J PROD INNOV MANAG 2008;25:347–369r 2008 Product Development & Management AssociationWhen Is a Disruptive Innovation Disruptive?ÃGlen M. Schmidt and Cheryl T. Druehl A disruptive innovation (i.e., one that dramatically disrupts the current market) is not necessarily a disruptive innovation (as Clayton Christensen deﬁnes this term). To aid in understanding why some innovations are more (or less) disruptive to the long-term health of incumbents, this article offers terminology and a framework complementary to Christensen’s work, focusing on the diffusion pattern of the new product. The framework and model presented herein suggest that when an innova- tion diffuses from the low end upward toward the high end, a pattern called low-end encroachment, the incumbent may be tempted to overlook its potential impact. Three possible types of low-end encroachment are illustrated: the fringe-market, detached-market, and immediate scenarios. Conversely, when the pattern is one of high-end encroachment, the impact on the current market is immediate and striking. A three-step framework is identiﬁed to assess the potential diffusion pattern and impact of an innovation, thereby helping a ﬁrm determine the threat or opportunity that an innovation represents.Introduction with regard to the primary performance dimension most appreciated by mainstream customers of the oldF ew terms in the recent literature on innova- product. However, the new product may perform bet- tion management have been as widely used as ter on an alternate dimension and thus open up a new the phrase disruptive innovation, as coined by market (or may simply be easier to use or of lowerClayton Christensen in his seminal and path-breaking cost). Then over time the disruptive innovation im-works, The Innovator’s Dilemma (Christensen, 1997), proves on the primary dimension to the extent that itThe Innovators Solution (Christensen and Raynor, eventually appeals to the very mainstream customers2003), and Seeing What’s Next (Christensen, Antho- that initially shunned it.ny, and Roth, 2004) (see Graziano, 1998 for a review Christensen and Raynor (2003) and Christensen etof Christensen, 1997 and Deck, 2005 for reviews of al. (2004) contend that incumbent ﬁrms often fail toChristensen and Raynor, 2003 and Christensen et al., recognize the threat posed by a disruptive innovation.2004). In these works, the essence of a disruptive in- That is, when incumbents are ‘‘overthrown,’’ it is gen-novation is described as follows. The new product (the erally by disruptive innovation. Thus, it is criticallydisruptive innovation) is de-rated (it underperforms) important that managers be able to recognize a dis- ruptive innovation when they see one. These works go Address correspondence to: Glen M. Schmidt, David Eccles School on to suggest that to succeed with a disruptive inno-of Business, University of Utah, Salt Lake City, UT 84112. Tel.: (801) vation, an incumbent should pursue it in a separate585-3160. E-mail: firstname.lastname@example.org. Ã The contributions of previous coauthors Evan Porteus and Jan business unit. Again, this points to the need forVan Mieghem were vital in helping develop this line of work. Journal clear recognition—a ﬁrm must be able to clearly de-editors Vish Krishnan, Christoph Loch, and Karl Ulrich providedvaluable feedback on related articles as did anonymous reviewers and lineate between what is a disruptive innovation andparticipants at the University of Utah’s annual conference on product what Christensen and Raynor (2003) and Christensenand service innovation, including Bill Moore, Kamalini Ramdas, andBo van der Rhee. The reviews of Professor Anthony Di Benedetto and et al. (2004) deﬁne as its converse: a sustaining inno-the anonymous referees led to signiﬁcant improvements in this article. vation.
348 J PROD INNOV MANAG G.M. SCHMIDT AND C.T. DRUEHL 2008;25:347–369Table 1. Mapping of the Type of Innovation to the Type of Diffusion Type of DiffusionType of Innovation to which It Maps Description ExampleSustaining Innovation High-end The new product ﬁrst encroaches on the high Pentium IV relative to encroachment end of the existing market and then diffuses Pentium III downward.Disruptive Innovation Low-end The new product ﬁrst encroaches on the low encroachment end of the existing market and then diffuses upward.New-Market Disruption Fringe-market low- Before encroachment begins, the new product 5.25 inch disk drive relative to end encroachment opens up a fringe market (where customer 8 inch drive needs are incrementally differenta from those of current low-end customers). Detached-market Before encroachment begins, the new product Cell phone relative to land low-end opens up a detached market (where customer line encroachment needs are dramatically differenta from those of current low-end customers).Low-End Disruption Immediate low-end Low-end encroachment begins immediately Discount relative to encroachment upon introduction of the new product. department storesÃ The distinctions between fringe and detached markets and between incrementally and dramatically different preferences are illustrated in the disk-drive examples provided herein. In spite of the ubiquity of the term, managers often innovation’s diffusion process is actually less disrup-have a hard time identifying a disruptive innovation. tive initially to an incumbent than that of a sustainingThis is true even if they have some familiarity with innovation. Said loosely, a disruptive innovation (inChristensen’s work, such as having been assigned to that it disrupts the current market) is not necessarily aread Bower and Christensen (1995), listed by Harvard disruptive innovation (as Christensen deﬁnes it).Business School Publishing as one of the most popu- In preparation for formalizing the terminologylar articles for executive education. Scott Anthony, a found in Table 1, Figure 1 illustrates two distinctlypartner at Innosight (the consulting ﬁrm founded by different diffusion patterns. First, consider the impactChristensen), also implies the term is often misunder- of a new generation of Pentium processor, as illus-stood: ‘‘The word disruption . . . has become loaded trated in the left frame of Figure 1. When the Pentiumwith meanings and connotations at odds with the III (P-3) was introduced in the last quarter of 1998,concept’’ (Anthony, 2005a, p. 3, italics in original). sales of the Pentium II (P-2) began to drop andFurther emphasizing the need for additional research dropped precipitously as the sales of the P-3 gainedwith regard to disruptive innovation are Danneels ground. Likewise, sales of the P-3 peaked at the point(2004) and Christensen (2006), for example. the Pentium IV (P-4) was introduced and dropped Because the term disruptive can be so easily mis- dramatically as the sales of the P-4 ramped up. This isconstrued, this article offers alternate terminology a familiar pattern: A new faster generation of micro-and a complementary framework. This terminology processor quickly cannibalizes the old, starting at theis based on the ﬁnding that new products diffuse high end of the market and diffusing downward to thethrough the market in different ways: A disruptive low end. The new generation has a disruptive impact on sales of the previous generation. Yet, in spite of this rapid cannibalization, Christensen, Anthony, and BIOGRAPHICAL SKETCHES Roth (2004) does not refer to a new generation of Dr. Glen Schmidt is associate professor in the David Eccles School of Business at the University of Utah. His Ph.D. is from the Grad- microprocessor as a disruptive innovation. Rather, it uate School of Business at Stanford University in the area of op- is a sustaining innovation. erations, information, and technology. Prior to his academic career It is important to recognize that even though the P- he worked in industry for 15 years. 2, P-3, and P-4 were all introduced by the same ﬁrm Dr. Cheryl Druehl is assistant professor in the School of Manage- (Intel), this is not the reason that a newer faster mi- ment at George Mason University. Her Ph.D. is from the Graduate School of Business at Stanford University, and her industry expe- croprocessor is sustaining rather than disruptive. rience includes engineering and consulting. Even if the P-4 would have been introduced by a competitor to Intel such as AMD, it would still be a
