INFLUENCER IMPACT ON BRAND AWARENESS: A MIXED METHOD SURVEY IN THE GERMAN FASHION SEGMENT

This study evaluates the impact of influencer marketing on consumers. Although online influencers are long established in German B2C markets, research in the field is scarce. Marketing managers are unsure how influencer marketing strategies should be drafted and which influencers should be recruited in order to maximize their marketing efforts at the consumer level. Based on a review of previous theoretical and empirical literature, a mixed method empirical study comprising semistructured interviews with three German fashion influencers and a consumer survey (N = 385) among the followers of these influencers are conducted to evaluate the impact of influencer activity on consumer followership, brand awareness and purchase intention. Results indicate that the intensity of influencer network involvement, intrinsic influencer motivation, authenticity of communication style and the real-life character of influencer posts increase the dependent variables. Hence, the empirical study has contributed to identify the most important determiners for the German B2C fashion influencer segment. Marketing managers are strongly advised to select influencers adequately and organize their marketing strategy in correspondence with their company and products.


INTRODUCTION
The use of the internet as an information and shopping medium by consumers has increased rapidly in recent years (Franklin, 2008). Ever more, buyers use virtual channels to find out about consumption options and products. Among others, social media are very popular.
In 2017, an average US American consumer spent around 1.72 hours on social media daily. Generally, 74% of buyers are influenced by social media in their shopping behavior (Woods, 2016). In this context, influencer marketing makes use of gatekeepers in virtual social networks for advertising. Businesses denying to actively engage in these new sales strategies risk a loss of competitiveness and melt-off of established outlets (Carbonaro and Votava, 2005). However, many companies, especially small and medium-sized enterprises, are still uncertain and cautious with regard to influencer marketing in the virtual space. Only around 68% of German businesses have budgeted for influencer marketing on the Internet in 2017. "Germany has got a more reserved relationship with social media than other countries like the U.S. and the U.K." observes Davies (2017), attributing this to concerns about data security and privacy. The reasons for the uncertainty of providers in terms of influencer marketing are more far-reaching.
To date, the determinants and moderators of social media marketing effectiveness on consumer brand awareness are not fully understood in practice. In order to optimally develop virtual markets by means of influencer marketing, a comprehensive marketing concept is required that uses various online channels to address customers as comprehensively and effectively as possible (Safko, 2010). Furthermore, academic research in influencer marketing is just establishing. The use of social media marketing requires clarity about the conditions, potentials and risks as well as an understanding of determiners of effective strategies and communication methods (Rowley, 2004). German businesses' reluctance concerning influencer marketing is therefore partly due to a lack of know-how and uncertainty how consumers in fact perceive online influencers and what makes them follow those influencers.
Thus, the main objective of this paper is to close the existing research gap by investigating the impact of influencer marketing at the level of consumers and assess their quantitative effect. The conducted study identifies determinants of influencer marketing use in German companies and evaluates how influencer marketing impacts consumers' readiness to follow those influencers, their brand awareness and a as consequence, purchase intention. It is localized in the B2C (business to consumer) segment, i.e. refers to marketing directed from providers to private end consumers. In B2C marketing, psychological factors influencing the purchase decision and post purchase phase are of particular importance. The concentration on the B2C segment is thus a logical consequence of the study subject "brand awareness". It is assumed that the influencer's involvement, his or her motivation, authenticity of communication and real-life contributions are the main drivers of the postulated dependent variables. In addition, it is argued, that this relationship is moderated by influencer audience and product fit.
The approach is unique in the academic context. For the first time a comprehensive set of determinants, moderators and effects of influencer marketing is developed and empirically tested. The study gives German businesses intending to utilize social marketing channel orientation, how to make the best out of this strategy. Findings indicate that determiners of the three target parameters influencer followership, brand awareness and purchase intention show different patterns. While followership is achieved by high-influencer motivation and authenticity, brand awareness is established mainly via the real-life character and authenticity of contributions. Purchase intention is primary based on the real-life character of contributions and the fit between the influencer and his or her audience.
The structure of the paper is as follows. Section 2 reviews the literature concerning the characteristics of influencer marketing and its success factors. Section 3 introduces the method and the data applied for this research. Section 4 provides partial evidence for the postulated relationships. In section 5, results are discussed and implications for researchers and practitioners derived.

