The path of substitution in an industry is a function of how RVP, the perception of RVP, switching costs, and the propensity of buyers to switch evolve over time. The rate of penetration of substitutes differs widely from industry to industry. Some substitutes gain quick acceptance, while others penetrate slowly or not at all and are discontinued. In many industries, however, the path of substitution for successful substitutes looks like an S-curve, when substitution as a percent of total demand is plotted against time (see Figure 8-2).
The S-shaped substitution curve is closely related to the familiar product life cycle curve, because products early in their life cycles are frequently substituting for some other products. An S-shaped substitution path is not characteristic of every successful substitution. However, it is important to understand why an S-curve might occur so that the economic factors that underlie it are recognized.7 Where the underlying economics of a particular substitution suggest an S- curve path, a number of techniques are available to help forecast the rate of substitution.
In an S-curve path such as that shown in Figure 8-2, substitution is initially modest and often continues at low levels for a considerable period of time in what can be called the “informing and testing phase.” Unless a product flaw emerges with the substitute or the threatened industry responds to nullify the substitute’s advantages, the penetration of the substitute then often climbs rapidly in a “takeoff phase” toward an upper bound that represents maximum penetration. This upper bound is determined by the number of buyers for whom the substitute is potentially valuable. The upper bound of substitution can itself change over time as changes in technology or buyer needs bring buyers in or out of the pool.
Figure 8-2. A Typical S-Shaped Substitution Path
The reasons that an S-shaped substitution curve occurs reflect a set of assumptions about the interplay of actual and perceived RVP, switching costs, and the buyer’s propensity to switch over time. Initially, the performance of a substitute is likely to be uncertain, and there may be few capable firms to supply it. Many buyers are not even aware of the substitute or what its characteristics are. The price of the substitute is likely to be high because of its low volume or because suppliers are skim-pricing. Though the value of the substitute is often uncertain, however, the cost of switching is usually very clear and made particularly high by the unfamiliarity of both buyers and suppliers with how the substitute should be used. Moreover, the cost of switching must be borne up front, before the benefits of the substitute are realized.
During the informing and testing period, some adventurous buyers or buyers who attach a particularly high value to the substitute’s qualities will switch to the substitute to experiment with it or will switch permanently because the RVP of the substitute to them appears particularly high. During this period, either the substitute begins to prove its value or product flaws become apparent. The product flaws prove uncorrectable (and penetration never increases), or the substitute is improved (sometimes the substitute is completely withdrawn from the market and then reintroduced). At the same time, marketing activity and word of mouth tend to widen the group of buyers that are aware of the substitute and thus the perception of value improves.64
Assuming that the substitute eventually achieves acceptable performance with early buyers, the rate of penetration of the substitute can start to increase rapidly along the S-pattern for a number of reasons. First, the uncertainty of perceived value and the risk of failure of the substitute both fall as early buyers have successful experiences. Second, once a few buyers successfully switch, competitive pressures force other buyers to switch to maintain their cost position or differentiation (or self-image for consumers). Third, the cost of switching may decline for reasons discussed earlier. Fourth, rising adoption leads to increasing awareness of the substitute and raises its credibility. Fifth, rising penetration of the substitute often reduces its cost through economies of scale and learning.9 Sixth, introduction of new varieties of the substitute opens up new industry segments. Finally, increasing penetration of the substitute prompts its suppliers to become more aggressive in pricing, marketing, and R&D, often as a result of entry by new competitors into the substitute industry. All of these factors tend to be self-reinforcing, and can lead to the extremely rapid penetration of the substitute.
Eventually, the penetration of a substitute begins to approach 100 percent of the buyers to which the substitute is attractive. As this happens, penetration tends to level off as the penetration of new buyers becomes more and more difficult. However, improvements in RVP or new varieties of a substitute may widen the pool of potential buyers beyond those initially foreseen, fueling new opportunities for growth of the substitute. The upper bound in Figure 8-2 can thus expand to contain a larger and larger number of buyers.
