Advertising that uses celebrity endorsers enjoys high popularity among brand managers (e. g. , Amos et al. , 2008). Each year, companies spend vast amounts of money to convince celebrities to endorse their products and brands (e. g. Jaiprakash, 2008; Klaus and Bailey, 2008; Lee and Thorson, 2008); for instance, Nike spent about $339 million on endorsements and their dissemination in advertising campaigns in 2004 (Thomaselli, 2004).
In the United States, approximately 25% of all televised commercials feature celebrities (Erdogan et al. , 2001); in Germany, approximately 12% of all advertising campaigns employ these endorsers (Ipsos Response, 2008). This communication strategy benefits from the widespread belief that celebrities positively influence the image of the advertised brands, such that a key outcome is a favourable effect on brand image (e. . , Erdogan et al. , 2001; TNS Sport, 2005).
Strong, unique, and favourable brand associations help companies differentiate their products from those of competitors and thus support a competitive advantage (Aaker, 1991; Krishnan, 1996). As a brand value driver, brand image also establishes an important foundation for a brand’s monetary value (Keller, 1993). Two prior narrative reviews (Erdogan, 1999; Kaikati, 1987) and one meta-analysis (Amos et al. , 2008) have attempted to summarise celebrity endorsement literature, yet no systematic investigation reveals success factors for brand image effects.
That is, in these reviews the contribution of celebrity endorsers to brand image has not been examined explicitly or separately from other measures of advertising effectiveness, such as attention, recall, or purchase intention. Furthermore, existing reviews are limited to literature pertaining to the selection of celebrity endorsers, including source effects such as credibility, attractiveness, fit or negative information. Erdogan’s (1999, p. 291) review, for example, “seeks to explore variables, which may be considered in any celebrity selection process by drawing together strands from various literature.
However, selection is only one facet of the development and execution of a celebrity campaign, and existing reviews cover only part of the body of extant research pertaining to celebrity endorsements. This study is the first to integrate literature from the broader field of celebrity endorsement literature. By adopting a communication process perspective, this research can derive implications regarding the characteristics of the celebrity (e. g. , perceived personal attributes), message (e. g. arguments used in the advertising copy), advertising channel and recipient (e. g. , personality traits that influence advertising receptiveness).
A systematic review of 36 studies identifies 24 drivers of brand image in prior literature. Furthermore, by arranging prior literature around distinct elements of the communication process, this review provides insights into which success factors have received strong empirical support and which have generated equivocal findings. Managers who hope to improve their brand? image can benefit from these insights regarding the entire development and execution of an endorsement campaign, rather than just the selection decision. Finally, this study identifies gaps in current research pertaining to celebrity endorsement and concludes by delineating directions for further research.
Celebrity endorsement and brand image A celebrity endorser is “any individual who enjoys public recognition and who uses this recognition on behalf of a consumer good by appearing with it in an advertisement” (McCracken, 1989, p. 310).
This definition explicitly encompasses celebrities who appear to have expertise or a long-term association with the manufacturer, but it excludes typical customer endorsements featuring noncelebrities. Because celebrities appear to be gaining increasing influence in society (Choi and Berger, 2010), marketing managers try to exploit the process of meaning transfer from an endorser to products or brands involved (McCracken, 1986, 1989). Associative learning theory details that celebrity endorsements influence brand image through a transfer of meaning from the endorser to the brand (Till, 1998).
Communication activities establish a pattern of connectivity between the image of the celebrity and the image of the brand. Both entities represent nodes in a cognitive network, whose connectivity can be modified according to experience. An image transfer occurs when an advertisement can establish contingency between the two entities (Till, 1998; Till et al. , 2008). Page 2 Managers pursue such connectivity with the goal of obtaining a favourable brand image outcome (TNS Sport, 2005). Brand image encompasses all perceptions of a brand, stored as brand associations in consumer memory (Keller, 1993).