WHEN IS A DISRUPTIVE INNOVATION DISRUPTIVE? J PROD INNOV MANAG 349 2008;25:347–369 Units sold per quarter, millions Units sold per year , millions 25 12 (in 100,000 for the 8-inch drive) 20 ” 15 8 P-4 P-3 ch h in inc 10 5 P- P-2 2 5. 3. 5 5 4 8 inch 0 1st Qtr. 1st Qtr. 1st Qtr. 1st Qtr. 1st Qtr. 0 1998 1999 2000 2001 2002 1980 1982 1984 1986 1988 1Figure 1. Sales of Successive Generations of Microprocessors and Disk Drives (Microprocessor data are from Dataquest, Inc., 2003,and disk-drive data are from Christensen, 1992)sustaining innovation. (Recall the deﬁnition of a dis- pared with the way the P-3 disrupted the P-2, the 5.25ruptive innovation given at the outset: The classiﬁca- inch drive wasn’t initially very disruptive to the pre-tion itself has to do with the characteristics of the vious 8 inch generation. Similarly, sales of the 5.25innovation and nothing to do with which ﬁrm intro- inch drive continued to ramp even as sales of the new-duces it.) Indeed, a key point of Christensen’s work is er 3.5 inch drive accelerated. Again, compared withto get incumbents to introduce disruptive innovations, the way the P-4 disrupted the P-3, the 3.5 inch driveinstead of focusing primarily on the sustaining type. wasn’t at the outset very disruptive to the 5.25 inchAlthough historically it may be the case that sustain- generation. Yet the disk-drive example is Christen-ing innovations have more often been associated with sen’s (1997) classic example of a disruptive innovation.incumbents and disruptive innovations with entrants, Thus, the microprocessor example represents anthey are not linked this way by deﬁnition. The Intel innovation that is disruptive to sales of the old prod-example is used only because it is so well known and is uct right from the outset but is not a disruptive inno-unarguably a sustaining innovation. For an example vation (per Christensen’s (1997) deﬁnition, asof a sustaining innovation introduced by an entrant, discussed on p. 347), whereas the disk-drive exampleconsider Apple’s ﬁrst iPod, introduced in late 2001, represents an innovation that is not very disruptiveseveral years after Rio and Creative Technologies in- initially but is in fact a disruptive innovation, con-troduced more primitive players. The iPod was an ex- tributing to the confusion surrounding the term. Apensive $399 high-end product, giving customers reason that it took longer for the new 5.25 inch drivemore of the key performance attribute that they to encroach on the old 8 inch drive market is that itcraved—memory—via the innovation of a tiny disk ﬁrst opened up a new low-end market for desktopdrive instead of ﬂash memory (in addition to offering computers before diffusing up-market to higher-endsuperior industrial design). Thus, the iPod ﬁts the de- products such as midrange and mainframe computers.scription of a sustaining innovation yet was intro- It is inferred from Christensen’s work that it is induced by an entrant that subsequently encroached on part because of the nondisruptive nature of a disrup-competitive products to grow its market share to tive innovation in the short run that an incumbent80%. For an example of a disruptive innovation in- may fail to take action and may therefore end up be-troduced by an incumbent, consider Intel’s Celeron. ing disrupted catastrophically in the long run. His Contrast the microprocessor experience with the work is groundbreaking and seminal—an incrementaldata shown in the right frame of Figure 1 for succes- contribution of the framework presented herein is thatsive generations of computer disk drives, which are it offers the alternate terminology of low-end en-hard drives contained inside computers where infor- croachment and high-end encroachment and outlinesmation is stored. Subsequent generations of disk the steps a ﬁrm can take to determine the potentialdrives were physically smaller and smaller: the 8 market impact of a new innovation.inch drive was superseded by the 5.25 inch drive, The term encroachment denotes that the new prod-which in turn was followed by the 3.5 inch drive. Note uct takes sales away from the old product. Cannibal-how sales of the 8 inch drive continued to grow even ization is a special form of encroachment where bothas the sales of the 5.25 inch drive ramped up. Com- products are sold by the same ﬁrm. Low-end
350 J PROD INNOV MANAG G.M. SCHMIDT AND C.T. DRUEHL 2008;25:347–369encroachment describes the scenario where the new Mapping Type of Innovation to High-Endproduct ﬁrst displaces the old product in the low end Encroachment versus Low-End Encroachmentof the old product market and then diffuses upward(the new product may open up a new market before Table 1 provides a mapping between the terminologyencroachment begins). The low end of a product’s of the encroachment framework and the theory ofmarket is deﬁned to consist of those customers with disruptive innovation. As described earlier, the newlowest willingness to pay for the product (they have microprocessor generation (a sustaining innovation)the lowest demand for the product’s key performance and the smaller disk drive (a disruptive innovation)attributes). Similarly, the high end of the market is differed in the way they diffused through the market.composed of customers with the highest willingness to The microprocessor is an example where the ﬁrst buy-pay. As suggested by the previous disk-drive example, ers are high-end customers: They have high willing-disruptive innovation maps to low-end encroachment. ness to pay, and their appetite for faster processingHigh-end encroachment progresses in reverse fashion, speed and more computing power is nearly insatiable.starting at the high end of the old-product market. A The laggards in adopting the new microprocessor aresustaining innovation diffuses via high-end encroach- lower-end customers who are less demanding (whosement, as indicated by the previous microprocessor ex- willingness to pay is lower). Thus, a new generation ofample. microprocessor diffuses via a pattern of high-end en- While in the previous examples encroachment pro- croachment: The new product ﬁrst displaces the oldgressed to the point where the new product fully sub- product at the high end, followed later by diffusionstituted for the old, this may not always be the case down toward the low end. The low-end customers areunder either form of encroachment. Encroachment the last to adopt the new product. A north wind is soimplies that the new product has some impact on the named because it blows from the north to the south;old but does not deﬁne the extent of that impact; sim- similarly, high-end encroachment progresses from theilarly, a disruptive innovation ultimately has at least high end of the old market toward the low end.some impact on an existing market but need not nec- In contrast, in the case of disk drives, the high-endessarily totally displace the market. customers (i.e., mainframe users) were the laggards In the current article three types of low-end en- rather than the ﬁrst to buy. The