How positioning strategies affect co-branding outcomes

Abstract Co-branding is a widely applied strategy, with research indicating differential benefits to the parent brands. Past studies suggest the source of these differences may be due to the partners’ relative market position, and characteristics such as brand familiarity, brand equity and proximity to the consumer have been explored. However, the role of brand positioning has received little attention in the context of co-branding. The current study attempts to address this gap, considering the positioning of a brand and the impact of a co-branding strategy on customer perceptions. Using the Blankson and Kalafatis positioning typology, we explore the impact of co-branding on the parent brand perceptions from a hedonic vs. functional (utilitarian) focus. The results suggest that for hedonically oriented positioning strategies, fit between the brands is more important than fit between the product categories in driving positive brand perceptions. For a functionally oriented positioning strategy, the reverse holds, with product fit a more important factor than brand fit in driving post-alliance perceptions.


Introduction
The popularity of co-branding as a brand management strategy is a growing area of academic focus. Co-branding combines the competencies and reputations of two partnering brands to create a new product (e.g. Faems, Van Looy, & Debackere, 2005;Park, Jun, & Shocker, 1996). Past research has identified important determinants of consumer attitudes to co-brands (also referred to as brand alliances), such as the familiarity (Levin & Levin, 2000), perceived quality (Rao, Qu, & Ruekert, 1999;Rao & Ruekert, 1994) and relative brand equity (Washburn, Till, & Priluck, 2000) of the partnering brands. In particular, the extent to which they are perceived to "fit", i.e. be congruent in terms of brand perceptions and product categories, has been found to influence consumer attitudes to the co-brand (Helmig, Huber, & Leeflang, 2007;Lafferty, Goldsmith, & Hult, 2004;Simonin & Ruth, 1998). One would expect the similarity between the parent brands to increase the perceived fit (Simonin & Ruth, 1998), but moderate incongruity has been found to foster favourable evaluations as well (Meyers-Levy & Tybout, 1989). The same studies also reveal that co-branding can subsequently influence perceptions of the original partner brands (Baumgarth, 2004;Simonin & Ruth, 1998;Washburn, Till, & Priluck, 2004). The nature of these "spillover" effects remains an area of debate; some studies have found only positive effects (e.g. Washburn et al., 2000), while others demonstrated negative effects under certain conditions (Till & Shimp, 1998;Votola & Unnava, 2006). A limitation of research to date is that characteristics of the parent brands have not been widely explored. Most studies have addressed [fictitious] co-brands based on familiarity (e.g. Baumgarth, 2004;Simonin & Ruth, 1998), equity levels (e.g. Besharat, 2010;Washburn et al., 2004), vertical integration (Desai & Keller, 2002) or country of origin (Lee, Lee, & Lee, 2013). However, real-life examples of brand alliances (e.g. Boodles and the Royal Ballet; Google & Luxottica; Alexander Wang; and H&M) may not fit neatly into such schemes. A few papers have considered additional parent brand attributes. A study of fictitious co-brands by James, Lyman, and Foreman (2006) suggests that the brand personality of the parent brands can influence perceived fit and subsequent attitudes towards an alliance. Johan Lanseng and Erling Olsen (2012) distinguished between functional brands (based around performance or utilitarian attributes) and expressive brands (where associations are built around the consumer's self-image). They found that the importance of product and brand fit on attitudes towards the alliance varied depending on whether the parent brands were functional or expressive. Singh, Kalafatis, and Ledden (2014) found consumer perceptions of positioning strategies of parent brands to be a significant factor influencing their view of the co-brand.
The current study builds on these papers in two ways. We move beyond the functional vs. expressive definition of Johan Lanseng and Erling Olsen (2012) to consider a more complete positioning taxonomy of the parent brands, in line with Singh et al. (2014). Rather than looking at attitudes towards the co-brand, we focus on the spillover effects on the parents, exploring how the alliance influences perceptions of the parent brands with similar/dissimilar positioning. We draw on theories of signalling and attitude formation to explain the mechanisms by which individuals adjust their brand perceptions in light of co-branding activity. The paper is organised as follows. First, we review the literature on positioning strategies and co-branding. We then link these literature streams to propose hypotheses and test these in a study of 168 consumers. We conclude with some managerial and theoretical implications and suggestions for further research.