Influencer Marketing and Brand Awareness
The Marketing strategy of a company has to continuously adapt to new market developments, trends and requirements. Marketing anticipates consumption needs and creates new desires in consumers (Cannon et al., 2010). At the same time, consumers are not just passive targets of marketing information any longer. Instead, they participate actively in the formation of consumption trends. The internet enables consumers to share their consumption decisions and desires with peers and these readily imitate the behavior of model personalities and gatekeeper persons on the web. The internet has become a virtual environment of presentation and status demonstration and thus endows marketing with a new dynamic element (Homburg and Koschate, 2007). The virtualization and democratization of marketing in the age of internet has increased the importance of branding in consumer marketing. According to American Marketing Association (AMA) a brand comprises "name, term, sign, symbol, or design, or a combination of them, intended to identify the goods and services of one seller or group of sellers and to differentiate them from those of competition" (Gnann, 2008). Brands create a feeling of togetherness and differentiate social groups. Brands also reduce psychological insecurity and perceived social risks (Court et al., 2009). They create a common basis of reference, communication and identification (Biel, 2001). Brands stand for experience, have got an integrating effect in social live, they are statements of personality and connect people sharing the same attitude (Schmitt, 2012). Brands thus play an important role in the integration of the individuals in social networks and even have a community-building effect. Customers reconnect to brands inner conceptions and ideals which go far beyond the product itself but comprise a whole philosophy and comprehensive context (Tan and Ming, 2003). Brand associations are partly concrete and refer to the product and its properties in more frequently abstract categories, e.g. luck, welfare or status (Bauer et al., 2002). The complex meaning of brands causes consumer convictions which are distinct from practical experience and are deeply rooted at the level of sentiment and desire (Keller, 1993).
The virtualization and growing consumer involvement with marketing has multiplied the effectiveness of brands and has endowed new dynamics to brand advertisement. Brands are created by both provider and customer (Herrmann and Schaffner, 2005). Consumers increasingly contribute to the formation of brands by their consumption behaviour and by communicating brand image through electronic word of mouth (eWOM) above all via social media. Thus, brands gain strength in a dialogue involving both parties, customers and providers (Janson, 2012). In this context, brand awareness, i.e. the conscious perception and reflection of a brand, is probably the most crucial factor in the communication process between consumer and provider (Fournier, 1998, p. 368). Shaping brand awareness is crucial in corporate brand marketing and evokes brand image, buying intention and consumer loyalty (Meyer and Schwager, 2007).
Nowadays, influencer marketing is a powerful instrument of brand communication in the age of internet and social media. Although there is no homogenous definition of influencer marketing, the term basically refers to the effect that influential communicators contribute to create awareness for a product in social media and in this way make consumption trends emerge (Evans et al., 2017;Biaudet, 2017;Baker, 2014). Influencers are consumers in a central communicative function who impact on other consumers in a targeted way to promote the consumption of certain products via social media (Keller and Fay, 2016;Dron and Mohamad, 2015). Marketing increasingly uses influencers to explicitly develop brand awareness of particular target groups from inside the group (McQuarrie et al., 2013;Dlodlo, 2014).
Influencers control and guide eWOM among their followers and are recognized as opinion leaders in their social ingroup (Dalstam et al., 2018). Influencer messages expressing positive attitudes towards brands or passion for particular brands or products, (Baker, 2014), "manipulate consumers' buying decisions" (Hu et al., 2015) and exert significant influence on prone users (Woods, 2016) Consumers' involvement with the influencer induce positive emotional, rational attitudes on the contributions and motivate consumers to follow the influencer (Riedl and von Luckwald, 2019).
Perceiving influencers trustworthy and experienced, consumers develop positive brand attitudes (Breves et al., 2019). eWOM initiated and promoted by influencers, in whom consumer trust, contributes to consumers brand knowledge (Lock, 2016). Consumers thus transfer their positive sentiments for the influencer to the brand (Evans et al., 2017). The spread of positive brand attitudes and brand involvement influences the development of a positive brand image among a majority of consumers, which again increases individual consumers brand awareness (Lock, 2016), which finally motivates consumers' purchase intention (Gadalla et al., 2019).

