Line Extensions: a Categorization and an Information Processing Perspective

Kalpesh Kaushik Desai, The University of Texas at Austin
Wayne D. Hoyer, The University of Texas at Austin
ABSTRACT - This paper addresses the theoretical principles involved in the processing and evaluation of line extensions by the consumers. We propose a theoretical model of processing of line extensions and identify some moderating factors influencing the evaluation of line extensions. We also develop a classificatory framework and discuss the implications of the theoretical concepts for the framework.
[ to cite ]:
Kalpesh Kaushik Desai and Wayne D. Hoyer (1993) ,"Line Extensions: a Categorization and an Information Processing Perspective", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 599-606.

Advances in Consumer Research Volume 20, 1993      Pages 599-606

LINE EXTENSIONS: A CATEGORIZATION AND AN INFORMATION PROCESSING PERSPECTIVE

Kalpesh Kaushik Desai, The University of Texas at Austin

Wayne D. Hoyer, The University of Texas at Austin

ABSTRACT -

This paper addresses the theoretical principles involved in the processing and evaluation of line extensions by the consumers. We propose a theoretical model of processing of line extensions and identify some moderating factors influencing the evaluation of line extensions. We also develop a classificatory framework and discuss the implications of the theoretical concepts for the framework.

INTRODUCTION

With the cost of launching new products greatly on the rise and with an ever increasing failure rate for new products, companies have started paying closer attention to the importance and value of existing brand names. Clear evidence of this fact is that companies have launched many new variants of brand which are called line extensions. Line extensions are using the same brand name to launch new variants in the same product category. Of the 10,000 new products launched in the packaged goods industry in 1988, about 90 % were line extensions (Marketing and Media Decisions 1989). Take any product category and you will find that there is hardly any brand which has not extended its line, be it toothpastes, automobiles, music systems, etc. In the toothpaste product category, for example, a consumer can purchase Crest Fluoride, Crest Tartar control (regular flavor and cool mint flavor), and Crest Sparkle.

In light of ever increasing launches of line extensions, managerial guidelines regarding which brands should extend into what variants is badly needed. Despite its importance, however, it is surprising to find that this area has hardly been studied by marketing and consumer researchers. It is the opinion of the authors that theory and research on categorization and consumer information processing have much to offer marketers in developing managerial guidelines. Thus the objectives of this paper are : (1) To discuss the theoretical concepts involved in the introduction of line extensions, (2) To propose a model of processing of line extensions, (3) To highlight the factors moderating the evaluation of line extensions, (4) To put forth a framework to classify line extensions.

LINE EXTENSIONS VS BRAND EXTENSIONS

In relation to present interests, an evaluation of line extensions involves the task of trying to judge the similarities between two instances of the same product category. For example, line extensions involve the task of judging the similarities between "Crest tartar toothpaste" and "Crest fluoride toothpastes". This is in contrast to brand extensions where the consumers judge the similarities between two instances bearing the same brand name but which are in two different product categories ("Crest toothpastes" and "Crest chewing gum"). As a result, Barsalou (1989) concludes that intraconcept similarity (line extensions) should generally be higher (and hence more acceptable to consumers) than interconcept similarity of two related concepts (brand extensions). A second difference between the two concepts is that with brand extensions, the company is trying to exploit the potential of the brand in different product categories. Therefore, the brand is trying to woo the same group of consumers who are currently using the parent brand. Alternatively with line extensions, the company is trying to exploit the potential of the brand within a product category. Thus, the brand usually is trying to woo new users (in most of the cases) and extend the usage of brand by making the current users use the different variants. Most of the research in this area has focused on brand extensions. The present paper represents an attempt to extend previous work in four important ways. First, none of the previous studies has examined issues related to line extensions, the focus of the present paper. Second, this paper attempts to go beyond previous efforts by introducing new variables from the categorization theory C "concept conjunction"; "dominant concept"; "concept confusability"; and "concept incompatibility" to explain the evaluation of line extension. A third contribution is the identification of moderating factors which influences the acceptance of line extensions and finally, this paper develops a classificatory framework which will help us classify the different types of line extensions so that meaningful strategies for each of the types could be developed.