Brand positioning
An extensive body of literature documents that positioning is a central success factor to a brand's performance and firm's competitive advantage (Hooley, Greenley, Fahy, & Cadogan, 2001;Porter, 1996). To avoid confusion about the meaning of the positioning concept, it is important to distinguish between brand positioning and strategic market positioning (Fuchs & Diamantopoulos, 2010). Strategic market positioning refers to the competitive market standing of a firm against its competitors, where firms seek ways to deploy firm-specific resources to build positional advantages (Day & Wensley, 1988). Brand positioning, on the other hand, focuses on perceptions of consumers about a firm's products or brands (Crawford, 1985). Many studies have highlighted the importance of brand positioning as a mechanism to deliver added value (Hooley, Broderick, & Möller, 1998;Knox, 2004). A consumer forms perceptions of a brand based on a range of factors including the communicated positioning, previous experience, word of mouth, personal goals and values, usage situations and comparisons to competitor brands (Ellson, 2004;Friedmann & Lessig, 1987;Ries & Trout, 1986).

Co-branding
The theoretical foundations of co-branding's impact on consumers come from theories on signalling (Rao & Ruekert, 1994;Rao et al., 1999) and attitude formation (e.g. Anderson, 1981;Hillyer & Tikoo, 1995). Previous co-branding studies demonstrate that partner brands are beneficial if they can signal high-quality cues that transfer to the host brand or provide information on product attributes that benefits the alliance (Rao & Ruekert, 1994). Attitude formation frameworks such as the elaboration likelihood model (Petty & Cacioppo, 1986) and context effects (Lynch, Chakravarti, & Mitra, 1991) suggest that pre-existing, salient and accessible brand attitudes and close, observable cues can both influence consumer perceptions of a brand partnership. Further, Berthon, Pitt, and Campbell (2009) argue that consumers' existing knowledge plays an important role in how they interpret brand communications, such that a brand can have multiple meanings in the market depending on the stakeholder. Research suggests consumers assimilate visual cues, such as branding and product form, into their judgments of product performance even when presented with conflicting feature information from an objective source (Hoegg & Alba, 2011). This reliance on pre-existing attitudes and external cues makes the perceived fit between the two parent brands an important factor influencing customer perceptions.
When two brands are combined in a brand alliance, there are several possible bases for fit: category fit, brand associations, culture, product usage, self-representation and consumer goals, (Loken, Barsoaloou, & Joiner, 2008;Martin & Stewart, 2001). Most research has focused on "product fit" and "brand fit" as the primary variables of interest. Defined as the extent to which consumers perceive two product categories to be complementary and well matched, product fit has generally been shown to have the stronger positive relationship with consumer attitudes towards the co-branded product (Simonin & Ruth, 1998;Washburn et al., 2000). Most research has considered brand fit to be the general association that consumers draw about the perceived congruence of the partnering brands; for instance, a partnership between BMW and Rolex is likely to elicit high brand fit since both brands are associated with high quality and good taste (Baumgarth, 2004;Lafferty et al., 2004;Simonin & Ruth, 1998). Empirical findings for brand fit have been somewhat inconsistent (e.g. Baumgarth, 2004). Like its product counterpart, brand fit is often operationalised as a two-or threeitems scale capturing complementarity, consistency and sensibleness.
Arguing that these simple measures do not fully capture the range of information consumers use to form perceptions of a co-brand, recent studies have considered additional measures of fit. Bouten, Snelders, and Hultink (2011) considered the match between the new product's category and the parents, finding the fit between the new product and the parent brand associations (but not their product categories) to be a more important influence on consumer attitudes than the fit between the parents. Lee et al. (2013) considered the fit between the parent brands country of origin, finding that a partner with a positively viewed country of origin influenced perceptions. Xiao and Lee (2014) introduced brand identity fit (the perceived congruence/incongruence between two brands' cultural meanings) as an important factor influencing co-branding success and recommend that future studies combine brand identity with other types of fit. In summary, the literature suggests that the bases of fit extend beyond a simple heuristic of general complementarity and/or similarity, and are related to the original consumer value associations for the parent brands. The additional fit metrics can be viewed as proxies for identifying additional characteristics of the parents (country of origin, brand personality and brand identity). Only three papers (Johan Lanseng & Erling Olsen, 2012;Singh et al., 2014; van der Lans, Van den Bergh, & Dieleman, 2014) have directly considered an expanded set of parent brand characteristics. Johan Lanseng and Erling Olsen (2012) distinguish between brands built around functional consumer needs (i.e. the associated attributes are objective, instrumental or utilitarian) and brands built around expressive needs (where the brand fulfils a consumer's need for self-enhancement, group membership or ego identification). They found product fit to be important for functional and mixed-brand concept-based alliances, but not for an alliance between two expressive brands. Singh et al. (2014) found consumer perceptions of parent positioning strategies of the parent brands influenced co-branded products; however, the differing positioning strategies were not compared. van der Lans et al. (2014) found that conceptual coherence in brand personality profiles between parent brands predicts attitudes towards a co-brand in some cases.