Determinants of Influencer Marketing Success
The question, which features of influencer marketing result in success -i.e. influencer followership, brand awareness and finally purchase intention -has been discussed to extent in previous literature, in order to enable marketing agencies to systematically select and recruit influencers, who reach an audience that buys and recommends the advertised products (Cakim, 2009). A systematic review using an evaluation method suggested by Webster and Watson (2002), has found the following six major determiners of influencer marketing success.
The first one refers to influencer network involvement. A high level of social presence, i.e. frequent posts and intense network participation, increase influencers' fame and followers' trust in the influencer personality (De Veirman et al., 2017). Brand attitude is strengthened by famous and socially involved influencers (Jin et al., 2019). Influencers showing strong social engagement and standing, develop higher impact on consumers' brand perception and brand attitude (De Veirman et al., 2017). Youtubers' social influence contributes to perceived information credibility and consumers' involvement with product and brand (Xiao et al., 2018).
Secondly, influencer motivation plays a major role. Extrinsic as well as intrinsic factors can motivate influencers to participate in particular campaigns (Biaudet, 2017). Influencers motivated both intrinsically and extrinsically are found most effective and reliable (Hutter and Mai, 2013).
Thirdly, the motivation is followed by the influencers' communication style. Language style as well as contents of influencer communication have been found important determinants of influencer communication success with consumers. High word and image counts enhance influencers' recognition as specialists in some sectors (Chae et al., 2016). The choice of positive words connected to leisure, fun and holidays as well as the employment of positive emojis increases the reception of contributions (Jaakonmäki et al., 2017).
Fourthly, auality of influencer contributions has to be considered. High perceived argument quality contributes to influencer credibility for a sample of Chinese followers (Chae et al., 2016). High transfer of meaning by the influencer enhances Malayan consumers' brand attitude and buying intention . Consumers estimate eWOM reliability and high credibility information sources (Lock, 2016). According to a consumer survey, influencers' previous Instagram activity positively moderates their credibility (Breves et al., 2019).
Last but not least, the fit between the influencer and the audience as well as the product as to be discussed. The effectiveness of influencers on consumers is high when the influencer person corresponds to the target audience in character and brand identity perception (Dalstam et al., 2018). An audience perceiving influencer's con-gruence to own inner attitudes shows positive emotions and is ready to adapt the influencers' attitudes and accept his or her messages as a fact (Belanche et al., 2019).
In terms of the product-fit, perceived influencer credibility depends on his/her expertise with the particular product and similar products. High influencer product match indicates a positive relationship with consumer attitude towards the product and purchase intention . Instagram influencers dis-posing of high brand fit enjoy higher image, recognized expertise, trustworthiness and advertising success than their unexperienced colleagues (Breves et al., 2019). Influencers' brand experience contributes to perceptions of trustworthiness, expertise and attractiveness with followers (Eroğlu and Bayraktar Köse, 2019). Consumers recognize experienced and expert influencers as opinion leaders and follow their suggestions to interact, recommend and buy the product (Istania et al., 2019).

DATA AND METHODS
Theoretical model investigates influencer network involvement (IN), influencer motivation (IM), authentic communication style (IA) and quality of contributions (IR) as independent variables with a postulated influence on consumers' influencer followership, brand awareness and purchase intention. Audience (IP) and product fit (IF) act as moderators.
The model leads to the following hypotheses: There is a positive relationship between IN, IM, IA and IR and H1) consumers' influencer followership, H2) consumers' brand awareness and H3) consumers' purchase intention, positively moderated by IP and IF.
The model is investigated in the German market, as none of the evaluated studies refers to this particular market so far. Instead, the wide range of empirical scientific literature is conducted in other Northern European countries or in the USA. Nevertheless, the German influencer business, is of high interest due to its strong growth. The market volume of influencer marketing in Germany, Austria and Switzerland is expected to almost double to 990 million euros until 2020 as compared to 2017 (market volume 560 million euros).
Furthermore, previous quantitative studies are not conclusive concerning the evaluated influencer marketing strategies and success effects. Although some recent studies refer to a set of several parameters (Belanche et al., 2019;Breves et al., 2019;Eroğlu and Bayraktar Köse, 2019), none has so far evaluated the comprehensive set of possibly relevant design elements of influencer marketing found in the above review. Studies excluding one or the other relevant aspect however risk spurious results, since important side effects are neglected.