THEORETICAL CONCEPTS

Before presenting a theoretical model of line extensions, it is necessary to define a few key concepts. These are: brand concepts; concept conjunction; and dominance concept.

Brand Concepts

"Concept" is one of the important variables in the categorization literature (Barsalou 1987; Brooks 1978). Barsalou (1987) defines the "concept" as the particular information used to represent a category (or an exemplar) on a particular occasion. According to Barsalou (1987), the concept has two parts, one which is context independent information (cii), which is the information used to represent the category in most of the contexts. For example, whenever a person thinks of Honda s/he immediately thinks of "reliability" and "good quality" or for Macintosh the ciis might be "user friendliness." Because the ciis are so accessible, they may be automatically incorporated into all the concepts constructed for it. This would result in their being "cores" for the concepts. Such cores provide the consumers with expectations that are useful for interacting with a category in most contexts. According to Barsalou (1980), properties high in diagnosticity (useful in distinguishing instances of one concept from other concept instances) and properties relevant to how people typically interact with instances of respective concept would become the ciis. We also posit that since context independent information form the core of the brand concept, unique brand associations, important associations common to all the variants of the brand and some evaluative attributes like "reliability", "quality" etc would also form part of the ciis. However, it is not essential that the ciis be same for different consumers (or group of consumers), therefore there can be more than one ciis for the same stimuli or brand.

The other part of "concept" is the context dependent information (cdi) or the information used to represent the category in that context. For example, the concept of "power" and "excitement" would be felt by a person driving a Honda motorbike (but not a Honda car) and similarly for Macintosh, it could be the "good graphics facility when one is using graphics. Properties of a concept not typically found for familiar referents may become cdis through disuse (Barsalou 1980). We also posit that context dependent information would consist of context specific attributes and specific functional attributes of the brand. Just as in case of ciis, different consumers (or group of consumers) might hold different cdis for the brand. Therefore the brand concept (against which the new line extension will be compared) will differ from one consumer to another.

FIGURE 1

A MODEL OF EVALUATION OF LINE EXTENTSIONS

Concept Conjunction

Hampton (1987) proposed a model of concept conjunction to explain the nature of the resulting concept which arises when the two concepts are combined. The principle of concept conjunction will explain as to which attributes of the two concepts (to be combined) will be important for the new line extension as perceived by the consumers. Thus, his model would help us know what the new concept of "Head & Shoulders natural shampoo" be, when the two concepts of "Head & Shoulders shampoo" and "Natural shampoo" are combined. The model proposes that the attributes of the new conceptC Head & Shoulders -Natural shampoo are formed by the union of the attribute sets of the two existing concepts, namely those of "Head & Shoulders shampoo" and "Natural shampoo". According to this model, (1) attributes that are necessary for either of the two concepts will also be important for the new concept and (2) attributes that are impossible for either of the concepts will also be impossible for the new concept. For example, good quality engine is necessary for Honda concept, therefore this attribute of "engine quality" will also be important for Honda's extension into sports car, but the attribute of "more leg space" is impossible for a sports car and therefore "more leg space" is not going to be important for the new extension of Honda into sports car. However, if one of the attribute is necessary for one of the concept and impossible for the other concept, then we posit that this attribute would not pass on to the conjunction because the conjunction can't be formed with an impossible attribute. The relative importance (relative to other important attributes) of an attribute will depend on the importance of that attribute for both the concepts. The attribute which is important for both the concepts will be more important for the new line extension than the attribute which is important for only one of the concepts. For example, "fuel economy" is more important for Honda concept than it is for "sports car" concept, whereas the attribute of "durability" is important for both the concepts, therefore the latter attribute will be more important than the former for Honda's extension in sports car. In this way, the "important" attributes of the new extension concept reflects consumers' expectations of the nature of the new extension concept. Now, how would the ciis and cdis influence the formation of concept conjunction ? We hypothesize that since the ciis form the core of the brand concept, they would pass over to the conjunction whereas, the transfer of cdis would differ from one line extension to another.