We now briefly consider how a brand alliance affects perceptions of the parent brands. Research has shown that these spillover effects are often asymmetrical, with one brand benefiting more from the alliance than its partner (see Helmig et al., 2007 for a review). Pre-alliance brand perceptions shown to influence relative post-exposure attitudes include the level of brand familiarity (Kumar, 2005;Simonin & Ruth, 1998), perceived quality (Rao et al., 1999;Voss & Gammoh, 2004), brand equity and reputation (Vaidyanathan & Aggarwal, 2000;Washburn et al., 2000) and loyalty (Swaminathan, Reddy, & Dommer, 2012). Although most studies have shown positive (if varying) spillover effects for both partners (e.g. Baumgarth, 2004;Simonin & Ruth, 1998;Washburn et al., 2000), a few suggest darker outcomes. Swaminathan et al. (2012) found positive spillover effects occurred only where the perceived brand fit was relatively high. In a study of 10 different partnership scenarios with real brands, Lebar et al. (2005) found that co-branding led to reduced brand esteem scores on average and recommended that high-esteem brands be wary of embarking on a brand alliance. Votola and Unnava (2006) identified conditions where negative spillover effects might occur.
Overall, these findings suggest that the impact of co-branding on a parent brand's reputation could be positive or negative and is influenced by the brand's characteristics as well as fit with the partner. The impact of a co-branding strategy on brands with different positioning is not known. Only three papers have considered brand positioning or personality perceptions in the context of co-branding. James et al. (2006) found that a brand alliance is more positively perceived when the alliance partners have similar brand personalities. Singh et al. (2014) found consumer perceptions of positioning strategies of partner brands to be significant determinants of the positioning perceptions of a co-brand and found some evidence for spillover effects on partner brand positioning. van der Lans et al. (2014) found that conceptual coherence in brand personality profiles between parent brands predicts attitudes towards the co-brand. These papers suggest that brands' positioning strategies have a direct influence on the perceptions of their co-brand offering, but do not consider feedback effects on the partners. In light of this gap, we now look at how a brand's positioning and choice of partner (in terms of fit) impact brand perceptions post alliance.

Conceptual framework
In looking at the spillover effects of a co-branding strategy on the parent brands, we draw on signalling theory and research on co-brands and brand extensions to develop the hypotheses.
Research has demonstrated that pre-existing attitudes to brand partners remain relatively consistent over time (Baumgarth, 2004;Simonin & Ruth, 1998). It is therefore important to control this for the highly predictive relationship between attitudes towards a brand. This is echoed by findings from the brand extension literature, showing that initial brand image conditions the final brand attitudes (Lee & Ulgado, 1993;Martinez & Pina, 2003). Although this relationship is already well researched, its presence is a necessary precursor before considering the fit relationships of interest. Therefore, we propose: H1: Customer perception of a partner brand prior to a co-branding strategy is positively related to post-exposure perceptions of the partner's positioning after the co-branding activity.