new drive had muchcroachment are deﬁned and related to Christensen less storage capacity, and thus these high-end custom-and Raynor’s (2003) ﬁndings. The encroachment ers, having an insatiable appetite for capacity, did notframework presented herein helps managers relative- initially even consider the new smaller drive. Instead,ly quickly grasp the impact of a potential new product a new market segment began purchasing the drives—and thus helps them more readily recognize whether when the 5.25 inch drive began displacing the 8 inchan innovation is disruptive or not. A key contribution drive in the mid 1980s, the new market segment wasof this article is to cohesively map the theory of dis- the desktop computer. Only as the new smaller drive’sruptive innovation to the encroachment framework. capacity was upgraded over time did it begin to en- The encroachment framework is based on an eco- croach on (i.e., to displace) the old larger drive. Thisnomic model referred to as a linear reservation price encroachment ﬁrst occurred at the low end of the oldmodel. This model is formally developed in Schmidt drive’s market (for the 8 inch drive in the mid 1980s,and Porteus (2000), Schmidt and Druehl (2005), and this was the mid-range computer market), where cus-Druehl and Schmidt (2006): A key contribution of this tomers were more price sensitive and were less drivenarticle is to use basic elements of the linear reservation by a need for capacity as compared with the high-endprice model to develop a three-step process that a ﬁrm market (i.e., mainframes). Accordingly, this pattern ofcan follow to assess whether a potential innovation diffusion is called low-end encroachment: The newwill diffuse via low-end or high-end encroachment. product encroaches on the old product from the low Another key contribution of this article is to cri- end upward toward the high end, with the low-endtique the encroachment framework against Christen- customers the ﬁrst to switch and the high-end cus-sen and Raynor (2003) and Christensen et al.’s (2004) tomers being the last to switch (if ever).deﬁnitions and to further validate the framework by The terminology of high-end and low-end en-considering more than 70 innovations as classiﬁed by croachment was ﬁrst presented in Schmidt and Porte-Christensen and Raynor (2003). Discussion and man- us (2000). Subsequently, Christensen and Raynoragerial recommendations are also offered. (2003) divided disruptive technologies into two types:
WHEN IS A DISRUPTIVE INNOVATION DISRUPTIVE? J PROD INNOV MANAG 351 2008;25:347–369new-market disruptions and low-end disruptions. The business ofﬁce segment. At the low end of the spec-disk-drive example is a new-market disruption, be- trum were second lines in homes (e.g., ‘‘teen lines’’).cause the new smaller generation of drive ﬁrst attract- The cell phone started out as a very expensive producted buyers in a new fringe-market segment as opposed that initially sold to mobile business users such asto immediately selling to customers previously asso- building contractors, who had dramatically differentciated with the existing (older) generation. For exam- willingness to pay (rather than incrementally differentple, as mentioned, the 5.25 inch generation of disk preferences) as compared with the low-end fringe ofdrives opened up the desktop computer market seg- the current market: In this sense the early new marketment. But when the new product did eventually begin was ‘‘detached’’ from the low end of the current one.stealing the older generation’s customers it did so However, the ﬁrst major group of users to drop thefrom the low end upward. Thus a new-market dis- old product (the land line) in favor of the innovationruption results in low-end encroachment. (the cell phone) was the land line’s low-end market: With a low-end disruption (as Christensen and Almost every ofﬁce still has a land line, and mostRaynor, 2003 are understood to deﬁne it), there may homes still have their primary line—but many stu-be little or no market expansion: The ﬁrst sales of the dents and apartment dwellers now exclusively use thenew product are to customers who would have oth- cell phone.erwise purchased the old product, as opposed to buy- The detached-market version of low-end encroach-ers in a new market segment. The low-end disruption ment helps explain a conundrum generated by Chris-also encroaches from the low end upward, ﬁrst selling tensen and Raynor (2003) and Christensen et al.to price-sensitive low-end customers. In other words, (2004). While they continually emphasize that disrup-both new-market and low-end disruptions result in a tive innovations are low priced, they proceed to listdiffusion process called low-end encroachment, the exceptions, such as the cell phone. The current articleonly difference being whether the encroachment starts shows how a product can encroach from the low endimmediately (as in the case of a low-end disruption) or even if it starts out as expensive. Digital cameras areafter the new product has opened up a new market another exception.and subsequently improved enough to become attrac- As Christensen and Raynor (2003) point out and astive to the low-end customers of the older product (as the framework herein will support, when an entrantin the case of a new-market disruption). ﬁrm introduces a new product that has the potential To distinguish between these scenarios, low-end to encroach from the low end, the incumbent ﬁrmencroachment is further broken down into three types may have reason to view the new product as non-(Table 1). When the innovation immediately sells to threatening. Indeed, as seen in the case of the smallerthe existing low-end market it is called immediate low- disk drive, the new product initially takes no salesend encroachment. If it opens up a new market, one away from the old product but rather may sell to apossibility is that the new market is on the fringe of new market segment. Even if some sales are impacted,the old market, in which case the diffusion pattern is the new product sells to low-end customers who areone of fringe-market low-end encroachment. A new not that highly valued anyway, since they have suchmarket is deﬁned to be on the fringe of the old market low willingness to pay. When Toyota ﬁrst encroachedif buyers in this new market would have bought the on the U.S. car market, it did so with economy carscurrent (old) product if only the old product were a that garnered low margins. General Motors and otherlittle less expensive. In other words, the preferences of U.S. manufacturers readily gave up this low-end mar-the new fringe market are only incrementally different ket because the bulk of their proﬁt was made on high-from those on the low end of the current market. er-end cars. Of course, over time, Toyota encroached Alternately, the new market may be ‘‘detached’’ upward into more lucrative midsize cars, eventuallyfrom the old market: Preferences in this new market producing the best-selling Camry, and ultimately en-are so divergent (detached) from the current market croached upward into the luxury car market with thethat reducing the price of the current product a bit Lexus.would not have enticed the detached market to buy it. If the entrant introduces a new product that en-For example, consider the innovation of the cell croaches from the high end, the incumbent tends tophone, classiﬁed by Christensen and Raynor (2003) defend its market quickly and vigorously, because inand Christensen et al. (2004) as disruptive. The high- the model presented herein, the incumbent is losing itsend market for land lines (the old product) was the best customers (those with highest willingness to pay).