Dominant concept

The dominant concept is related to the previous principle of concept conjunction. This principle tries to explain which of the existing two concepts involved in a line extension context will be dominant in determining the concept of the new line extension. The basic principle is that if one of the concepts has a greater number of salient and important attributes (for the new line extension) than the other concept, the new concept (line extension) will bear greater similarity to the former concept. Thus, the former concept would be the dominant concept. For example, if Rolls Royce were to extend into a sports car segment, then the dominant concept will determine whether this new car would be (consumer expectations) more similar to the "Rolls Royce" concept or the "sports car" concept. Therefore the dominant concept would by definition include only the important attributes of the constituent concept.

A MODEL OF EVALUATION OF LINE EXTENSIONS

Having introduced the theoretical concepts in the previous section, we now propose a model of evaluation of line extensions (fig 1.). The purpose of this model is to trace the process underlying the acceptance or rejection of the new line extension.

As shown in the figure, the starting point in the process of evaluating line extensions are the ciis and the cdis which together form the brand concept. For example, the ciis for Rolls Royce car could be "good engineering"' and a "luxury " car, whereas the cdis could be "good after sales service", and "excellent interior decor" of the car. Once the consumer becomes aware of the launch of the new line extension, s/he will decide whether the brand could launch this new extension. This is the first stage of "matching". The input to the first stage of matching is the brand concept and the extension product category associations. These two determine "product similarity" and "brand concept consistency" which are involved in the first stage of matching. The actual matching involves using "product similarity" and "brand concept consistency" to compare the new extension concept and the current brand concept (Murphy and Medin 1985; Park et al 1991).This is an overall evaluation i.e., without comparing attribute by attribute. Product similarity in case of line extension is not very important as the extension is in the same product category. The brand concept consistency would involve asking whether there is any inconsistency in the brand launching a particular extension. For example, can a conservative luxury car like Rolls Royce launch a sports car? A negative answer will lead to the rejection of the new extension whereas, a positive answer would lead to the next stage of matching. In the second stage, using the principles of concept conjunction and dominant concept, the consumers try to visualize what the concept of new line extension should be (i.e which attributes of the original two concepts should be incorporated in the new extension). This concept is defined as the "expected" concept of the new extension. Let us assume that in our case, after using the principles of "concept conjunction" and "dominant concept" the consumers expect that the new Rolls Royce sports car should have the following features of the "Rolls Royce" concept : "excellent engineering"; "smooth ride"; "good interior decor"; and "high price" and it should also incorporate the following features of the "sports car" concept : "quick acceleration"; "good styling of the car"; and "one door." Looking at the desired features of the two concepts (Rolls Royce and sports car), the new line extension in this case will be more similar to the concept of "Rolls Royce car" than that of "sports car", since the number of features or attributes of the "Rolls Royce" concept desired in the extension is greater than those desired of the "sports car concept". Therefore the concept of "Rolls Royce car" will be the dominant concept in this case. This means that the consumers expect the new extension to be more similar to Rolls Royce car than to sports car. However for some consumers, the product concept might be dominant (for e.g., those who are indifferent to the different variants, those who are "variety seekers"). This expected concept is then compared to the "actual concept" of the new extension. This is the second process of "matching". Since the concept conjunction and dominant concept reflect consumer's expectations about the new extension concept, a mismatch between the expected and the actual concept would lead to a negative evaluation and possible rejection of the new line extension CC Satisfaction literature (Oliver and Desabro 1988). However, if the new concept is as per their expected concept, the line extension will be accepted. So for example, for those consumers, whose dominant concept is Rolls Royce car, if the actual concept of Rolls Royce sports car is tilting more toward the sports car concept than the Rolls Royce concept, the new extension will be rejected. For example in its attempt to be a good sports car, the new extension compromised on the ciis of the "Rolls Royce concept" like "good engineering" and "smooth driving", the consumers might not perceive the new extension to be a "Rolls Royce" car and might reject the new extension.

How would the ciis and cdis influence brand evaluation? As we discussed above, since the ciis would always be passed over to the concept conjunction and the passing of cdis would vary from one case to another, the chances of rejection of a line extension are much greater if there is a mismatch on the ciis than if they are on the cdis.

A related issue is how would the dominant concept impact evaluation of a line extension ? We posit that since dominant concept receives a greater weight in the formation of a conjunction, the chances of rejection of a line extension are greater if there is a mismatch on the dominant concept than if there was a mismatch on the nondominant concept.