Signalling theory suggests brands serve as signals of product quality (Erdem & Swait, 1998), particularly where attributes are not easily observed (Rao et al., 1999). Where product attributes can be clearly assessed, complementarity and/or reinforcement of specific attributes or benefits can occur (Washburn et al., 2004). Analogous to how partnering firms in a traditional strategic alliance combine complementary resources (Makri, Hitt, & Lane, 2010), co-brands that bring together complementary functional attributes (such as a camera and a phone) strengthen the performance of the joint offering. Where partner brands offer similar functional attributes (i.e. they share similar strengths and weaknesses), there is limited benefit to the relationship (Newmeyer, Venkatesh, & Chatterjee, 2014). Studies have shown that new functional attributes can positively affect a co-branded product (Desai & Keller, 2002), as well as the parent brands (Park et al., 1996;Rodrigue & Biswas, 2004). Radighieri, John Mariadoss, Grégoire, and Johnson (2014) found spillover effects were greater in functionally oriented product categories (mobile phones and shaving cream) than in a sensory-based one (cookies). These findings suggest that product fit is particularly important for functionally positioned brands. Research on cognitive processing supports this view (Sujan, 1985). In evaluating product fit, consumers assess whether the two parents can complement or substitute each other (Aaker & Keller, 1990), possess the same physical product characteristics or perform the same practical functions (Park, Milberg, & Lawson, 1991). This process is likely to involve an attribute-by-attribute or piecemeal comparison of two product categories, where attributes are evaluated individually and overall evaluations are formed by combining these (Fiske & Pavelchak, 1986). This evaluative approach will be inherently easier for consumers considering functionally positioned brands, where their associations are more likely to be discrete and based around objective, performance-related attributes. In contrast, brand fit involves more abstract associations between two brands. Brand concepts are category structures in consumers' minds and consist of attribute interrelations, product beliefs and emotions developed through experience (Cohen & Basu, 1987;Fiske & Taylor, 1991). Brand fit is evaluated by matching these abstract, superordinate associations. Following this reasoning, we suggest that product fit between the functionally positioned parent brands plays a greater role than brand fit on co-brand perceptions, and this is echoed in the postalliance attitudes towards the brands (Simonin & Ruth, 1998). Therefore, we hypothesise: H2: For partner brand positioning that relies more on functional benefits (e.g. concrete, performance-related benefits), product fit will have a greater impact than brand fit on postexposure attitudes towards the brand.
Turning to hedonically positioned brands, the literature suggests that co-brands are more positively viewed if the attributes that capture the sensory/emotional feelings and personality traits of the partner brands are perceived to be similar (Broniarczyk & Alba, 1994;Newmeyer et al., 2014). Research on sensory cues has shown that dissimilarities can lead to varying preferences (Biswas, Labrecque, Lehmann, & Markos, 2014). Congruence of hedonic attributes should result in more cohesiveness and consistency in the brand image of the partners and therefore higher brand fit. Studies of brand extensions have found brand fit to be a more important factor than product fit for hedonically positioned brands (Bhat & Reddy, 2001;Park et al., 1991). Logic suggests that brand fit is a more important factor than product fit for brand alliances, where parent brand positioning is based around hedonic attributes, but the empirical evidence is somewhat mixed. Lau and Phau (2007) find brand fit for prestige brand alliances did not subsequently influence parent brand perceptions. In contrast, van der Lans et al. (2014) find that conceptual coherence in brand personality profiles predicts attitudes towards a brand alliance. Fleck, Michel, and Gatignon (2012) argue that there are hedonic and utilitarian components of attitude towards a co-branded product. The utilitarian component is built on the functional performance and perceived credibility of the new product, and the hedonic component on the novelty, exclusiveness and sensory perceptions, with each reflecting underlying brand positioning. Building on this notion, we hypothesise: H3: For partner brand positioning that relies on hedonic attributes or benefits, brand fit will have a greater impact than product fit on post-exposure perceptions.

Method
In line with most of the co-branding research to date, this study took the form of a scenario-based experimental design (e.g. Baumgarth, 2004;Simonin & Ruth, 1998), using hypothetical co-brands created from existing brands. Perceptions of brand positioning strategies were obtained using the eight-dimension typology developed by Blankson and Kalafatis (2004). This typology was selected as it is based on customer-perceived positioning strategies rather than managerially derived strategies. The positioning strategies are defined as "Top of Range", "Value for Money", "Attractiveness", "Service", "Country of Origin", "the Brand Name", "Reliability" and "Selectivity". See Table 1 for a description of the positioning typology.