In order to test the model, an empirical survey in the German fashion influencer segment was conducted. The survey combines qualitative and quantitative research methods and evaluates two perspectives of the influencerconsumer communication process Combining a qualitative and quantitative study, the categories derived from the review are validated and then tested statistically (Cooper and Schindler, 2014).

Qualitative Interview Design
Semi-structured interviews with German fashion influencers are chosen as an approach for the qualitative section. Interviews provide in-depth and expert insights on the issue of influencer marketing, but at the same time allow the researcher to remain a less biased observer (Torbert and Taylor, 2008). The elements of the research model are addressed in the semistructured part questions (Mayring, 2002). The model is validated by evaluating results by issue in a comparative fashion.
Influencers are a competent target group for a survey assessing the impact of influencer activity on consumers' brand awareness and purchase intention. They earn prestige and gain followers due to their personal brand commitment and the online-demonstration of this . They understand the mechanism of developing brand awareness by effectful product communication and accordingly the essence of the research model. Three female fashion influencers are recruited for the interviews, which is an acceptable number for an in-depth qualitative study (Yin, 2014). The selected influencers dispose of several 100,000 followers in the fashion segment each as of March 2020, hold central gatekeeping positions and are thus experienced enough to represent the provider perspective.

Quantitative Consumer Survey Design
A quantitative consumer survey is conducted to test the causal relationships suggested by the model using statistical hypothesis tests. It addresses consumers following the interviewed influencers. 385 participants (completed surveys) are acquired by posting a surveylink to the survey together with a positive comment below recent YouTube-contributions of the interviewed influencers. The consumer survey is anonymized and refers to the categories of the research model using several contingent research questions (items) for each category. The items scales are adopted from similar validated scales to assess followership (Kelley, 1992), brand awareness (Keller, 2001) and purchase intention (Barber et al., 2012). The model constructs are formed from the survey items using reliability analysis. The causal relationships and hypotheses of the model are analyzed in SPSS using regression models and ANOVA tests. This methodology of causal inference is adequate for the metrically scaled data set.

Quantitative Study Representativeness
To link the quantitative study back to the interviews, the consumer survey specifically addresses the audience of the interviewed fashion influencers. To do so, a link to the recent YouTube posts of the fashion influencers containing the survey was added, asking video viewers to participate. Solely persons checking the influencers videos and reading through the YouTube comments retrieve the survey link. This method cannot avoid a further distribution of the survey link. Nevertheless, we can assume that these secondary participants are equally interested in the target influencer, since sharing happens if the secondary addressee is already involved. Altogether, 385 completed questionnaires have been collected in this way and the participants are about equally distributed between the three relevant influencers. Consequently, the survey is representative for consumers' following the interviewed influencers in the German fashion market. This strategy limits the reach of the study to the population of the audience of these fashion influencers, however, allows to calculate the necessary sample size to reach represen-tativeness of results. Assuming that there is no overlap between the followers of the three interviewed influencers, the total population size of their YouTube followers is 104,000. The size of the survey to reach representativeness is calculated from the size of the total population and for a certain confidence level (here 95%, resulting a z-value of 1.96) and a certain predefined error range e (here 5%), using the following rule of thumb formula for sample size (ss): in where Z is the z-value, p stands for the percentage picking a choice (expressed as decimal and c indicates the confidence interval (also expressed as decimal).