The proposed model of line extensions is different from the current models of brand extensions in the following ways:

(1) The current models of brand extensions consider only the first stage of matching (i.e between the existing brand concept and the new concept of the extension) whereas, our model takes one more matching process into account and that is between the expected new concept and the actual concept of the extension. This second stage is important because, even if there is a consistency between the extension concept and the old brand concept, if the actual performance of the extension (as determined by the "actual concept" of extension) is not as per the consumers' expectations (as reflected in the "expected concept" of the extension), the extension would be rejected.

(2) The model identifies various factors (discussed below) that could moderate the two processes of matching. The moderating factors could either increase or decrease the importance of the two matching processes described above. For example, an expert consumer will give more weight to these matching processes than a novice consumer because an expert would like to maintain the coherence of her / his product knowledge, whereas the novice might not be even able to define the concept (expected concept) of the new extension.

(3) This model also takes into account the difference in individual brand concepts positing that it might not be possible for all brands to launch a particular line extension. For example, it would be difficult for Rolls Royce to launch a sports car because of its conservative, exclusive, luxury brand concept, however, it would be relatively less difficult for Lexus to launch a sports car because of its modern, and liberal luxury car brand concept.

However, as our model deals with categorization constructs like brand concept, context independent information, context dependent information, category differentiation etc, and since the theory of categorization is equally applicable for brand extensions, this model would also be applicable to brand extensions.

MODERATING FACTORS

After discussing the process of "matching" between the line extension concept and the old brand concept, let us now look at some of the factors which could moderate these matching processes. These factors could be divided into two broad categories CC individually related and product related.

Individually Related Factors

(1) Expertise: Experts, with their detailed and complex knowledge structure of the target product category, should be able to see the relationships between different attributes more clearly than the novices (Alba and Hutchinson 1987). Hence, in the line extension context, experts are more likely to detect any inconsistency between the brand concept and the new line extension concept (whose knowledge of the category is not very deep and who normally process information via the peripheral routeC Sujan 1985; Alba and Hutchinson 1987). Since our model posits that the extensions will be rejected if the consumers detect any inconsistency between the old brand concept and the new extension concept, experts would accept lesser number of line extensions than novices.

(2) Model of Categorization Used: Smith and Medin (1981) divided the models of categorization into three basic types. The classical theory holds that all instances of a category share common properties that are necessary and sufficient conditions for defining the category. For example, according to this rule all brands of toothpastes should be white in color, have sweet taste, should freshen breath and should be in paste form. Any brand which does not satisfy any of the above mentioned criteria (a very stringent restriction) would not be categorized as toothpaste. The exemplar view says that a category may be represented by their individual exemplars and assignment of a new instance to a category is determined by whether the instance is sufficiently more similar to one or more of the category's known exemplars. For example, an individual might have two exemplars in the toothpaste product category, Crest and Colgate. To determine whether a new product s/he encounters is a toothpaste or not, s/he will compare the new product with either Colgate or Crest. If it is perceived to be quite similar to either of these brands, the new product would be categorized as toothpaste. The probabilistic or prototypical view says that there are no necessary and defining properties, rather categories or concepts are represented in terms of properties that are only characteristic or probable of class membership. Thus, membership in a category is graded such that "better" members have more properties than the "poorer" ones i.e the membership in a product category is evaluated on a continuum (rather than a yes or no condition) of how good or bad the new instance is of the brand concept (a more flexible restriction compared to the classical model). Hence, it is clear that the chances of a new instance being accepted as a member of the base category would be highest in individuals following the prototype model and lowest for the individuals following the classical model.

(3) In Group / Out Group Members: According to the theory of in-group vs out-group members (Park and Rothbart, 1982), in group members judge their own group members as more dissimilar to one another than they judge the out group members. This results mainly from the more extensive interpersonal contact which occurs between the in group members in relation to the contact between out group members. This contact aids the in group members in observing more detailed information about the in group members. Therefore, in the line extension context, brand users (in group members) would see more dissimilarities in the variants of their own brand than they would in the variants of the other brands. Thus, the line extensions of other brands would be perceived as more similar to the parent brand concept and, hence, accepted as legitimate line extensions of those brands. For their own brands, however, since they see more differentiation between the various variants, the chances of their accepting the line extensions of their own brands is relatively less.