Its importance mainly reflects its ability to provoke unique perceptions of the brand in competitive settings (Aaker, 1996; Kamins et al. , 1989). It also constitutes a meaningful brand value driver and thus influences the formation of brand equity (Faircloth et al. , 2001). Keller (1993) distinguishes among attitudes, attributes and benefits as types of brand associations, and Barich and Kotler (1991) regard brand attitude, belief and impression as ingredients of brand image. Both perspectives imply that improved attitudes toward the brand favourably influence brand image. In line with this argument, this review does not distinguish attitudes from image but instead adopts a holistic view of the impact of celebrity endorsements.
Celebrity Endorsement From a Communication Process Perspective
Communication, including marketing communication that employs celebrity endorsers, can be characterised as a process in which the sender conveys stimuli to influence the behaviour of others (Hovland et al. , 1953). Lasswell’s (1948) classical framework of persuasive communication differentiates the elements of the communication process according to his wellknown formula: “who says what in which channel to whom. The first element, „who? , refers to the message source or sender (Ajzen, 1992; Hovland et al. , 1953). Message sources can be impersonal (e. g. , advertisements), interpersonal (e. g. , friends, colleagues) or experiential (Keaveney and Parthasarathy, 2001).
In an endorsement context, the celebrity is an impersonal message source, so his or her characteristics, including perceived personal attributes like attractiveness or credibility, may determine endorsement success in terms of the favourable brand image outcome. The second, „says what? lement pertains to the message content itself, as conveyed in the communication process. Message characteristics relate to the execution and arrangement of the celebrity endorsement’s communication.
For example, marketing decisions involve the degree of endorsement strength and one- versus two-sided appeals (i. e. , only positive arguments about a product or claiming positive aspects on important determinants while simultaneously conceding minor negative aspects) (Kamins, 1989). The third element, „in which channel? , refers to the means the sender and receiver use to communicate.
The channel bridges any distance between senders and recipients of the message Page 3 (Ajzen, 1992); celebrity endorsement advertising usually employs mass media. The advertising channel characteristics, or properties of the media vehicles in which celebrity-endorsed advertising appears, could strengthen or weaken its effectiveness. Finally, the „to whom? element pertains to the audience or recipient of a message (Hovland et al. , 1953).
In an endorsement context, recipients differ in their degree of susceptibility to celebrityendorsed advertising. Therefore, recipient characteristics (e. g. individual personality traits, gender, and age) should influence the campaign with regard to the target audience that has the highest probability of providing a favourable brand image result.
To determine the state of the art of research from a communication process perspective, a literature search identified studies in the relevant domain. The included sources were empirical studies published in scientific journals that provided major implications regarding the use of celebrities as endorsers; studies that primarily focused on different types of (noncelebrity) endorsers (e. g. experts, endorsers with long-term associations with the manufacturer) were excluded. The systematic investigation encompassed the following journals: Advances in Consumer Research, European Journal of Marketing, International Journal of Advertising, International Journal of Research in Marketing, Journal of Advertising, Journal of Advertising Research, Journal of Consumer Marketing, Journal of Consumer Psychology, Journal of Consumer Research, Journal of Marketing, Journal of Marketing Research, Journal of Product and Brand Management, Journal of the Academy of Marketing Science, and Psychology and Marketing.
Perceived Personal Attributes
The roles of likeability, physical attractiveness, trustworthiness, expertise and credibility have been discussed widely in the celebrity endorsement domain. Likeability indicates affection toward a celebrity as a result of his or her physical appearance and behaviour (McCracken, 1989).
