drive business was strategically less criticalfor those companies tban it was for Seagate.

Locate the initial market for the disruptive tech-nology. Once managers have determined that a newtechnology is disruptive and strategically critical,the next step is to locate the initial markets for thattechnology. Market research, the tool that man-agers have traditionally relied on, is seldom helpful:at the point a company needs to make a strate-gic commitment to a disruptive technology, noconcrete market exists. When Edwin Land askedPolaroid's market researchers to assess the poten-tial sales of his new camera, they concluded thatPolaroid would sell a mere 100,000 cameras overthe product's lifetime; few people they interviewedcould imagine the uses of instant photography.

Because disruptive teclinologics frequently sig-nal the emergence of new markets or market seg-ments, managers must create information aboutsuch markets - who the customers will be, whichdimensions of product performance will mattermost to which customers, what the right pricepoints will be. Managers can create this kind of in-formation only by experimenting rapidly, iterative-ly, and inexpensively with both the product andthe market.

50 HARVARD BUSINESS REVIEW lanuary-February

For established companies to undertake sucb experiments is very difficult. Tbe resource-allocationprocesses that are critical to profitahility and com-petitiveness will not - and should not direct re-sources to markets in which sales will be relativelysmall. How, tben, can an established companyprobe a market for a disruptive tecbnology? Letstart-ups-either ones the company funds or oth-ers with no connection to the company - conductthe experiments. Small, hungry organizations aregood at placing economical bets, rolling witb tbcpunches, and agilely changing prt)duct and marketstrategies in response to feedback fiom initial torays into the market.

Consider Apple Computer in its start-up days.The company s original product, rhe Apple I, was aflop when it was launched in 1977. But Apple badnot placed a huge bet on the product and bad gottenat least somcthin^i into the hands uf early usersquickly. The company leamed a lot from tbe Ap-ple I about tbt." new technology and about what cus-tomers wanted and did not want. )ust as important,a group of customers learned about what they didand did not want from personal computers. Armedwith this information, Apple launched tbe Apple IIquite successfully.

Many companies could bave learned tbe samevaluable lessons by watcbing Apple closely. In fact,stime companies pursue an explicit strategy of be-

F'-verv company that has triedId manage mainstream andiisruptive businesses within\ single organization failed.

ing ux'ond to invent -allowing small pioneers tolead the Vk̂ay into uncharted market territory. Forinstance, IBM let Apple, Commodore, and Tandydefine the personal computer. It then aggressivelyentered the market and built a considerable person-al-computer business.

But IBM's relative success in entering a new mar-ket late is the exception, not the rule. All too often,successful c(unpanies hold the performance ofsmall-market pioneers to tbc financial standardsthey apply to tbeir own performance. In an attemptto ensure that they are using their resources well,companies explicitly or implicitly set relativelyhigh thresholds for the size of the markets theyshould consider entering. This approach sentences

them to mnkmg late entries into markets alreadytilled with powerful players.

For example, when the 3.5-mch drive emerged,Seagate needed a S300-million-a-year product toreplace its mature flagship 5.25-inch model, theST412, and the 3.5-inch market wasn't largeenough. Over the next two years, when the tradepress asked wben Seagate would introduce its 3.5-inch drive, company executives consistently re-sponded that there was no market yet There actu-ally was a market, and it was growing rapidly. Thesignals tbat Seagate was picking up about the mar-ket, influenced as they were by customers wbodidn't want 3.5-inch drives, were misleading. WhenSeagate finally introduced its 3.5-inch drive in1987, more tban S750 million in 3.5-ineb drives badalready been sold. Information about the market'ssize had been widely available tbrougbout tbe in-dustry. But it wasn't compelling enough to shift tbefocus of Seagate's managers. They continued tolook at the new market through the eyes of theircurrent customers and in the context of their cur-rent financial structure.

The posture of today's leading disk-drive makerstoward tbe newest disruptive technology, 1.8-incbdrives, is eerily familiar. Eacb of the industry lead-ers bas designed one or more models of the tinydrives, and the models are sitting on shelves. Tbeircapacity is too low to be used in notebook comput-

ers, and no one yet knows where theinitial market for 1 .S-ineb drives willbe. Fax machines, printers, and auto-mobile dashboard mapping systemsare all candidates. 'Tbere iust isn'ta market," complained one industryexecutive. "We've got the product,and tbc sales force can take orders forit. But there are no orders becausenobody needs it. It just sits tbere."This executive has not considered

the fact that his sales force has no incentive to sellthe 1.8-inch drives instead of the higher-marginpnxlucts it sells to higher-volume customers. Andwhile tbc I .H-inch drive is sitting on the shell at hiscompany and others, last year more than S50 mil-lion worth of 1.8-inch drives were sold, almost allby start-ups. This year, the market will he an esti-mated S150 million.

To avoid allowing small, pioneering companiesto dominate new markets, executives must per-sonally monitor the available intelligence on theprogress of pioneering companies through monthlymeetings with technologists, academics, venturecapitalists, and other nontraditional sources of in-formation. They cannot rely on tbe company's tra-

HARVARI> BUSINESS REVIEW Mniury-Fcbruin'

DISRUPTIVE TECHNOLOGIES

ditional channels for gauging markets becausethose channels were not designed for that puqiose.

