Designing new products to meet the requirements of a ‘horizontally differ- entiated market’ presents a different set of challenges for a firm. In such a
Handbook of New Product Development Management
market, customers vary in the extent to which they value different attributes of a product. However, horizontally differentiated customers can neither be rank-ordered according to their willingness to pay for quality improvements, nor are they unanimous in ordering different products based on their perfor- mances. The main objective for firms when facing such variety is to provide a line of products that meets demands of different customer segments suffi- ciently, yet economically. Economic models are useful in understanding the trade-offs between the cost of variety and benefit of providing customers their ideal products.
A traditional approach to modeling horizontally differentiated markets is to consider different segments as being separated in space. This can be traced to Hotelling (1929) who assumed that customers are distributed in a line of varying preferences with customers on either end of the line representing those with completely different preferences with respect to product char- acteristics. Naturally, meeting the specific needs of an infinite number of customers is impossible. A more realistic approach is to identify the exact preferences of different clusters of customers that exist in the market. An effective approach to identifying these segments is to infer the weights dif- ferent customers place on various attributes by performing a conjoint anal- ysis (Green and Krieger, 1989). While the marketing models are useful in understanding the need for product variety itself, models that are more sophisticated are needed to understand the costs of providing this variety and to explore alternatives in design & development of the product line.
Krishnan, Singh, and Tirupati (1999) present the following model that inte- grates customer-demand information with the design cost information to make platform and product family-planning decisions in a horizontal differentiation context.
Firms target a performance interval Q in which to introduce products, which is generally aimed to cover the customers at the ends of the preference spectrum. Suppose the firm considers n variants in the product family with respective performance levels q1 q2 qnqiQ i=1n such that q1<
q2< < qn. The maximum performance qT in the product family that the firm can offer (such that qT≤qT) is determined by the technology T that the firm develops at a cost CDqTP, where P is due to the level of technology already possessed by the firm, or in other words, the quality of the platform.
As with vertically differentiated markets, the pricepia consumer will pay for the product will depend on the product’s characteristicsqi as well as on other products that he has the option of purchasing. The life-cycle demand for variantidi(Eq. 2), can be obtained from a general distribution of consumer tastes (or preference for performance levels). For instance, if fzdenotes the distribution of consumer attributesz, andzirepresents the proportion of
• • • • • 102
consumers with performance levelszthat purchase variant i, the demand for this variant can be calculated as follows.
di= 0
zif z dz (6)
Such a model also permits the incorporation of demand uncertainty by allowing the distribution fz to be a random variable at performance level z. When zi is independent of fz, the expected demand Edi is given by:
E di=
0
ziE f zdz (7)
The demand model described above is similar to a consumer choice model in which consumers choose products ‘closest’ to their ideal points (Carpenter and Nakamoto, 1990). The ideal-point model of demand is based on the premise that (i) consumers have different preferences for each price-performance com- bination; and (ii) each consumer prefers one price-performance combination over any other, where product performance defines the ideal product for the consumer.
While this model is adequate to calculate demands for different variants in a product line, it is also useful to understand how consumers seek the best fit before deciding the variants that should be offered in the product family. A consumer with an ideal performance levelzassociates a utility valueuzqi pi with variant I (for convenience, we will henceforth simply use the termuzi for this utility function). Consumers are performance-sensitive, and do not consider a product for purchase if a product with a performance closer to their ideal point is already available. Thus, the only products considered by the consumer are the two neighboring productsiandi+1 such thatqi≤z≤qi+1. Using a share of utility-based choice rule (see Green and Kreiger, 1989) to describe the probability of purchase of a product by any consumer in the market, the purchase probabilities (when the choice is limited to the two productsqiandqi+1) can be described by.
zi zi+1
=
⎛
⎜⎝
uzi uzi+uzi+1+uz0
uzi+1 uzi+uzi+1+uz0
⎞
⎟⎠ (8)
Handbook of New Product Development Management
whereuz0is the consumer’s utility if neitherinori+1 is chosen (‘balking’).
With the above model to determinezi, the expected life-cycle demand can be written as follows:
E di=qi+1
qi−1
ziE f zdz fori=12 n−1 E d0=q1
0
ziE f zdz E dn=
qn−1
ziE f z dz
(9)
On the cost side, a platform comprising of reusable components is assumed to be architected in the planning phase and reused in all product variants. Each variant consists of components that are either unique to the variant, are a part of the platform, or are adapted from existing variants (see Fig. 4.2). Development costFaggin the aggregate-planning phase involves the development of the platform and depends on the scope of the platform P (the number and complexity of functional elements that are reused throughout the family).
Creating a platform that allows for greater reusability (larger) may require more effort in the specification of task structures and components, which in the individual development phase would pay off in the simplification of development tasks (Clark and Baldwin, 1993).
Development costs for an individual variant are incurred in the phase following aggregate development, and consist of (i) creative design cost I qj that involves creation of functional elements which are unique for each variant, and (ii) adaptation costs g qi qjthat involve the reuse and adaptation of components designed for earlier variants. Effort invested in creating a platform can contribute to the decrease of both the creative design and the adaptation costs. The cost of creating components unique to variant j is g qi qj, where is an efficiency parameter that is depen- dent upon the process of design creation: all else being equal, a firm with a more efficient design process, or greater (perhaps as a result of invest- ments in systems and processes such as CAD/CAM and rapid prototyping), will incur smaller costs in the development of the platform and individual variants.
This comprehensive consideration of development costs can be succinctly summarized as the total development cost function below, where S= 12 nis the set of products in the family.
G S=Fagg+n
i=1
I qi+g qi−1 qi (10) Production costs are affected not only by both the specific configuration of the variant, but also by the design of the platform: qi. Now the total
• • • • • 104
cost function for the output (volume) vector dS=d1 d2 dn can be written as.
CD dS=G S+
i∈S
di qi (11)
Such cost models, as shown in this instance, can be developed without inter- fering with the model of demand. A simplified model of the market may capture the price offered for any variant aspqi. The optimization problem for the firm may be written as.
q1 q2 qn n= n
i=0
p qi−v qidi−Fagg
− n
i=1I qi+g qi−1 qi
−CTqn qP (12) st
q0≤q1≤q2≤ ≤qn ∈1 2 Interpreting the optimal design formulation
Interestingly, despite its complex structure, the entire problem of selecting the optimal product family from any platform can be reduced (with a few assump- tions) to a well-known and readily solvable network optimization problem called the Shortest Path problem (Fig. 4.4). The optimal solution to the prob- lem selects the least cost path to meeting customer needs with the platform’s derivatives. The nodes through which the shortest path flows represent the variants that are actually offered in the optimal product line.
There are many benefits to reducing the formulation to a simple form. This formulation is now easily extendable for a firm that is considering extensions to or deletions from an existing product family. The solution gives ready and intuitive insights about the width and spacing of the product line, and even allows managers to identify the platform. This is an important function of modeling these decisions because a big challenge for many managers who are interested in developing platform-based product families is identifying the basic platform. These explicit models also allow managers to explicitly calculate the benefits of using a platform. The model is amenable to incor- porating competitive forces into the decision-making process. Firms can use these models to evaluate even dynamic, strategic benefits of platforms. Fur- ther, these models and the structure of the product line are robust to extensions such as consideration of multiple dimensions of performance.
Handbook of New Product Development Management
S 0 1 2 N E
C2N
C1E C2E
C1N
C12 C01
C02 CSE =0
CS0
C0N
Figure 4.4
Network representation of the product family design problem.