McGuire (1985) assumes that higher likeability is associated with greater effectiveness in terms of message persuasiveness. Kahle and Homer (1985) find no difference between a highly likeable and less likeable celebrity in brand image effects, but they reveal a significant interaction between likeability and involvement, such Page 5 Table 2 Impact of Celebrity Endorsement on Brand Image – Results of the Literature Analysis Success factor Brand image driver Source Focal effect Celebrity characteristics Perceived personal Likeability Kahle and Homer (1985) o attributes Ranjbarian et al. (2010) + Physical Kahle and Homer 1985) + attractiveness Kamins (1990) o Silvera and Austad (2004) Eisend and Langner (2010) Priester and Petty (2003) Siemens et al. (2008) Eisend and Langner (2010) Lafferty and Goldsmith (1999) Goldsmith et al. (2000) Lafferty et al. (2002) La Ferle and Choi (2005) Silvera and Austad (2004) Kamins (1990) Lee and Thorson (2008) Lee and Thorson (2008) Till et al. (2008) Koernig and Boyd (2009)
- under low involvement, n. s. implicitly tested:
- for product domain in which attractiveness is relevant
- for product domain in which attractiveness is relevant, n. .
- for product domain in which attractiveness is relevant
- for weak product-related argument strength
- in the long term
- for highly attractive endorsers
- for high corporate credibility, n. s.
In line with this argument, Silvera and Austad (2004) reveal that physical attractiveness associated with a product category positively influences brand image but physical attractiveness unrelated to the product category does not. Kamins (1990) neither observes a main effect of attractiveness nor finds a significant interaction between attractiveness and product category, though his results are directionally supportive of the assumption that attractiveness associated with the product category enhances brand image.
Eisend and Langner (2010) distinguish between immediate and delayed effects of attractiveness on brand image and find a positive impact of high attractiveness for both conditions. Trustworthiness and expertise both represent subdimensions of the more general credibility construct. Trustworthiness refers to the perceived willingness to make valid assertions; expertise entails the ability to make valid assertions (Hovland et al. , 1953; Sternthal et al. , 1978). For celebrity endorsements, the empirical results indicate strong evidence of a key influence of credibility and its subdimensions on brand image.
Priester and Petty (2003) find a positive impact of trustworthiness, as well as an interaction of trustworthiness with product-related argument strength. When weak product-related arguments mark the advertising copy, the influence of trustworthiness is more pronounced than it is for strong product-related arguments. Siemens et al. (2008) confirm the positive impact of expertise, whose impact on brand image is fully mediated by perceptions of endorser credibility. Eisend and Langner (2010) report no immediate but only a delayed positive effect of expertise on brand image.
In addition, expertise interacts with attractiveness so that the favourable impact of high expertise increases with higher levels of attractiveness. At the more general celebrity credibility level, several studies find evidence of a positive influence on brand image (Goldsmith et al. , 2000; La Ferle and Choi, 2005; Lafferty and Page 9 Goldsmith, 1999; Lafferty et al. , 2002). Lafferty and Goldsmith’s (1999) assumption of a more pronounced effect of celebrity credibility when corporate credibility is high receives no support though.
Social learning theory (Bandura, 1977) can help explain why perceived similarity between the celebrity and recipient facilitates brand image effects. That is, consumers may infer from their perceived similarity with the celebrity whether they can expect gratification from adopting attitudes or following the celebrity’s recommendation. However, Silvera and Austad’s (2004) empirical results regarding the impact of this perceived similarity are equivocal. Only in one of their two experiments do they find a positive and significant correlation between similarity and brand image.
Congruence With the Brand
At a general level, congruence describes a match between the endorser and the brand (Misra and Beatty, 1990). Arguments based in schema theory frequently indicate that congruence constitutes a prerequisite for the effectiveness of a celebrity endorsement. When an object seems to have high schema congruence, it should receive favourable evaluations, in line with a transfer of affect (Wansink and Ray, 1996). However, an analysis of research on congruence in the celebrity endorsement domain reveals that the notion has been conceptualised differently across studies.
To structure this extant body of research, this review classifies the concept of congruence into
- attractiveness-based (e. g. , matching a physically attractive celebrity with a beauty-related brand),
- expertise-based (e. g. , matching an athlete with a sports-related brand)
- image-based (e. g. , matching highly accessible celebrity associations with highly accessible brand associations) congruence.