Although each positioning strategy appeared clearly functional or hedonic in orientation to the authors, the strategies were also reviewed by 12 marketing professionals to confirm orientation. The professionals were former students contacted via LinkedIn and who held positions as a product manager (6), customer service manager (3), marketing manager (2) or entrepreneur (1). Through independent assessments, the professionals determined that three of the positioning strategies, Value for Money, Service and Reliability, are based on functional attributes and benefits and the remaining five-Top of Range, Country of Origin, Brand Name, Attractiveness and Selectivity-are hedonically focused positioning strategies. The rationale in Singh et al. (2014) guided the selection of brand alliances.
The professionals were then presented with the Blankson and Kalafatis (2004) descriptors of each strategy and asked to indicate how strongly each of the 16 presented brands was associated with the strategy, using a five-point scale anchored at "very weak" and "very strong". From this, the eight brands used in the study were selected, based on clear and different positioning strategies to ensure the effects of the brand alliances were prominent and the results easy to interpret. The four brand pairings created were Panadol (pain relief) and Ferrari (cars), Samsonite (luggage) and Michelin (tyres), Hugo Boss (clothing and accessories) and Emporio Armani (clothing and accessories) and British Airways (airline) and Facebook (social media site). A final step of the pre-test saw product fit and brand fit (reflective constructs) for the brand pairings measured using two-item semantic differential scale (is/is not consistent; is/is not complementary) on a five-point Likert scale anchored at "strongly disagree" and "strongly agree" (Simonin & Ruth, 1998 Following the pilot test, a questionnaire was developed as the survey instrument based on the approach in Simonin and Ruth (1998). Similar to other co-branding research, the main study respondents were undergraduate and postgraduate students from a UK university and each respondent was randomly assigned to one of the four brand alliances. The respondents were first presented with each of the brand allies and requested to indicate their perceptions of their market positioning (pre-alliance) for each dimension of the Blankson and Kalafatis typology (Blankson & Kalafatis, 2004); questions related to brand and product fit followed before each of the brand alliances were presented in the form of pictorial representations (an advertisement). Questions about attitudes towards each brand alliance and post-alliance questions about market position completed the survey. A total of 168 usable replies were obtained; the number of replies was comparable between the four co-brands (Tables 2 and 3).

Analysis and results
Given the exploratory and predictive nature of the study (as opposed to theory building), the data were subjected to partial least squares (Chin & Newsted, 1999;Haenlein & Kaplan, 2004;Tenenhaus, Vinzi, Chatelin, & Lauro, 2005) using the software SmartPLS 3 (Ringle, Wende, & Becker, 2014) with a bootstrap analysis of 500 samples. PLS analysis was adopted because of its advantages over covariance-based modelling (assumptions of normality and minimum sample size requirements). The composite reliability of brand fit and product fit were .898 and .939, respectively, while the corresponding AVE values were .814 and .884. These indices are above the suggested benchmarks of .70 for the former and .50 for the latter (Fornell & Larcker, 1981), and thus confirm the psychometric properties of the latent constructs.  The results related to the research hypotheses are presented in Table 4. Before debating each hypothesis, it is worth noting that satisfactory fit for all positioning strategies; variance explained (R 2 ) greater than .30; and predictive relevance (Q 2 ) positive. Hypothesis 1 predicts positive pre-and post-alliance relationship of perceptions of market positioning. The evidence in Table 2 confirms this hypothesis for all the positioning strategies.
We propose that for functional positioning strategies, product fit is the main determinant of postalliance perceptions of market positioning (H 2 ) . This hypothesis is supported for Value for Money (both significant with β product fit = .267 > β brand fit = .192) and Service (β product fit = .227 and β brand fit is not significant), but not for Reliability (both significant with β product fit = .331 = β brand fit = .330).
For hedonic positioning strategies, brand fit is hypothesised to be the stronger determinant of postalliance perceptions of market positioning (H 3 ). Of the five hedonic positioning types, we find support for three: Brand Name (β brand fit = .136 > β product fit = .009), Attractive (β brand fit = .272 > β product fit = .057) and Country of Origin (β brand fit = .217 > β product fit = .046). However, for Top of the Range, product fit was higher than brand fit (β product fit = .228 > β brand fit = .153). For Selectivity, neither brand nor product fit was significant.