(4) Affect Toward the Parent Brand: Isen (1984) have shown that people store not only cognitive information about a product category in their memory but, also, the affect or feelings associated with that cognitive information. This means that consumers will have not only stored the product information about the brand but also the affect (positive or negative feelings) associated with that information. Research on the effects of affect on categorization (Isen 1984, 1987) has shown that people who are feeling more happy (positive affect) sort objects into more inclusive or broader categories. Therefore, in case of line extensions, consumers with a more positive affect toward the parent brand might get into a positive affect state on seeing or hearing the brand name and might see more extensions similar to the parent brand, leading to the acceptance of a greater number of extensions as compared to consumers with lesser positive affect towards the parent brand.

Product Related Factors.

(1) Degree of Product Differentiation: Product differentiation refers to the variation in the different types of brands available in a product category (Mervis and Crisafi 1982). As discussed previously, objects within a category are more similar compared to objects between two categories. As a result, distinguishing between different types of objects within the same product category becomes more difficult. For example, tartar control toothpastes and fluoride toothpastes are the two different subcategories of the toothpaste category. Not only are the different members of tartar control toothpaste category quite similar, but, in general, they are quite similar to fluoride toothpastes. Because of this high "within category" similarity and "between category" similarity at the subcategory level , the differentiation at this level is low. Therefore, it is relatively difficult to discover the differentiating features between the tartar control and fluoride toothpastes. However, the subcategory level differentiation might vary from one product category to another. For example, in the "automobiles" category there are many types of cars : "sports cars"; "luxury cars"; "compacts"; "subcompacts"; "sedans"; "convertibles", etc and each of this type has further subtypes. Thus, it would be easier to discover the differentiating features among the various types in automobiles category than in the toothpaste category. Furthermore, the easier it is to discover the differentiating features, the more difficult it is to see the two variants of a brand as similar. Also, the more difficult it is to see the two variants as similar (and hence the degree of perceived fit between the old brand concept and the new line extension concept), more difficult it is to get the new extension accepted.

(2) The Position of the "Extension" Attribute in the Product Hierarchy: In the field of consumer behavior, the use of hierarchies to organize products and attributes has been proposed for some time. Howard (1977) views consumers as systematically grouping and distinguishing products into hierarchies on the basis of similarity. He posits that consumers make choices at different levels of these hierarchies or at different levels of abstraction. Category level choices occur at the most abstract level of a hierarchy, while the brand level choices occur at the more concrete level. His model also posits that consumers form hierarchies of attributes from the abstract to the concrete levels that correspond to their product hierarchies. For example, the product hierarchy for Crest toothpastes might look as shown in figure 2.

The horizontal extension (Crest Fluoride to Crest Tartar) involves a change in the concept of the brand, since according to our discussion on concept conjunction, the new line extension concept will be a conjunction of the two conceptsC Crest (till now only a fluoride toothpaste) and tartar control toothpastes. In the vertical extension C Crest Fluoride C regular flavor launching a 4.6 oz and 6.4 oz pack sizes, the dominant concept is still that of a tartar toothpastes since the flavors are included in the higher concept of tartar. The acceptance of horizontal extension will therefore be relatively difficult compared to the vertical line extensions which does not involve any concept modification (since the brand is still a tartar toothpaste).

FIGURE 2

A PRODUCT HIERARCHY OF CREST TOOTHPASTE

(3) Abstractness of the Current Positioning: According to Johnson (1984; 1987), choice among non-comparable alternatives (i.e among different product categories) occurs by comparing these product categories on some abstract attributes (more general aspects of a brand) which these product categories share . Therefore, as the choice moves from comparable to the non-comparable alternatives (i.e from "within category" to "between categories"), not only does the choice process involves comparing a greater number of product categories, but the comparison among those product categories is made in terms of more abstract attributes. This is in contrast to the more concrete attributes (which directly describe some specific aspects of a brand) used to compare alternatives in the same product category. Therefore, a given abstract attribute will encompass a greater number of product categories than a concrete attribute would. Extending this principle a step further, current positioning of a brand could center on a concrete attribute as in case of "fluoride" toothpaste, or it could be an abstract positioning as in the case of "a total dental care" toothpaste. Because of the specificity of the positioning in the concrete case, the chances of an extension being perceived as incompatible to the existing concept is much higher in the concrete case than in the abstract case, where the overall global positioning of "good toothpaste" can subsume almost all of the new attributes used by the new line extensions.