352 J PROD INNOV MANAG G.M. SCHMIDT AND C.T. DRUEHL 2008;25:347–369Its potential to reap high margins is dissipated. Thus, new. And with continual upgrading, the new productthe incumbent may be more likely to introduce its eventually becomes acceptable even to the high-endown new product to encroach from the high end, even customers of the old product, who then also switchif it means cannibalizing its own product, as Intel does from buying the old product to buying the new. Notewith a new generation of microprocessor. The point is that this is a description of a low-end encroachmentnot that high-end encroachment is necessarily a bad process: The new product diffuses through thestrategy and low-end encroachment a good strategy, market from the low end upward toward the highor vice versa. Rather, the point is that both incum- end (possibly after ﬁrst selling only to a new marketbents and entrants must be aware of and make use of segment).the strategy (or strategies) that offer maximum ben- This conﬁrms the basic mapping of a disruptiveeﬁt. innovation to low-end encroachment. If the new prod- uct opens up a new-market segment, these new customers may have only incrementally different pref-Is the Encroachment Model Consistent with erences as compared with the old product’s low-endthe Theory of Disruptive Innovation? market (this new market is called a fringe market) or can have diverse preferences (this is referred to as aDruehl and Schmidt (2008) formally establish the detached market). In some cases the new product maymapping shown in Table 1. To describe brieﬂy why sell immediately to existing low-end customers with-the Christensen and Raynor (2003) and Christensen et out opening up a new segment (this corresponds to theal. (2004) terminology maps to ours as shown in Table ‘‘immediate’’ type of low-end encroachment).1, one can ‘‘go back to basics.’’ Christensen’s theory is In other words, by going back to the source ofbased on a ‘‘trajectory chart’’ developed from the Christensen’s theory (the trajectory curve), the map-disk-drive industry. Such trajectory charts are given in ping between his terminology and ours becomes ap-Christensen (1997, pp. xvi, 16), Christensen and Ray- parent. The encroachment terminology offers annor 2003, pp. 33, 44), and Christensen et al. (2004, p. alternate means of describing the impact of an inno-xvi). Their basic premise is that in trying to please vation, and the framework presented herein offershigh-end customers with regard to a key performance insights beyond those illustrated by the trajectorydimension, an incumbent eventually develops a prod- chart. Speciﬁcally, the three-step process presenteduct that ‘‘overshoots’’ the performance needs of mid in this article is a way to qualitatively or quantitativelyto low-end customers along that key dimension. Then assess a potential new product’s impact over time, inalong comes a new product (a disruptive innovation) terms of prices, quantities, proﬁts, and market seg-that falls short of the needs of most (if not all) current ments. This article further shows that characteristicscustomers along this ﬁrst performance dimension but of a new product determine whether it will encroachthat is lower cost or performs better along a second from the high or low end, and if it is the latter, wheth-dimension. While existing high-end customers dislike er the product is of the fringe-market, detached-mar-the new product (they despise its poor performance ket, or immediate low-end encroachment variety.along the ﬁrst dimension), a new market segment (orthe existing low-end segment) gladly accepts the de-rated performance along the ﬁrst dimension in favor A Framework to Help Recognize a Low-Endof lower cost or the enhanced performance along the Encroachment Threat (or Opportunity)second dimension. In the disk-drive example of the1980s, this was the desktop computer market segment. As discussed previously, the terminology and insights Over time, however, the new competitive product is presented herein stem from a linear reservation pricecontinually and incrementally upgraded, particularly model. This section of the article goes on to offerwith regard to that ﬁrst performance dimension where insight into how a ﬁrm can begin to assess how a newit was initially woefully inferior to the old product. product might impact the market. The linear reserva-Gradually, because of this continual upgrading, the tion price model further provides a mechanism to as-new product eventually becomes acceptable to the sess the seriousness of a low-end encroachment threatlow-end customers of the old product (assuming it and possibly to turn such a threat into an opportuni-ﬁrst sold only to a new market segment), who then ty. Of course, high-end encroachment offers its ownswitch from buying the old product in favor of the opportunities and threats.