EMPIRICAL RESULTS
The interview results and consumer survey are evaluated and the results are condensed to accomplish and quantify the research model. The interviewees' reflections on determiners of influencer marketing success confirm the proposed effects of influencer activity on consumers. The review has found that influencer characteristics contribute to influencer followership (H1), brand awareness (H2) and finally purchase intention (H3). In correspondence with H1, the influencers declare that businesses get into personal contact to consumers through influencers and encourage consumers to identify with particular products and brands. Followership results due to the proximity of influencers and consumers. Furthermore, followers develop brand awareness if influencers are authentic and personally identify with their model rather than just playing their role artificially. Influencers' communication policy mediates particular emotions and a special identity, which consumers assign to the product or brand. Influencer followership thus ideally contributes to brand awareness, if designed adequately (H2). In addition, followers develop own intentions to buy the product due to influencers' communication policy. Authenticity, personal engagement and the creative and innovative presentation of the product motivate consumers to adopt the product as part of their personal consumption portfolio (H3). Hence, the three interviews confirm the relevance of the six determinants of influencer credibility derived from the review and concretize them. Participants agree that high relationship quality, intense contact and con-tinuity are more important than a broader network. All three influencers dispose of IM originating in their personal interest in the advertised products, their creative spirit and their desire for freedom and independence on their job. They agree that the communication style should mediate honesty, correspond to personal inner attitudes, mediate true friendship, avoid exaggerate perfection and show true emotional engagement. Also content quality is of importance. Mediated contents should disclose private information, show true images, mirror daily life, provide diverting information, be plausible and evoke emotions as well as personal thoughts. Moreover, participants agree on the high importance of IP, e.g. influencers should define themselves as part or even friends of their community and feel like their followers to be perceived as authentic. IF is required according to the interviewees, as influencers have to be experienced with the products, like them personally and consider them as truly important.
A quantitative survey among followers of the interviewed influencers is conducted to test the validity of the proposed relationships. The manifest constructs measured by the questionnaire are assumed to form the latent constructs proposed in the model. To check if these constructs are unidimensional their internal consistency has been calculated using Cronbachs Alpha. Following the guidelines from Blanz (2015) all Cronbach alpha values are classified as high (α > 0.80) and for TF and TB as very high (α > 0.90), confirming the reliability of the latent constructs presented in Tab. 1. In addition, a confirmatory factor analysis was conducted to statistically analyse the fit between the data and the coherences between manifest and latent constructs.

Tab. 1: Construct reliability -Cronbach Alpha
The confirmatory factor analysis refers separately for the dependent independent variables due to the fact, that all independent variables (IV's) have been derived from a qualitative study, while all dependent variables (DV's) have been extracted from different already validated questionnaires. Moreover, those variables have been shortened and partly adapted to fit the practical research requirements. Analysing both simultaneously could lead to constructs which are not interpretable in the conceptional realm of this research question.
Following interpretation guidelines from Hu and Bentler (1999) the results of the confirmatory factor analysis, in regard of the IV-part of the model, indicates a very good global fit, χ 2 = 188.83, df = 215, p = 0.901, as well as a very good relative fit as shown by the Tucker-Lewis Index, TLI = 1.01, and the Root Mean Square Error, RMSEA = 0.00 [0.00, 0.01].
Regarding the global fit of the DV-part of the model, the probability value of the quisquare test is below 0.05, χ 2 = 2437.46, df = 101, p < 0.001, indicating a rejection of the null hypothesis which states that the proposed model fits the data. Regarding the results for the relative fit, the Tucker-Lewis Index, TLI = 0.57, and the Root Mean Square Error, RMSEA = 0.25 [0.34, 0.25], furthermore indicate a poor model fit following interpretation guidelines from Hu and Bentler (1999). Therefore, a principal axis factoring analysis is conducted to explore which items may be responsible for the poor fit, to check if the hereby observed factors are interpretable and if they can be located in the conceptional realm of this research. As uncorrelated DV's are unlikely in this context and in light of the respectively poor modelfit an oblique rotation is used (Promax with a Kappa of 4), allowing for correlated constructs to obtain and evaluate a more data driven model. According to Eckey et al. (2002) as well as Bühner (2006) this usually leads to satisfactory results. Following the results from the monte carlo study conducted by MacCallum et al. (1999) the communalities in combination with the sample size will be used as an indicator if the results of a factorial analysis can be interpreted. Following the guidelines from Bühner (2006) the data can be classified as satisfactory for conducting a factorial analysis, as all communalities are above 0.50 and the sample size exceeds 300. The Kaiser-Meyer-Olkin test, KMO = 0.86, is satisfactory and does further indicate that the data is suitable for a factor analysis. The results of the Henze-Zirkler-, Mardia-and the Royston-test indicate all that no multivariate normal distribution is given. As the Bartlett test of sphericity requires such a distribution, it cannot be interpreted.