(4) Pervasiveness of the New Line Extension: The chances of acceptance of a line extension which is already being introduced by competitors will be much higher than that of a line extension which is being introduced for the first time in the market. This would occur because a more common extension would be perceived as more of a prototype (i.e as having feature shared by majority of the brands in the market) than a new line extension which is introduced for the first time in the market.

Having looked at the theoretical aspects, what are some of the problems associated with line extensions ? As competition in different product categories intensify, companies are resorting to indiscriminate launches of line extensions, to protect their flanks from ever growing competition.

This has led to the problems of brand concept confusability (when the brand tries to be everything to everyone), and cannibalization (where the extension takes share from existing variants rather than from competition.

CLASSIFICATION OF LINE EXTENSIONS

To make some sense out of the myriad of line extensions available in the market, a framework classifying the different types of line extensions is proposed (Table 1).

One way of classifying line extensions would be to see : (1) Which attributes (new vs old) are used to extend the brand and (2) To whom is the new line extension targeted (new vs old segment). This is related to brand objectives of increasing the market share either among the current users or nonusers of the brand. In doing this a 2X2 matrix is developed. The principle of concept conjunction will be used to hypothesize the nature of the new line extension and the implications of concept confusability; concept incompatibility; and cannibalization would be discussed for each type of extensions.

Cell 1-Same Attribute Same Segment

The first cell refers to those cases where the parent brand uses the same attribute to target the new line extension to the current users of the brand. The different types of line extensions which are possible in this cell are as follows:

- Pack sizes : This would refer to any variant/s of the current brand extending in new pack sizes. For example, Crest Fluoride could launch a new pack size of 10 Oz. Assuming that the concept of "new pack size" is not important in most cases, the existing concept of "Crest Fluoride" would be the dominant concept and hence the question of incompatibility does not arise. The new extension might cannibalize an existing variant if its per unit price is lower than those of other variants. Too many pack sizes for each variant might make the consumers spend more time and effort (and hence irritate) in his / her choice decision.

- Pack types : This would refer to any variant/s of the current brand extending itself in a new packaging. For example, Crest Sparkle launching a pump version of its toothpaste. For the same reasons as mentioned above for "pack size," Crest Sparkle would be the dominant concept and hence there is no question of any incompatibility between the two concepts. The arguments for "cannibalization" and "consumer irritability" would be the same as mentioned for "pack types."

TABLE 1

A CLASSIFACTORY FRAMEWORK OF LINE EXTENSIONS

- Flavors: One way to extend the usage of the brand among current consumers is to launch different flavors, provided each flavor is positioned as a different concept. For example, in category of fruit juices, orange juice could be positioned as a breakfast drink whereas, apple juice could be positioned as a "lunch" drink. The same arguments for concept dominance as in case of pack sizes and pack types would be applicable here. However, there could be an incompatibility if the new flavor is not compatible with the existing taste of the product (more sour when the general taste of the brand is sweet). Cannibalization and Consumer irritability might not arise if the new extension is not positioned distinctively.

Cell 2-Same Attribute New Segment

The second cell refers to those cases where the brand uses the same attribute to target it to the new users. The different types of line extensions which are possible in this cell are as follows:

- Pack sizes and Pack types and Flavors: (As in cell one) could also be used to target the new segment/s. Same arguments as those made for Cell 1 would be applicable here. However, since the new extension is targeted to new users, the problem of cannibalization does not arise.