WHEN IS A DISRUPTIVE INNOVATION DISRUPTIVE? J PROD INNOV MANAG 353 2008;25:347–369 The framework is developed assuming there is an would likely accept an even deeper trade-off with re-incumbent ﬁrm (or set of ﬁrms) selling an existing, gard to capacity. And even further detached from thewell-established product. Analysis is pursued from the current market was the possibility of handheld dataperspective of the incumbent, but an entrant ﬁrm loggers. Memory requirements would be even morecould use a similar process. The disk-drive example minimal for this device compared with a laptop oris used here because it is a classic example familiar to a desktop computer. But the requirement for compact-large audience. Also, by using this one example all the ness would be even greater. Of course, today thesedifferent types of encroachment can be illustrated types of devices are known as, for example, personal(high end plus the three types of low end): The sce- digital assistants (PDAs) and MP3 players.nario that actually materializes depends on the char-acteristics of the product actually introduced into the Step 2: Assess Each Market Segment’s Willingness tomarket. Pay for Each Attribute The next step is to plot each of the current and po- tential market segment’s willingness to pay for each ofStep 1: Identify Market Segments and Primary Attri- the attributes identiﬁed to this point, both the key at-butes of the Product tributes of the current product and the alternative at-The ﬁrst step is to formally identify the various mar- tributes preferred by fringe markets. These plots areket segments that currently use the incumbent’s cur- given in Figures 2 and 3 for the two key disk-driverent product and to put them in order from high end attributes: capacity and compactness (physical size).to low end. In 1985 when the 8 inch disk drive was the The plots qualitatively reﬂect the relative preferencesmainstream product, the key market segments were of the mainframe, midrange, desktop, and laptop seg-mainframe and midrange computers (workstations). ments as given by Christensen (1992), but the exactThe key performance attribute for disk drives was ca- shapes of these plots (as will be further described la-pacity (MB). Mainframe users represented the high- ter) are hypothetical. The relative position of the spe-end segment, demanding the highest capacity and cialty segment is inferred from market conditions.offering the highest willingness to pay for this key Thus, the plots are based only in part on actual data:parameter. A second performance parameter, that of They are hypothetical, but plausible.physical size of the disk drive (i.e., compactness), was Note in Figure 2 that the market segments are or-not a real consideration for the mainframe users be- dered along the x-axis in terms of customer willing-cause these computers were housed in a separate large ness to pay, from highest to lowest. The width of eachroom or even a basement where it made virtually no segment indicates the number of potential users indifference whether the drive was 8 inches or a more that segment. That is, each point along the x-axiscompact 5.25 inches or a very compact 3.5 inches. effectively represents one customer—realistically the But the analysis must not stop with current users. It x-axis should show discrete points, but for mathemat-must also consider new market segments that mightmaterialize if the product were de-rated to reduce Higprice or if alternate features were improved. In the her capdisk-drive market of the mid 1980s, desktop comput- Cu ac ity rreers were on the drawing board but not yet widely nt c Low a pacmarketed. An essential component that facilitated the er capa ity Willingness citydevelopment of the personal computer (PC) was a to pay forsmaller disk drive, one compact enough such that a capacity Lowest ca paci tydesktop computer could actually ﬁt on a desktop.Also, this market required a disk drive that was cheapenough such that the ﬁnished product would be Main- Mid- Desktop Laptop Specialtyaffordable by home users. Early users were willing frame rangeto accept a de-rated disk drive (with much less capac- Current Applications Potential Applicationsity) if it instead met the size and price requirements. Even further on the fringe was the possibility of a Total number of potential customersnotebook computer. Such computers would need Figure 2. Willingness to Pay for Capacity as a Function of Mar-drives that were even more compact, but buyers ket Segment
354 J PROD INNOV MANAG G.M. SCHMIDT AND C.T. DRUEHL 2008;25:347–369ical convenience it is shown as a continuous line. The new drive of smallest size and lowest capacity; and (4)large ‘‘dots’’ in each graph denote the average will- a drive of the existing size and capacity, but withoutingness to pay within each market segment: The line sales support.through each set of dots suggests that even withineach market segment there is some variation in will- Product 1: A new drive of current size but with a thin-ingness to pay. Further, the plots suggest that the ﬁlm head that facilitates higher capacity (high-end en-market segments ‘‘come together’’ at the boundaries croachment). First, consider the case where the newbetween segments; for example, the mainframe cus- disk drive is of current size but with some new tech-tomer with the lowest willingness to pay for capacity nology, such as a thin-ﬁlm head, that facilitates pack-in Figure 2 has roughly the same willingness to pay as ing more capacity into the drive. An analogousthe midrange customer with the highest value. This is, example would be that of Intel upgrading the Pent-of course, an approximation, and such an assumption ium processor from say, the P-3 to the P-4; the newmay not precisely hold (i.e., the curves may not always processor offers more of what the high-end customersbe linear as shown here). want—processing power. A customer’s total willing- The desktop computer market is said to represent ness to pay for the new drive (called the customer’sthe fringe segment because preferences in this market reservation price for the new drive) is calculated as theare closest to the preferences of the current (1985) customer’s willingness to pay for the new drive’s ca-low-end market (midrange) segment. The specialty pacity plus his or her willingness to pay for its size. Insegment is ‘‘detached’’ (preferences are quite disparate other words, since the new drive is of higher capacityfrom the current low-end segment). and current physical size, the plot of all customers’ Note from Figures 2 and 3 that that the stronger the reservation prices is obtained by adding the line forcustomer’s preference for capacity, the weaker his or ‘‘higher capacity’’ in Figure 2 to the line for ‘‘currenther preference for compactness, and vice versa. That physical size’’ in Figure 3. This results in the line la-is, the strength of preference for capacity is negatively beled ‘‘new’’ (for new product) in the left frame ofcorrelated with the strength of preference for com- Figure 4. This line represents the new product’s res-pactness, as shown by Christensen (1992). These im- ervation price curve. Analogously, the line for ‘‘currentplications are discussed in a later section. capacity’’ in Figure 2 is added to the line for ‘‘current physical size’’ in Figure 3 to ﬁnd the reservation price size curve for the current (old) product. Figure 4 no longer ical p hys plots the ‘‘dots’’ or averages for each market segment llest Sma but simply plots the two resulting reservation priceWillingness curves, one for the current (old) product and one for to pay for ical size Smaller phys the new product.compactness For purposes of gaining insight, it is assumed that Current physical size part-worth curves are always linear (thus, no matter Main- Mid- Desktop Laptop Specialty how many attributes the reservation price curves will frame range also be linear), and it is assumed there are exactly two products in the market: the current (old) product and Current Applications Potential Applications the innovation (the new product). While real-life Total number of potential customers problems may often be more complex, possibly in-Figure 3. Willingness to Pay for Compactness as a Function of cluding, for example, more than two key performanceMarket Segment attributes, the objective here is to broadly determine what inﬂuences diffusion patterns rather than to gen- erate precise numbers. Using an example of bicycleStep 3: Assess which Segments Will Buy a Given New pumps, for which there are four attributes—inﬂationProduct over Time time, ease of use, size, and durability—Schmidt andFour innovations that could be introduced into the Porteus (2000) show data that suggest the assumptionmarket just described are considered: (1) a new drive of linearity is an approximation but is not necessarilyequal in size to the current drive but with higher ca- an unreasonable one.pacity resulting from new thin-ﬁlm head technology; There are two possible scenarios: Either both res-(2) a new drive of smaller size and lower capacity; (3) a ervation price curves slope downward (if both slope