Discussion and managerial implications
This study aims to explore how the brand positioning of partner brands influences co-branding outcomes. We find that post-alliance brand positioning beliefs are strongly associated with prior beliefs and these have a slightly stronger influence for hedonic positioning than functional positioning. We also find product fit and brand fit to have different effects on post-alliance perceptions, depending on the original positioning strategies. Where firms have functional positioning strategies built on value or service attributes, product fit is more important than brand fit as hypothesised (significantly so for Service positioning). However, for Reliability positioning, brand fit and product fit are equally weighted. One possible explanation for this finding is that as Reliability is somewhat less "concrete", or visible to the consumer than other functionally oriented positioning attributes (such as specific features, price, value metrics or a service offering), complementarity in the features between the product categories is less easy to assess. Under these circumstances, the consumer may focus on the brand fit in equal measure to the product fit as a credibility signal.
Looking at hedonic positioning dimensions, Brand Name, Attractiveness and Country of Origin were influenced by the brand fit as hypothesised, with product fit having no significant influence. These positioning strategies contain strong symbolic elements, fulfilling a consumer's need for selfexpression and prestige. Here, the fit between partner brands in terms of image, associations and equity is more important than the perceived fit between parent products in determining how consumers view the brands partnership activities and the brands themselves. We have no theoretically grounded explanation for the lack of significant results for Selectivity and can only speculate as to its cause. It is possible that it is a measurement issue with the brand pairings that did not resonate with the respondents. Alternatively, the discriminatory and principled nature of selectivity positioning may encourage individuals to disregard any partners. Studies of prestige brand suggest that co-branding can present a risk to the prestige-oriented brand. Where positioning relies on giving an impression of exclusivity, introducing any type of brand extension may make the brand seem less exclusive and commonplace (Kim, Lavack, & Smith, 2001;Lye, Venkateswarlu, & Barrett, 2001;Sharp, 1993;Wang, Soesilo, Zhang, & Di Benedetto, 2012). Geylani, Inman, and Hofstede (2008) considered the spillover effects of co-branding activities on the image of partner brands and identified conditions in which a brand's image can be reinforced by the partner brand or impaired through increased uncertainty introduced via dissimilar brand images. They found that it is not always in a brand's interest to choose the highest performing partner brand. Our findings provide contextual support for this view, indicating that brand positioning plays an important role in determining what type of partner is most beneficial. A high-performing service brand might not necessarily benefit from a partner with strong Country of Origin positioning, regardless of the partner's performance level. Likewise, a brand with strong hedonic positioning might benefit more from partnership with another hedonically positioned brand rather than one reliant on concrete features or attribute positioning, even if the hedonic partner is lesser known. However, Geylani and his co-authors also found co-branding for image reinforcement may not be a viable strategy if the brand already has a very strong reputation. The finding that Top of Range (ToR) positioning is influenced by product fit more than brand fit (although the difference is not significant) may relate to this point. For ToR positioning, the fit between the brands and product categories is of similar importance. It may be that ToR is a more "all encompassing" strategy driven by both brand and product dimensions. This argument might also apply to Value for Money positioning; future research could explore these strategies further, perhaps with another set of brands that match this positioning.
If these results are generalisable, a key managerial takeaway is that managers should carefully consider potential trade-offs between complementarity of features and of fit in brand associations with impending partners. Where brand positioning is based on functional attributes, partners should be carefully assessed for feature complementarity, in addition to general brand equity considerations. Where brand positioning is hedonic in nature, managers should give weight to the fit in brand associations over product fit considerations. Finally, where a brand has prestige positioning, co-branding presents a greater risk, and both brand fit and product fit should be carefully considered.

Limitations and directions for future research
Only a few brands and product categories were used in this study; therefore, generalisations must be made with caution. Special attention must be paid to replicating the research for different brand pairings to clarify the relationships that received mixed support in the study. Data could be collected to test the model under different brand knowledge, product category, familiarity and consumer involvement conditions. This study called for respondents to reach to a hypothetical pairing of two real brands. Only highequity brands were considered and familiarity (a control variable) was high. In situations where the brands have different levels of brand equity, we might expect consumers to revise their prior beliefs more. With novel pairings, consumers' lack of familiarity with the attributes inherent in the product category may mean that product fit plays a less important role in spillover effects than it did in this study. There is scope for further research in this area to enhance and complement the data collected in this study.