- Price variants: This would refer to any variant/s of the brand extending to a different price level within the same attribute positioning. For example., Rolls Royce launching a cheaper version of its current brand to broaden its target market base. The dominant concept in this case would be that of old Rolls Royce (for a not very well known brand name like Pontiac, the market concept of "economy" cars might be the dominant concept). There could be incompatibility in terms of "image" of Rolls Royce (status symbol vs an "economy car"). Whereas, incompatibility in terms of "capability" might arise in case of an "economy" brand like Honda launching a "luxury car," and this could be one of the reason why Honda used a different brand name (Accura) to launch its luxury car. Since the new extension is targeted to new users, the question of cannibalization does not arise. Brand confusability might arise if the brand "positioned" at one end of a price continuum launches an extension at the other end of the continuum (e.g., a "premium" car launching an "economy" extension).

- Age variant: In this case, the brand uses the same attribute. For example, Johnson and Johnson used the attribute of "softness" to extend its baby shampoo to adults. The dominant concept in this case could be either the concept of "J & J " shampoo (since the "adult" consumers might strongly desire the presence of "J & J " shampoo attributes in their new shampoo) or it could be the concept of "adult shampoo," if the consumers are not sure about the performance capability of a baby shampoo for adults. To the extent that "J & J " is no longer exclusively associated with children products, there would be brand concept confusability.

Cell 3-New Attribute Same Segment

The third cell refers to those case where a new attribute is used by the parent brand to target the extension to the current consumers.

- Improvements : In this case, the brand improves one of its variants to target the current users. For example, Tide liquid launching Tide with bleach. The dominant concept in this case would be that of old Tide because the improvement in performance would be with respect to the old Tide and the consumers would not be able to form expectations of "laundry detergent with bleach" as the improvement would be launched for the first time. The problem of incompatibility might arise if the new extension does not perform as well as the parent brand on some important attributes. The chances of cannibalization are high unless the per unit price differential between the two is high. As the new variant replaces the old (in most extensions of such types), there is not much scope of concept confusability. However, consumer irritability might increase if this improvement is launched in all the different pack sizes and pack types of the brand currently available.

- Combination of Benefits: Here one or more new benefits are combined with one of the existing benefits (currently offered by the brand) to launch a new line extension. For example, Tylenol launched a new line extension Tylenol cold and flu, after combining the benefit of "flu" with the existing variant of Tylenol cold. Depending upon individual preferences, the dominant concept in this case could be either, Tylenol cold or flu remedies. There might be an incompatibility problem if the new attribute added is inconsistent with the old brand concept (very rare). There could be some cannibalization if the per unit price difference between the two variants is not very high. Confusability of such extensions is high because the brand is now no longer associated with one attribute.

Cell 4-New Attribute New Segment

The fourth cell refers to those cases where the brand uses a new attribute to target the extension to new users. The change in the brand concept (relative to other three cells) is maximum. The different types of line extensions which are possible in this cell are "improvements," "combinations," of the third cell. A horizontal line extension would also be classified in this category. Referring to the figure 2 above, the extension of Crest from fluoride to tartar control would be an example of a horizontal extension. The dominant concept in this case could either be the "core" Crest or it could be the tartar prototype, depending upon such factors as "is the consumer a "brand user" or a "segment user"? [if a user is a brand user (i.e., if s/he does not mind using any variant of Crest), then the concept of "Crest Fluoride" might be the dominant concept; and if the user is a "segment" user (the user prefers to use "tartar" toothpastes, but does not mind which tartar brand is s/he using), then the dominant concept will be that of "tartar" toothpaste]. Incompatibility could arise in few instances as in the case of a "tough on dirt" detergent extending into a detergent which is "soft on clothes". Confusion will arise since the core of the brand is getting diffused.

Our framework is superior to Kotler's (1991) framework (product line stretching) because whereas Kotler's framework could apply for line extensions and multibrand strategies (with different brand names), our framework is more specific in terms of it being relevant for line extensions only. Also Kotler's (1991) framework could account for only two (price and improvements) out of the eight types of line extensions as classified by our framework. Since our framework is more detailed, we expect it would be more useful to the practitioners.

Therefore, depending upon the brand objectives (increasing market share among new users vs current users), and other factors like brand concept confusability, incompatibility, consumer irritability, and cannibalization, product managers could use the above framework in deciding which variant/s should a brand extend into.