WHEN IS A DISRUPTIVE INNOVATION DISRUPTIVE? J PROD INNOV MANAG 355 2008;25:347–369 At introduction After several years After a few more years Ne Ne w Ne w Reservation price w Reservation price Reservation price Ol d Ol d Ol d Sales of old Sales of new Sales Sales Sales Sales of of of of old new old new Main- Mid- Desktop Laptop Specialty Main- Mid- Desktop Laptop Specialty Main- Mid- Desktop Laptop Specialty frame range frame range frame range Actual or Potential Market Segments Actual or Potential Market Segments Actual or Potential Market Segments (width of segment indicates number of customers) (width of segment indicates number of customers) (width of segment indicates number of customers)Figure 4. Product 1 First Sells to High-End Customers and Then Encroaches Downward, Representing High-End Encroachmentupward, reorder the x-axis so they slope downward), Note in the left frame of Figure 4 that mainframeor they are opposite sloping (one downward, one up- customers have the strongest preference for the newward). As Figure 4 veriﬁes, new product 1 yields the drive as compared with the current drive (the differ-downward-sloping case. ence between the two reservation price curves is great- If the production costs of the old and new products est for mainframe customers). The new drive’sare also known (in addition to the reservation price reservation price curve has a steeper slope than thatcurves), then each product’s selling price can be of the old drive, and per Characteristic 1, mainframefound, which in turn determines each product’s sales customers will be the ﬁrst to buy it.quantity and its proﬁt. If the two products are sold by For purposes of illustration, it is assumed that allcompetitors, then prices are found using the concept of the hypothetical new products discussed herein areof a Nash equilibrium, whereas if sold by the same introduced by a competitor and that each new prod-ﬁrm it is assumed that the ﬁrm maximizes proﬁts. uct eventually fully displaces the old product. But vir-Each customer is assumed to buy either the old prod- tually all successful innovations take time to realizeuct or the new product or nothing—a customer buys their success—when the new product is ﬁrst intro-the product that offers him or her the most surplus, duced, it starts with minimal sales, and over time itswhere surplus is the difference between the product’s sales grow as it diffuses through the market. Theactual selling price minus the customer’s reservation product under current consideration is no exception—price—if each product’s selling price is greater than the leftmost frame of Figure 4 illustrates that when thethe customer’s reservation price, then that customer new disk drive of current size but with higher capacitybuys nothing. For further technical details, refer to is ﬁrst introduced, it achieves minimal sales that, perSchmidt and Porteus (2000) for the downward-slop- Characteristic 1, are derived from the high-end mar-ing case, and refer to Druehl and Schmidt (2008) for ket. The width of the darkened rectangle labeledthe opposite-sloping case. ‘‘Sales of new’’ indicates the number of units sold The linear reservation price framework just de- for that drive, whereas the height of the rectangle in-scribed exhibits several notable characteristics. Char- dicates selling price, such that the area of the rectangleacteristic 1 follows directly from Theorem 1 of equals sales revenues. A similar interpretation appliesSchmidt and Porteus (2000). to the rectangle labeled ‘‘Sales of old.’’ Note that the rectangle representing the sales of the old product in- Characteristic 1 (applies to the downward-sloping sce- tersects the reservation price curve for the old product nario): The product whose reservation price curve has a at the upper right corner of the rectangle. No cus- shallower slope sells to the low end of the market, and tomers to the right of that point will buy the old the product with the steeper slope sells to the high end, product because their willingness to pay is less than assuming both generate positive sales. the product’s selling price. At the boundary between
356 J PROD INNOV MANAG G.M. SCHMIDT AND C.T. DRUEHL 2008;25:347–369the two rectangles (the one representing the sales of The ﬁrm selling the old product reacts to this trendthe old product and the one representing the sales of in reservation price curves and costs, as does the ﬁrmthe new) the surpluses for the two products are equal. selling the new product. Speciﬁcally, it is assumed thatTo the left of that boundary, surpluses are higher for each ﬁrm reacts by resetting its price to maximize itsthe new product, so customers buy the new. To the proﬁt, given the other ﬁrm’s price. This is the Nashright of that boundary, surpluses are higher for the equilibrium. Of course, when a ﬁrm resets its price,old product. The prices of the two products determine this impacts the volume sold, which the ﬁrm takes intothe exact position of that boundary. account in setting price. Christensen and Raynor (2003) suggests that diffu- Effectively, the described trends in reservationsion and substitution of the new product for the old price curves and costs push more buyers toward themay in some cases be primarily due to improvements new product, as shown qualitatively in the progres-over time in the new product’s attributes and costs sion from the left frame to the middle frame and(Schmidt and Druehl, 2005). Thus, for a ﬁrm to pro- eventually to the right frame in Figure 4—the out-ject how encroachment might progress, it is impera- come in each frame is a Nash equilibrium based ontive that the ﬁrm make projections of how product the hypothetical but plausible reservation pricesattributes and costs will change over time and to iden- shown and a set of assumed costs. Per Characteristictify how these changes in performance and cost may 1, the new drive maintains the high-end market, whichimpact customer preferences over time. continually grows, and thus the encroachment starts An innovation such as the P-4 might be introduced from the high end and progresses down market. In theat a speed of, say, 500 MHz. Then over time, it will be rightmost frame of Figure 4, high-end encroachmentupgraded to 600 MHz, 700 MHz, and eventually to 1 has progressed to the point where the new product hasGHz or more. Similarly, if a new drive of current size the bulk of the volume and the current drive is rele-and a new thin-ﬁlm head is introduced, this new high- gated to selling to a small low-end market.end new disk drive might be expected to be further While this is illustrated in the context of a new diskupgraded with even more capacity over time (in fact, drive (for the reasons explained earlier), the patternthe thin-ﬁlm head might be introduced precisely to described for this type of new disk drive mimics whatallow for such further upgrades). Christensen (1992) has been (and continues to be) experienced with a newshows this trend in the actual disk-drive data. This generation of microprocessor. The new microproces-continual upgrading would tend to push the new sor encroaches from the high end for reasons analo-product’s reservation price curve upward over time gous to those just described.as shown in the progression in the three frames of Note that although a reduction in the price of aFigure 4 (the new drive’s curve becomes a bit steeper new product such as a new Pentium microprocessorover time—since the high-end customers have a large may over time contribute to high-end encroachment,appetite for capacity, they appreciate the capacity en- the encroachment framework is not based on any as-hancements more than the lower-end customers). sumption of a price trend per se. Rather, it is based onOver time the old product looks more and more an assumption that product performance changesinferior because it can’t keep up with customer over time (as do customer valuations of that perfor-appetites—its reservation price curve degrades over mance), which is precisely the notion conveyed bytime. Christensen’s trajectory curves. The framework pre- Simultaneously, empirical observation has veriﬁed sented herein adds richness by allowing for possiblethat costs in the electronics industry continually de- cost changes due to learning effects and further con-crease over time. For example, the cost of a transistor siders the trade-off that customers make between per-has decreased by roughly 30% per year, as shown in formance and price; recall that each customer buysSchmidt and Wood (1999). This type of learning and the product that offers not the lowest price, but thecost reduction contribute to Moore’s Law, for exam- highest surplus, where surplus is the difference be-ple. These cost decreases are often more dramatic for tween the customer’s reservation price and the ﬁrm’sthe new product; learning theory says that cost goes actual selling price. High-end encroachment occursdown by some percentage for every doubling of out- when high-end customers are the ﬁrst to buy and low-put and that the output of a new growth product er-end customers, over time, view the price of the newdoubles quickly while for a more mature product it product as less and less excessive, until eventually theytakes longer. consider it to be a better value than the old product.