At the same time, the buyers’ usage of the substitute may change in ways that increase or decrease demand. In TV sets, for example, sales of black and white sets were very robust even after color sets began to penetrate because buyers bought second, third, and even fourth TV sets. Similarly, in the substitution of electric shavers for conventional razors, the recent introduction of small, portable electric razors may well be shifting the upper bound of penetration of electric shavers by making them more versatile and less expensive. Thus the upper bound in Figure 8-2 can include a growing volume of units.
The length of the informing and testing phase is a function of a number of factors. Clearly important is the size of the RVP improvement offered by the substitute—the bigger the inducement, the shorter the period will be. The length of time necessary to prove the performance of a substitute also differs by product, and can greatly influence the length of the testing period. The performance of an automatic drip coffee maker can be proven in weeks or months, for example, while a new piece of capital equipment may require years of production line testing to assess its true performance. The period required for the industry producing the substitute to make necessary improvements in performance or cost, and to build adequate capacity to serve major buyers, also influences the length of the testing phase. Finally, the intensity of competition of the buyer’s industry and the significance of the specific RVP of the substitute to that competition influence the length of the testing phase, by determining the pressures for imitation if one buyer switches to a substitute.
The steepness of the takeoff phase is a function of how compelling in an industry are the reasons for rising penetration described above. In selling a substitute to an intensely competitive buyer industry, for example, takeoff can be very rapid. The steepness of takeoff is also a function of the time required to change over to a substitute and the adequacy of capacity. The purchasing cycle in the buyer industry is also important, because buyers are more likely to switch to a substitute when they would normally reorder or replace the product anyway. This reduces the switching costs associated with scrapping a product which has years of useful life remaining or where the buyer has substantial inventory on hand. For similar reasons, switching to a substitute in durable goods tends to occur faster when the buyer industry is growing and therefore investing in new facilities and equipment.
The response of the threatened industry is also clearly important to the path of substitution. An aggressive response by the threatened industry can sometimes stop the penetration of a substitute altogether or delay it considerably. Conversely, entry into the substitute industry by credible competitors can accelerate penetration, as IBM’s entry into personal computers and Kodak’s entry into instant cameras seem to have done. Smooth substitution curves such as the one portrayed in Figure 8-2 are most characteristic of industries where there are many buyers of a substitute. Where there are few buyers, a decision by a major buyer can dramatically shift the curve overnight. In such cases, substitution analysis is best conducted on a buyer-by-buyer basis.
1. Segmentation and the Substitution Path
The substitution path in an industry is often very closely related to industry segmentation. Early penetration occurs in industry segments where a substitute offers the highest RVP, requires the lowest switching costs, and/or encounters the most adventurous or highest value buyers. Early segments support the cost reduction or performance improvement necessary to penetrate later segments. The high value of the substitute to early segments either offsets the high initial cost of the substitute, or allows its producers to earn extremely high profits. Margins often will fall over time in the substitute industry as additional segments are penetrated where the substitute has lower value.
The substitution of minicomputers provides a good example of how the substitution path relates to industry segmentation. The early segments penetrated by minicomputers were scientific and computer center applications where computational power was needed but where users could do their own programming, adapt machines to their needs, and perform some maintenance in- house. Later penetration was in applications such as industrial controls, where the buyer was still sophisticated and support requirements were moderate. Only after some time did minicomputers develop the service and support capabilities to penetrate small business applications.
The same factors that lead to early penetration in some segments also mean that the rate of penetration within segments will vary. Since segments vary in RVP and switching costs, penetration will proceed much faster in some segments than others. Hence the industry substitution curve is actually a collection of segment substitution curves.
2. Substitution Forecasting Models
The observation that successful substitution often follows an S-shaped penetration curve can be used in forecasting. A variety of models that are based on the assumption of S-shaped penetration have grown out of research on diffusion processes. Data from the early years of a substitution can be used to forecast the entire substitution curve using such models, under the assumption that an S-shaped process will occur. The forecasted substitution curve may then become a base case from which an analysis of the underlying economics of a substitution can begin. Adjusting the standard S-shaped curve to reflect the economics of a particular substitution allows a prediction of the extent of substitution in future years. The premise underlying such a procedure is that the tendency toward an S-shaped penetration curve is strong enough so that the plotting of such a curve—derived from the early penetration data—is a useful beginning point for analysis.