MANAGERIAL RECOMMENDATIONS

Having discussed the various issues involved in the introduction of line extensions, we now propose some managerial recommendations.

(1) The concepts of "concept conjunction" and "dominant concept" have implications for advertising strategy and product development . Both should emphasize "brand" features (brand name and brand attributes respectively) if brand concept is the dominant concept. Whereas, the "extension category" features should be emphasized if the market concept is the dominant concept. This could be done by advertising the new extension as how good a member the new extension is of the extension product category and by incorporating more features of the extension product category than those of the brand.

(2) The problem of confusability could be reduced if the manufacturers make the new extensions distinctive from the current brand concept C satisfying a distinct need. The example of a laundry detergent for "colored" clothes given above would be relevant here. The packaging could be made more distinctive in tune with the modified concept of the brand.

(3) To attract the nonusers, the brand should try to rectify its "negative image" among them.

(4) The managers should take advantage of moderating factors (wherever possible), to get a favorable response from the consumers. For example, by positioning the parent brand on an "abstract attribute," by keeping the advertising of the extension simple so as to attract the novices.

REFERENCES

Alba, Joseph W and J Wesley. Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, vol 13 (Mar), 411-454.

Barsalou, Lawrence W. and G H. Bower (1980), "A Priori Determinants of Concepts' Highly Accessible Information," Paper presented at the Annual Meeting of the American Psychologists Association, Montreal

Barsalou, Lawrence W. (1987), "The Instability of Graded Structure : Implications for Nature of Concepts" in Concepts and Conceptual Development, ed. Ulric Neisser. Cambridge : Cambridge University Press

Brooks L (1978), "Nonanalytic Concept Formation and Memory for Instances" in Cognition and Categorization, eds E. H Rosch and B. B Lloyd, Hillsdale, NJ : Lawrence Erlbaum Associates.

Bullmore, J (1984), "The Brand and its Image Revisited," International Journal of Advertising, 3, 235-38.

Fannin, Rebecca (1989), "Where are the New Brands," Marketing and Media Decisions, 24 : 7, 20-7.

Hampton, James, (1987), "Inheritance of Attributes in Natural Concept Conjunctions," Memory and Cognition, 15 : 1, 55-71.

Howard, John A (1977), Consumer Behavior : Applications of Theory, New York : Mcgraw-Hill.

Isen, A. M. and Kimberly A Daubman (1984), "The Influence of Affect on Categorization," Journal of Personality and Social Psychology, 47: 6, 1206-17.

Isen, A. M., K A. Daubman and G P. Nowicki (1987), "Positive Affect Facilitates Creative Problem Solving," Journal of Personality and Social Psychology, 52 :6, 1122-31.

Johnson, Michael D (1984), "Consumer Choice Strategies for Comparing Non-Comparable Alternatives," Journal of Consumer Research, 11 (Dec), 741-53.

Kotler Philip (1991), Marketing ManagementC Analysis, Planning, Implementation and Control Englewoods, NJ : Prentice Hall Inc.

Mervis , C. B and Maria A. Crisafi (1982), "Order of Acquistion of Subordinate, Basic, and Superordinate Level Categories, " Child Development, 53, 258-66.

Murphy, Gregory L and Douglas L. Medin (1985), "The Role of Theories in Conceptual Coherence," Psychological Review, 92 (July), 298-316.

Murray, Noel, H. Sujan, E. R Hirt, and M. Sujan (1990), "The Influence of Mood on Categorization : A Cognitive Flexibility Interpretation," Journal of Personality and Social Psychology, 59 (Sept), 298-316.

Newman, J W (1957), "New Insights, New Progress for Marketing," Harvard Business Review, Nov-Dec, 95-102.

Oliver, Richard L and Wayne S. DeSabro (1988), "Response Determinants in Satisfaction Judgements," Journal of Consumer Research, 14 (March), 495-507.

Park, B and M Rothbart (1982), "Perception of Out-group Homogeneity and Levels of Social Categorization : Memory for the Subordinate Attribute of In-group and Out-group Members," Journal of Personality and Social Psychology, 42, 1051-68.

Smith, E. E and D. L Medin (1981), Category and Concepts. Cambridge, MA : Harvard University Press.

----------------------------------------