Place responsibility for building a disruptive-technology business in an independent organiza-tion. The strategy of forming small teams intoskunk-works projects to isolate them from the sti-fling demands ot mainstream organizations is wide-ly known but poorly understood. For example, iso-lating a team of engineers so that it can developa radically new sustaining technology just hecausethat technology is radically different is a fundamen-tal misapplication of the skunk-works approach.Managing out of context is also unnecessary in theunusual event that a disruptive tech-nology is more financially attractivethan existing products. Consider In-tel's transition from dynamic ran-dom access memory (DRAM) chipsto microprocessors. Intel's early mi-croprocessor business had a highergross margin than that of its DRAMbusiness; in other words, Intel's nor-mal resource-allocation process naturally providedthe new business with the resources it needed.

Creating a separate organization is necessary on-ly when the disruptive technology has a lower prof-it margin than the mainstream business and mustserve the unique needs tif a new set oi customers.GDC, for example, successfully created a remoteorganization to commercialize its 5.25-inch drive.Through 1980, CDC was the dominant mdepen-dent disk-drive supplier due to its expertise in mak-ing 14-inch drives for mainframe-computer makers.When the 8-inch drive emerged, CDC launched dlate development effort, but its engineers were re-peatedly pulled off the project to solve problems forthe more profitable, higher-priority 14-inch proj-ects targeted at the company's most important cus-tomers. As a result, CDC was three years late inlaunching its first 8-inch product and never cap-tured more than 5% of that market.

When the .S,2S-inch generation arrived, CDC de-cided that it would face the new challenge morestrategically. The company assigned a small groupof engineers and marketers in Oklahtima City,Oklahoma, far from the mainstream organization'scustomers, the task of developing and commercial-izing a competitive ri.2f>-inch product. "We neededto launch it in an environment in which everybodygot excited abt)ut a $50,000 order," one executiverecalled. "In Minneapolis, you needed a Si millit>norder to turn anyone's head." CDC never

36.

the 70% share it had once enjoyed in the market formainframe disk drives, but its Oklahoma City op-eration secured a profitable 20% of the high-perfor-mance 5.25-inch market.

Had Apple created a similar organiziition to de-velop its Newton personal digital assistant (PDA),those who have pronounced it a flop might havedeemed it a success. In launching the product, Ap-ple made the mistake of acting as if it were dealingwith an established market. Apple managers wentinto the PDA project assuming that it had to makea significant contribution to corporate growth. Ac-cordingly, they researched customer desires ex-

In order thai it mav live,a coiporation must be willing to

see business units die.

haustively and then bet huge sums launching theNewton. Had Apple made a more modest techno-logiciil and financial bet and entrusted the Newtonto an organization the size that Apple itself waswhen it launched the Apple I, the outcome mighthave been different. The Newton might have beenseen more broadly as a solid step forward in thequest to discover what customers really want. Infact, many more Newtons than Apple I modelswere sold within a year of their introducticm.

Keep the disruptive organization independent.Established companies can only dominate emerg-ing markets by creating small organizations oi thesort CDC created in Oklahoma City. But whatshould they do when the emerging market becomeslarge and established'

Most managers assume that once a spin off hasbecome commercially viable in a new market, itshould he integrated into the mainstream organi-zation. They reason that the fixed costs associatedwith engineering, manufacturing, sales, and distri-bution activities can be shared across a broadergroup of customers and products.

Tbis approach might work with sustaining tech-nologies; however, with disruptive technologies,folding the spin-off into the mainstream organiza-tion can be disastrous. When the independent andmainstream organizations are folded together in or-der !o share resources, debilitating arguments in-evitably arise over which groups get what resourcesand whether or when to cannibalize establishedproducts. In the history of the disk-drive industry,every company that has tried to manage main-

52 HARVARD BUSINESS REVIEW [anuary-Fchnury 1995

stream and disruptive businesses witbin a single or-ganization failed.

No matter tbe industry, a corporation consists ofbusiness units witb finite life spans: thc technolog-ical andmarket hases of any business will eventual-ly disappear. Disruptive tecbnologies are part oftbat cycle. Companies tbat understand tbis processcan create new businesses to replace tbe ones tbatmust inevitably die. To do so, companies must givemanagers ot disruptive innovation free rein to real-ize tbe tecbnology's full potential-even if it meansultimately killing tbe mainstream husiness. For thccorporation to live, it must be willing to see busi-ness units die. If tbe corporation doesn't kill themoff itself, competitors will.

Tbe key to prospering at points of disruptiveebange is not simply to take more risks, invest for

tbe long term, or figbt bureaucracy. Tbe key is tomanage strategically important disruptive tech-nologies in an organizational context wbere smallorders create energy, wbere fast low-cost forays intoill-defined markets are possible, and wbere over-bead is low cnougb to permit profit even in emerg-ing markets.

Managers of established companies can masterdisruptive tecbnologies witb extraordinary suc-cess. But wben they seek to develop and launcb adisruptive tecbnology tbat is rejected by importantcustomers witbin thc context of tbe mainstreambusiness's financial demands, tbey fail - not be-cause they make tbe wrong decisions, but becausetbey make tbe right decisions for circumstancestbat are about tqj)ecomejiistory. ^Reprint

THE. MIGHT OF

"Good evening, lady and gentleman.

CARTOON BY H MARTIN 53