|
Article Excerpt
We present a model describing the demand dynamics of two new products competing for a limited target market. The demand trajectories of the two products are driven by a market saturation effect and an imitation effect reflecting the product experience of previous adopters. In this general setting, we provide analytical results for the sales trajectories and life-cycle sales of the competing products. We use these results to study the impact of launch time on overall life-cycle sales. We consider the perspective of one of the competing products and model the trade-off between the lost revenues resulting from a delayed launch and the lower unit-production costs. We find that the profit-maximizing launch time exhibits a counterintuitive behavior. In particular, we show that a firm facing a launch time delay from a competing product might benefit from accelerating its own product launch, as opposed to using the softened competitive situation to further improve its cost position. We identify conditions under which a marginal cost-benefit analysis leads to suboptimal launch-time decisions. Finally, we analyze the Nash equilibrium in launch-time decisions of the two competing products.
Subject classifications: new products: cross-functional performance metrics; marketing-operations coordination; competitive diffusion dynamics; cost of delay.
Area of review: Manufacturing, Service, and Supply Chain Operations.
**********
1. Introduction
A common problem in product development is the trade-off between the four performance metrics: product development lead time, product unit cost, technical product performance, and overall development cost (Smith and Reinertsen 1991). To find a balance between these conflicting objectives in their day-to-day decision making, development teams typically rely on simple trade-off rules. Such trade-off rules attempt to "dollarize" (i.e., assign a financial value to) changes along any of these metrics and thereby to create a common ground for comparison across organizational functions (Ulrich and Eppinger 1999).
While understanding the financial impact of changes in development cost and product unit cost is relatively simple, understanding the financial impact of a change in launch time is rather difficult: What is the cost of a one-day launch delay, e.g., in the automotive industry? One million dollars? Maybe two? How can we adequately capture the long-term impact on market share resulting from the launch delay? Clearly, whichever answer we choose will have a substantial impact on the quality of the development team's decision making. Yet, despite the pivotal role of trade-off rules in guiding decision making and enabling cross-functional coordination, they are commonly formed in an ad hoc fashion. One major shortcoming of the current decision-making process is its simplistic treatment of demand dynamics over the product life cycle. Standard models either assume that the life-cycle demand of a new product is exogenous and therefore not affected by a delayed launch (the sales curve is just shifted into the future), or that there is a predetermined market window, after which sales are reduced to zero (Ulrich and Eppinger 1999). Factors that have been identified as critical success drivers for a new product, such as time-to-market relative to competition (Porter 1985, Kalish and Lilien 1986) or product diffusion (Bass 1969, Krishnan et al. 2000), are, at best, only included qualitatively.
The first objective of this paper is to overcome this shortcoming by developing quantitative trade-off rules with respect to product development lead time. In contrast to the simplistic treatment of demand dynamics prevalent in existing models, the trade-off rules we derive are grounded on a detailed analysis of competition and product diffusion, and thereby allow for an endogenous analysis of the life-cycle demand with respect to changes in product development lead time. Our second objective is to demonstrate how such trade-off rules can be used in the case of a development team that balances the lost life-cycle revenue resulting from a delayed launch with the reduction in product unit costs.
Our work is contributing to three literature streams discussed in [section]2: product development decisions, normative models of competitive product diffusion, and marketing-operations coordination. Specifically, we provide the following novel results. First, we present closed-form expressions for the diffusion of two competing products within the same category (Theorem 1). We investigate how life-cycle sales are influenced by the values of diffusion parameters at the brand, as well as the category, level (Theorem 2). This extends the Bass model of new product diffusion to a competitive setting, including the interactions (word of mouth) between potential adopters of one brand with previous adopters from the same brand, as well as from a competing brand. While these effects have been observed empirically (Mahajan et al. 1993), they have never been formally modelled (see Table 1 for an overview of related research in competitive new product diffusion).
Second, we analyze the coordination between the marketing and the operations efforts of a team developing one of the competing products. In particular, we model the team's decision of finding the optimal product launch time. From a marketing perspective, the team prefers to launch the product sooner, as this would lead to higher life-cycle sales. From the operations perspective, the team prefers to spend additional time on the detailed engineering of the product and process, leading to lower product unit costs. We show how a project manager can resolve this tension by deriving (dollarizing) how a change in launch time impacts profits. We establish conditions under which the immediate launch of a new product is optimal. We also demonstrate that making product launch time decisions on a marginal profit basis can be misleading; we show that it might be optimal to delay product launch despite a negative marginal value of a longer development time (Theorem 3).
Third, we analyze the existence and the nature of the Nash equilibrium with respect to the product launch times of the two competing products. For the case of symmetric products, we derive sufficient conditions for the existence of a Nash equilibrium in which both competitors launch immediately (Theorem 4a). In addition, we identify the cost parameters for which a pure-strategy equilibrium does not exist. We also derive a set of conditions that characterize the nature of the equilibrium for asymmetric players (Theorem 4b).
2. Related Literature
Ulrich and Eppinger (1999) recommend a four-step procedure towards evaluating the trade-off between product development lead time and product unit cost: (1) Build a base-case financial model (including a spreadsheet and a representation of life-cycle demand), (2) perform sensitivity analysis to understand the key assumptions of the model, (3) use sensitivity analysis to understand the trade-offs (including the trade-off between cost and time), and (4) consider the influence of qualitative factors, including competition and other market characteristics. The strength of this approach is its simplicity and the little effort required for implementation. However, the approach following Steps (1)-(3) is biased towards the easily measurable costs, including idle production plants, the cost of capital, and the expenses related to additional development time, while ignoring the "hidden cost" associated with the negative impact on revenues that results from a delayed launch.