The most commonly used diffusion model is the so-called “logistic function,” a form of exponential function.65 The functional form of the relationship as applied to substitution is as follows:
where F = fraction of the total potential market that has switched to a substitute
K = a constant set equal to the early growth rate of a substitute
The logistic function makes two important assumptions: (1) if substitution has progressed as far as a few percent, it will proceed to completion; and (2) the fractional rate of fractional substitution of a substitute for the product is proportional to the remaining amount of the product left to be substituted for. It is the latter assumption that produces the S-curve shape. If F/(1 — F) is plotted as a function of time on semilogarithmic paper, the logistic function results in a substitution curve that is a straight line with the slope K as shown in Figure 8- 3.
If the logic underlying an S-shaped substitution curve is believed to hold in a particular substitution process, as well as the assumption about a constant fractional rate of fractional substitution, the logistic function can be used to forecast the path of substitution that will occur. To do this, data on substitution history are used to determine F/(l — F) for each year. Then F/(1— F) is plotted on semilogarithmic paper against time, and a straight line is fitted through the early history. Extending this line projects the future substitution path under the assumption that it will follow the logistic curve.
Figure 8-3. A Typical Logistic Curve
The procedure is shown in Figure 8-4, which plots the early substitution history of aluminum for steel in beer cans. The early history fits the logistic curve quite closely. Based on this history, we could expect the extent of substitution by 1982 to be approximately 91 percent if the assumptions of the logistic curve hold. The accuracy of this forecast, however, depends on how well the logistic curve reflects the economics of substitution in the particular industry. One important issue is the size of the potential market, and hence the upper bound. In the beverage can case the upper bound is fairly clear. Where the upper bound is increasing, the logistic curve will have a tendency to overestimate the rate of penetration. Another important issue later in the substitution process is whether the intensity of use of the substitute required to perform its function will change over time.
Perhaps the most important issue in whether substitution will follow the logistic curve growing out of the early substitution history is the extent to which the RVP that is driving the substitution will change over time. The logistic curve assumes a stable motivation for substitution. If RVP falls, the penetration of a substitute may actually decrease, something that never happens with the logistic curve. If the RVP of a substitute improves, on the other hand, the rate of penetration will increase from early history—the substitution path may jump to a new curve. Figure 8-5 illustrates how this has actually happened in beer cans.
After 1976, the introduction of two-piece steel can technology helped slow the penetration of aluminum cans. Working with steel companies, Crown Cork and Seal introduced a cheaper way of making steel cans that reduced the RVP of aluminum cans. In 1978, however, Miller Beer (number two in the beer industry) announced a major test of aluminum cans against steel. The rate of substitution slowed further pending the outcome. Miller’s decision in 1979 was to switch to aluminum. Once Miller’s switch occurred, substitution accelerated to a faster pace than before 1976, in part because Miller legitimized the substitute. An additional factor in the faster penetration of aluminum after 1979 was the growing recycling of aluminum cans, which are more recyclable than steel cans. As infrastructure allowed for more widespread recycling, the RVP of aluminum became even better than before because recycling reduced the cost of metal. By 1982, penetration of aluminum was actually about 98 percent instead of the 91 percent that would have been forecasted using the early history.
As the beverage can example illustrates, the logistic curve is not a substitute for substitution analysis. It is a simple tool that can serve as a starting point for more careful analysis of the economics of substitution. The substitution path will vary from industry to industry, and will be affected by technological changes and competitive moves. In order for the logistic curve to be meaningfully used, the factors that determine RVP, switching costs, and propensity to substitute in a particular industry must be understood and expected changes in them forecasted. Since the economics of a substitute often differ by industry segment, logistic curves should usually be plotted at the segment level and not the industry level. Beer producers, for example, have different needs in cans than other can users. A logistic curve for aluminum’s substitution against steel in cans overall would not be as meaningful as one for beer cans alone.
Source: Porter Michael E. (1998), Competitive Advantage: Creating and Sustaining Superior Performance, Free Press; Illustrated edition.