For attractiveness-based congruence, Kamins (1990) and Lee and Thorson (2008) find no empirical support for a positive effect on brand image. Nor do their results support the assumption of an inverse U-shaped relationship between the level of congruence and brand image; that is, mild incongruence does not outperform high and low congruence.
In one of their data sets, a significant attractiveness-based congruence by involvement interaction emerges and implies that when involvement is high, the positive effect of congruence is more pronounced (Lee and Thorson, 2008). For expertise-based congruence, Till et al. (2008) find empirical support for a positive effect on brand image, but neither Lee and Thorson (2008) nor Koernig and Boyd (2009) can discern a significant impact.
Therefore, the contribution of expertise-based congruence to brand image remains equivocal. Page 10 Finally, regarding image-based congruence, existing literature reveals a more conclusive picture. Misra and Beatty (1990) and Kamins and Gupta (1994) find a positive effect of image-based congruence on brand image. Even though Kirmani and Shiv (1998) do not confirm a direct effect, they find a positive influence of image-based congruence in conditions of high involvement. Sengupta et al. (1997) investigate the robustness of brand image enhancements induced by celebrity endorsements over a longer period of time.
They thus empirically reveal a stable, positive, long-term effect on brand image when image-based congruence is high and involvement is low. Batra and Homer (2004) also investigate whether celebrities with highly accessible associations can reinforce equivalent brand image beliefs; their results confirm the positive impact of image-based congruence. However, they find a significant increase on the respective brand image beliefs only in one of the two cases they consider.
Negative information or information about remuneration constitute types of external nformation that affect perceptions of the celebrity. Negative information includes harmful news about a celebrity, which may become public during or after an advertising campaign. After an associative link has been established between a brand and its endorser, negative press about the celebrity may directly harm the image of the brand. Bailey (2007) and Edwards and La Ferle (2009) find significant negative effects on brand image when they present respondents with a cover story about a celebrity endorser being arrested for domestic violence and child abuse, respectively.
Till and Shimp (1998) find no significant impact of negative information on brand image, but their cover story, about a cyclist using steroids, clearly was weaker concerning negativity. Information about a high endorsement fee paid to the celebrity also could influence consumers? attribution about whether the celebrity actually likes the endorsed product or provides the endorsement only for financial reasons. However, in two studies (Cronley et al. , 1999; Van der Waldt et al. 2007), no significant differences emerge between a scenario in which they provide information about a high endorsement fee versus a scenario in which the endorser received no fee.
The use of celebrity endorsers in marketing enjoys high popularity, largely because they may exert a positive impact on brand image. Noting the importance of brand image as a critical intangible asset, with a fundamental impact on brand equity, this systematic literature review pinpoints 24 brand image drivers of celebrity endorsements, arranged around four distinct elements of the communication process.
Thus, the framework includes not only celebrity selection issues but also the design of the advertising message, the appropriate media vehicle, and the target group. Extant empirical studies reveal certain success factors that seem somewhat ambiguous, whereas others receive relatively strong empirical support.
Therefore, this review can help managers derive practical implications for the development and execution of their celebrity endorsement campaigns: They should focus particularly on the well-documented success factors. Credibility, physical attractiveness relevant to the product domain, image-based congruence, twosided appeals and lower elaboration likelihood all positively influence brand image. When selecting an appropriate celebrity for an endorsement, brand managers must take care to achieve high credibility.
Extant research consistently verifies the importance of credibility and its subdimensions, trustworthiness and expertise. Furthermore, most studies show that physical attractiveness has a positive impact on brand image, though only if that attractiveness is relevant to the product domain (e. g. , a physically attractive celebrity for a make-up brand). This evaluation of prior literature also reveals that image-based congruence (as opposed to attractiveness- or expertise-based) constitutes the most important congruence dimension in a Page 15 celebrity endorsement context and has a positive impact on brand image.