[FIGURE 1 OMITTED]
A good example of this approach can be found in Clark (1989, p. 1260), who reports: "Research indicates that each day of delay in market introduction costs an automobile firm over $1 million in lost profits, not including the impact of lost market share." While such numbers are certainly effective in directing senior management's attention to even minor launch delays, they are of little value when guiding development teams in their operational decisions. Whether or not a development team in the automobile industry around 1989 was well advised spending $1.5 million to avoid a one-day launch delay depends on the exact size of the resulting marketshare loss. Thus, a more detailed model is needed that goes beyond treating the market side as a qualitative factor or a residual.
Several authors provide qualitative guidelines on how revenues are affected by a slow (delayed) versus a fast (accelerated) launch, four of which are displayed in Figure 1. Urban and Hauser (1993) argue that reducing product development lead time will increase life-cycle sales, however, at a diminishing rate. It is also suggested that an accelerated launch will increase overall development cost. Given that additional time in development reduces cost at a diminishing rate, the graph suggests the existence of an optimal launch time. While Urban and Hauser do emphasize the importance of competition, neither sequence of entry nor the duration of the first-mover monopoly is visible in their graph.
Rosenthal (1992) takes a slightly different approach. Based on a forthcoming competitor's product introduction, a late launch will give the product a shorter growth period, and thereby smaller peak sales. Moving from actual (late) introduction to the earlier, planned introduction seems to indicate sales increases at an increasing rate (peak moves up, so does duration of the monopoly period). The frame-work explicitly includes competition--assumed to begin at the time when sales start to fall--and takes the perspective of the first mover.
Kalyanaram and Krishnan (1997) suggest a convex-concave relationship between product development lead time and life-cycle sales. The convex part of their graph results from the product diffusion in its monopoly phase. The switching point (from convex to concave) indicates the beginning of the competitive phase. Similar to Rosenthal's model, the authors include competition and take the perspective of the first mover.
Finally, Wheelwright and Clark (1992) argue that there are steep gains associated with shortened development lead times, especially for companies who are "head to head" with their competition. Getting too far ahead does not yield the desired increase in profits, and can even result in profit loss.
Taken together, all four graphs in Figure 1 emphasize the impact of changes in product development lead time on a product's life-cycle sales and profits. However, by contrasting the four graphs, we can make two interesting observations. First, all four curves are purely qualitative and support, at best, Step (4) in the Ulrich and Eppinger procedure. Because none of the curves is described in a functional form, they are impossible to use for quantitative decision making. Second, none of the four graphs is derived formally from a transparent set of assumptions, making it hard for a project manager to judge if the corresponding model fits her current situation. In particular, the effects of competition and diffusion (the main differentiators from spreadsheet-based models) are included in a rather informal manner.
Detailed Models of Market Demand
As a first step towards developing a model of competition and diffusion, we turn to the marketing literature, which provides a rich stream of research in the area of new product diffusion models (Bass 1969, Mahajan et al. 1993). Most prominently, the Bass model stipulates that sales for a new product are initially low, as there exists limited word of mouth for it, and customers only adopt the product in response to external influences (captured by the coefficient of innovation). With more customers adopting the product, the word-of-mouth effect for the new product becomes stronger and the sales rate increases (captured by the coefficient of imitation). Finally, the sales rate decreases, reflecting the overall market saturation.
A model of category-level diffusion is important when analyzing the demand dynamics of two competing products within a category. Krishnan et al. (2000) discuss how Chrysler introduced the Caravan and the Voyager (brands) back in 1984 and thereby acted as a pioneer in the minivan (category) market. When a few years later Ford introduced the Aerostar, Ford was able to benefit from the category awareness for minivans and captured a sizable portion of the market. Thus, although the first product to market obtains 100% of the category sales up to the arrival of the second product, initial sales can be low, given the limited awareness for the new category. We will label this effect as the category-awareness effect.
Whereas the original models of new product diffusion were applied at the product category level, the last 15 years have witnessed the evolution of a significant body of research on diffusion models incorporating the effects of brand competition within a category (see Chatterjee et al. 1998 for an overview of competitive diffusion models). There are two effects of brand-level diffusion that are important to consider when modeling the impact of launch time on demand dynamics, brand-level word of mouth, and cross-brand word of mouth.
In addition to the word-of-mouth effect at the category level, customers also exchange information at the brand level. Consider a customer deciding which brand within the mobile phone category to adopt. A customer who has interacted with a prior adopter of Brand A will not only be more likely to also buy any mobile phone (category-awareness effect), but will be more (or less) likely to also adopt Brand A. To capture this effect, Krishnan et al. (2000) extend the traditional Bass model to the brand level and assume that each brand has its own coefficient of innovation as well as coefficient of imitation. The coefficient of imitation in the model presented by Krishnan et al. (2000) captures the effect that prior adopters of the category have on the future adopters of a brand as a "... collective force of all previous adopters that act on each brand's future adoption" (Krishnan et al. 2000, p. 271). While lumping the effects of brand-level word of mouth together into a collective force makes the resulting diffusion equations more elegant, it does not separate between the word of mouth for Brand A coming from customers who have adopted Brand A versus customers who have adopted Brand B.
However, such a separation can be important, especially when studying the sales loss of...
|