WHEN IS A DISRUPTIVE INNOVATION DISRUPTIVE? J PROD INNOV MANAG 357 2008;25:347–369 At introduction Ol After several years After a few more years d Ol Ol d d N ew Reservation price Reservation price Reservation price New New Sales Sales Sales of old Sales of of Sales of new of old new old Sales of new Main- Mid- Desktop Laptop Specialty Main- Mid- Desktop Laptop Specialty Main- Mid- Desktop Laptop Specialty frame range frame range frame range Actual or Potential Market Segments Actual or Potential Market Segments Actual or Potential Market Segments (width of segment suggests number of customers) (width of segment suggests number of customers) (width of segment suggests number of customers)Figure 5. Product 2 Initially Sells to Fringe Low-End Customers and Then Encroaches Upward, Representing Low-End Encroachmentof the Fringe-Market TypeThis change in customer valuation may progress due concomitant decrease in price) can be attributed toto any combination of price decrease, performance learning effects and pricing strategies. The increase inimprovement, and customer perception. capacity makes the disk drive more attractive to all customers, but particularly to high-end customers. Product 2: A new drive of smaller size and lower ca- The drive is no longer so deﬁcient in the eyes ofpacity (low-end encroachment of the fringe-market high-end users. With regard to low-end users, ittype). Next consider what would be projected to hap- wasn’t all that deﬁcient to begin with, so the increasepen if the new drive were instead of smaller size and in willingness to pay is relatively smaller for them.lower capacity. The left frame of Figure 5 shows the Thus, the reservation price curve for the new drivereservation price curves. Again, these are determined shifts upward over time, as shown in the middle andas the sum of the relevant part-worth curves shown in far right frames of Figure 5. The slope actually be-Figures 2 and 3. Note that now the new product’s comes steeper but is still shallower than that of the oldreservation price curve has the shallower of the two larger drive. The enhanced reservation prices of theslopes. new drive, along with its reduced cost, make it a more Per Characteristic 1, in this case the ﬁrst buyers of formidable product compared with the old drive suchthe new disk drive are customers on the low-end fringe that over time more buyers are attracted to it.of the existing market (continue to refer to the left Per Characteristic 1, the old drive continues to sellframe of Figure 5). Speciﬁcally, they are at the high to the high end of the market, but in continuallyend (left edge) of the desktop computer market seg- smaller quantities. This time, the result of the de-ment. Note that even the desktop computer segment scribed incremental changes is that the market for themight be willing to pay more for the old drive than the new drive diffuses upward, from the low end towardnew drive (in this region the reservation price curve the high end. As the capacity of the new smaller drivefor the old drive lies above that of the new drive); increases, midrange and mainframe computer usershowever, note that the old product prices itself out of view it more favorably, increasing their willingness tothis market. As suggested earlier, the selling price of pay for this drive relative to the old drive. In the righteach product is indicated by the height of its sales frame of Figure 5, encroachment has progressed torectangle. The new product is priced low enough to the point where the new product has the bulk of thecapture just a sliver of the desktop market. Its selling volume, but this time the last vestige of sales for theprice is lower than the reservation price for just a current product lies at the high (left) end of the overallsmall segment of the desktop market; that segment market. Also note in this case that there has been aadjacent to the midrange market. good degree of market expansion compared with the Over time, the capacity of the new drive would be ﬁrst scenario.expected to increase, and cost per MB would drop This pattern mimics what was actually experiencedprecipitously. Again, performance cost changes (and a in the disk-drive market: The new generation was
358 J PROD INNOV MANAG G.M. SCHMIDT AND C.T. DRUEHL 2008;25:347–369indeed a smaller drive with lesser capacity; the drive drive is of the smallest size, with the lowest capacityﬁrst sold to users in a new desktop market segment; level. The left graph in Figure 6 shows the resultingand the laggards in converting to the smaller drive reservation price curves—each curve is again obtainedwere mainframe customers. The number of drives sold by adding the two pertinent lines for willingness toincreased dramatically. Since the new drive ﬁrst sells pay from Figures 2 and 3.to customers who were previously on the fringe ofbuying, this represents low-end encroachment of the Characteristic 2 (applies to the opposite-sloping sce-fringe-market type. nario): Unless all market segments are served, each Using this linear reservation price framework, Sch- product sells to one end of the market, isolated frommidt and Van Mieghem (2005) assign values to the any impact from the other product (sales prices andcosts and reservation price curves. Using these num- volumes are not affected by the presence of the otherbers they determine the Nash equilibrium market product).prices, which in turn determine market volumes. Intheir example, upon introduction of the new drive, the Per Characteristic 2, which follows directly fromincumbent’s market actually expands because the en- Theorems 1 and 2 of Druehl and Schmidt (2006), intrant’s innovation has motivated the incumbent to this scenario the ﬁrst buyers would be the specialtyslightly reduce price (by 3%). The incumbent’s proﬁt users, as shown by the narrow rectangle on the rightdrops by only 6%. In this analysis it is assumed the side of the left frame of Figure 6: The drive wouldincumbent starts as a monopolist. Under an assump- have too little capacity to be of much use for anyonetion of competition in the market for the old product, else. Effectively, the current and new products wouldthe innovation might initially have even relatively initially sell to the two opposite ends of the market,lesser impact on the old product. After three years, and in this sense the markets for the two productsthe two drives are sold in equal volumes, but the in- would be ‘‘detached’’ from one another. Interestingly,cumbent still retains 70% of the industry proﬁts and both products would be priced high, as indicated byhas had to reduce price by only 4%. But after six years the somewhat comparable heights of the rectanglesthe innovation has swept through the market, and the labeled ‘‘Sales of old’’ and ‘‘Sales of new’’ in the lefttotal market is much greater than previously. They do frame of Figure 6.not claim the absolute values of these numbers dupli- Over time, capacity would be expected to increasecate actual market experience, but qualitatively these and cost per MB would decrease. Referring back tocalculated outcomes mimic the actual outcomes in the Figure 6, this would make the new drive begin to lookmarket. more favorable to higher-end users of the old drive, making the reservation price curve steeper. Eventual- Product 3: A new drive of smallest size and lowest ly, laptop users and then desktop users would adoptcapacity (low-end encroachment of the detached-mar- the new drive (sales of the new drive would expand asket type). A third possible outcome is that the new shown in the progression from the left to the middle to At introduction After several years After a few more years Old Ol New d New w Ne Ol d Reservation price Reservation price Reservation price Sales of new Sales of old Sales of old Sales of old Sales of new Sales of new Main- Mid- Desktop Laptop Specialty Main- Mid- Desktop Laptop Specialty Main- Mid- Desktop Laptop Specialty frame range frame range frame range Actual or potential market segments Actual or potential market segments Actual or potential market segments (width of the segment indicates number of customers) (width of the segment indicates number of customers) (width of the segment indicates number of customers)Figure 6. Product 3 Initially Sells to a Detached Market and Encroaches from the Low End Upward, Representing Low-End En-croachment of the Detached-Market Type
WHEN IS A DISRUPTIVE INNOVATION DISRUPTIVE? J PROD INNOV MANAG 359 2008;25:347–369the right frames in Figure 6). As suggested by the 200righthand frame in Figure 6, the mainframe users Land lines Millions of subscriptionswould be the last to buy, after the market had swept 150through the laptop, desktop, and midrange markets:Even at the lower price of the smaller drive the main- 100frame users would hold out until the small drive hadenough capacity to become attractive at its lower price Cell phonespoint. 50 Note in the left frame of Figure 6 that the new driveﬁrst sells to customers in a segment that is ‘‘detached’’ 0from the old market segments (along the x-axis, the 1985 1990 1995 2000 2005new drive sells to the far right, and the old drive to the Figure 7. Subscriptions of Cell Phones and Land Lines (Cellleft). Thus, this is called the detached-market type of phone data from CTIA, 2003, and land-line data from FCC, 2004)low-end encroachment. The scenario just described was technically infeasi-ble, as well as being undesirable from a strategic andmarketing perspective. It was not possible to cost- end encroachment is possible even when the neweffectively make such a small drive, and the market product starts out as being high priced, if the newwas not ready for PDAs at that point. It was strate- product opens up a new market segment that is in es-gically more favorable for an entrant to instead en- sence detached from the existing market. The newcroach by decreasing size and capacity less market segment so highly values the alternate attrib-dramatically. As previously suggested, the second sce- ute that it is willing to pay a high price for it, even if itnario (involving the smaller drive with lower capacity) means the new product is severely de-rated with re-is the one that actually materialized. gard to the performance parameter that current cus- As discussed earlier, cell phones exhibit the char- tomers most highly value.acteristics of this type of encroachment. Cell phonesales grew for roughly 25 years before beginning to Product 4: An innovation where sales support isencroach on land lines (Figure 7). Early cell phone dropped (low-end encroachment of the immediatesubscriptions were expensive. However, early users type). A fourth possible innovation simply involvesdid not give up their land lines in ofﬁces and homes eliminating sales support (the drive’s physical size andbecause cell phone coverage was initially unreliable: capacity do not change). Under the assumption thatRemember the TV commercial that continually asked, high-end customers more highly value this support‘‘Can you hear me now?’’ (Note that one person can than low-end customers, this innovation is similar tosimultaneously represent multiple customers if that what Christensen and Raynor (2003) identify as aperson is in the market for multiple products at the ‘‘low-end disruption.’’ Their classic example involvessame time.) As cell phones became a little less bulky familiar brand-name goods, which were historicallyand less costly and as coverage improved, the ﬁrst sold in full-service department stores. K-Mart, Wal-signiﬁcant direct encroachment on the land-line mar- Mart, and other discounters innovated by offeringket was with lower-end land-line users such as college these goods without the support that the departmentstudents and second lines in homes. Cell phones con- store sales personnel offered. By simply offeringtinue to encroach up-market: For example, apartment goods at lower prices with less service, they immedi-owners are now dropping their land lines. However, ately encroached on the low end of the departmentmost homes still have a land-line subscription, and store market (as opposed to ﬁrst opening up a newvirtually all ofﬁces still have a land line. In other fringe or detached market).words, the ofﬁce segment, which represents the high- It seems plausible that the loss of sales supportend segment of the market, would be expected to be impacts the high-end customers most signiﬁcantly.the last to switch to the cell phone innovation (if they The logic would be that high-end customersever do so). expect ‘‘name brand’’ service, and a loss of such This points to another key insight. There are ex- support signiﬁcantly degrades their willingness toceptions to Christensen et al.’s (2004) rule that dis- pay, at least relative to the low-end customers. Con-ruptive new-market innovations are low priced: Low- versely, the low-end customers may not be as sensitive