The advertising message also should include two-sided appeals in celebrity endorsements, because the empirical findings consistently verify their effectiveness. Finally, lower elaboration likelihood among recipients positively influences the effectiveness of celebrity endorsers for creating a favourable brand image outcome. Both high arousal and low involvement promote low elaboration likelihood and can enhance the impact on brand image. Rather ambiguous findings relate to negative information, multiple brand endorsements and endorsement strength.
In the context of negative information, Bailey (2007) and Edwards and La Ferle (2009) find that really negative information about the celebrity impairs brand image, but Till and Shimp (1998) cannot confirm this effect using a weaker cover story. Thus, the severity of the negative information and the person involved (e. g. , the celebrity versus friends of the celebrity) seem worthwhile candidates for further investigation. Mowen and Brown (1981) report a negative effect on brand image when a celebrity endorses numerous brands simultaneously, but Tripp et al. (1994) cannot confirm this result.
The omnipresence of multiple brand endorsement in practice and the high cost of exclusive contracts with celebrities therefore implies the need for further research to disentangle the boundary conditions for such negative effects. Moreover, research from the related field of brand alliances (e. g. , Simonin and Ruth, 1998) suggests that even positive effects might emerge in specific conditions (e. g. , a less familiar brand might benefit from more familiar brands in a multiple brand endorsement context).
In addition, Martin et al. (2008) report a positive effect of stronger endorsements, but Cronley et al. 1999) find no significant differences between stronger and weaker endorsements. In practice, weak and strong endorsements may represent two extremes. On the one hand, weak endorsements might not express the celebrity’s conviction about the product adequately and thus fail to achieve the greatest persuasive effect. On the other hand, strong endorsements might evoke psychological reactance, in response to perceived restrictions on freedom and control (e. g. , Brehm 1966; Brehm and Brehm, 1981). Whether moderate endorsement strength might outperform both weak and strong endorsements should be investigated in ongoing research.
Most studies in this body of research have focused exclusively on celebrity characteristics, to the detriment of in-depth investigations of other elements of the communication process (i. e. , message, advertising channel and recipients). Regarding message characteristics, a conceptual distinction of four endorsement modes suggests celebrities might endorse brands in an explicit mode (“I endorse this brand”), implicit mode (“I use this brand”), imperative mode (“You should Page 16 use this brand”) or co-present mode (i. e. , celebrity and brand depicted simultaneously without further explanation) (McCracken, 1989).
Yet, research has not considered these different endorsement modes empirically to determine their effectiveness. Regarding advertising channel characteristics, research is scant (Freiden, 1982). This gap seems surprising against the background of prior research that shows media selection exerts a strong impact on advertising success (e. g. , Batra et al. , 1996; Korgaonkar et al. , 1984). Furthermore, misplaced celebrity advertising might compromise its effectiveness (Seno and Lukas, 2007).
Therefore, an investigation of the impact of advertising channel characteristics on elebrity endorsements seems indispensable. With regard, finally, to recipient characteristics, additional research is necessary to enable marketers to tailor their endorsements to target groups that are more susceptible to celebrity advertising. Personality traits other than those addressed by prior research might influence this susceptibility, which would grant consumer behaviour researchers interesting insights as well. In addition, cross-cultural comparisons of acceptance of celebrity endorsement seem desirable.
For example, U. S. culture might promote a rather strong attachment to celebrities, such that U.S. consumers would be more amenable to marketing communications that use celebrity endorsers (McCracken, 1989); Scandinavian cultures appear more reluctant (Avant and Knutsen, 1993). Reconsidering celebrity endorsement from a communication process perspective thus highlights the diversity of elements that managers should take into account when designing their campaigns. In practice, this insight seems particularly meaningful, because most current attention focuses solely on the selection of an appropriate celebrity (Miciak and Shanklin, 1994).