Both, the Eigenvalue criterion and the Screeplot indicate the presence of three factors which somewhat differ in their composition in contrast to the proposed model. The items TF4, TF5 and TF6 from the scale 'Influencer followership' have cross loadings on the scale 'brand awareness'. Both scales do correlate largely and significantly with each other which further indicates certain overlap of both scales, r = 0.60, p < 0.001. The same is observed for the items TB3 and TB5 from the scale 'brand awareness' regarding the scale 'purchase intention'. Both scales show strong and significant correlations as well, r = 0.66, p < 0.001. The Item TB6 from the scale 'brand awareness' does correlate distinctly higher with the scale 'purchase intention' than with 'brand awareness'. The scale 'purchase intention' and 'influencer followership' do share no cross-sectional items which fits to their weak, although significant correlation, r = 0.23, p < 0.001. The observed factorial structure did not lead to a better or at least similarly interpretable solution. Given that the three constructs have been derived from different questionnaires it appears adequate to stick to the original model as it is based on original questionnaires. As just one Item (TB6) can be clearly assigned to another scale, the current model will not be changed and used for the hypotheses tests while keeping possible distortions in mind. The constructs required for the hypothesis tests and regression models are all confirmed and no further constructs are necessary to assess the relationship.
For assessment, we carried out a multiple regression analysis. To ensure that the results of the model tests, carried out by a multiple regression analysis, can be interpreted appropriately, the necessary requirements for this procedure have been checked. The matrix scatterplot using loess smoothing indicated linear relations between all IV's. No outliers or extreme outliers have been observed. The results of the Durbin-Watson tests lie between 1.50 and 1.83 and therefor inside the interval of 1.5 and 2.5, as suggested by Brosius (2011), indicating no autocorrelations. The scatterplots, juxtaposing the standardized residuals and the predicted residuals, show no heteroscedasticity. The VIFand TOL-values as well as the correlations seen in Tab. 2 display no presence of multicollinearity between the IV's. In conclusion all requirements are sufficiently fulfilled and the results can be interpreted without consideration of distortions.
The independent variables where standardized resulting in a mean of zero and a standard deviation of one. The usage of standardized parameters furthermore changes the unit of the beta weights from its original units to the unit of standard deviation. This eases the interpretability of the regression weights as follows: b 0 is now the intercept, and b 1 the slope for an average value of an IV. Adjusted R 2 0.83*** 0.86*** 0.81*** Note: *, **, *** denote significance at the 10, 5 and 1 per cent level. All variables have been standardized before analyses. Bootstrap has been used to generate more precise standard errors. Standard errors in parentheses from the Bootstrap table.
coefficients IN and IA on the ten percent level. The coefficients IF and IP are significant on the one percent level without being considered as moderators. A moderation on the five percent level of IF has been observed for the relation between IR and purchase intention. In this regard H3 is accepted. Based on the quantitative results the research model has been partially validated and quantified. To differentiate the results, each hypothesis is displayed in a separate part model. Fig. 2 classifies the results by relevance of input factors. Accordingly, influencer activity largely determines influencer followership. IA and perceived IM induce consumers to follow an influencer. The fit of IP or IF are insignificant in this regard. In addition, IP is moderating the relations between IM, IR and influencer followership.
The results concerning H2 show, however, that IN and IM are not relevant for consumers' brand awareness. Instead, the major determiners are IA (β = 0.552) and IR (β = 0.735), i.e. the factual output of the influencer in the form of posts. The moderator IP is of comparatively low importance (but still significant; β = 0.09), although no moderation effects have been observed. According to the results for H3, the major determiners of consumers' purchase intention are IR (β = 0.72) and the proposed moderator IP (β = 0.49), although moderation was just observed for the relation between IR and purchase intentions. IA (β = 0.04), IN (β = 0.04) and IF (β = 0.13) do also show significant impact, which however is less compared to the other factors.
These observations imply that the major determiners of the three target parameters influencer followership, brand awareness and purchase intention differ. While followership is achieved by high IM and IA, brand awareness mainly depends on IR and IR. Purchase intention mainly results from IR and IP.
One possible interpretation is that consumers' followership mainly depends on the bond influencers establish between their own personality and the audience, while consumers' product and brand related behavior rather depends on the contents of influencer messages than on the influencer herself. Yet, the target factors of the models -influencer followership, brand awareness and purchase intentionare strongly interdependent. Most consumers following the influencers equally show brand awareness and purchase intention.
These results imply that each of the determiners of influencer activity as identified from the review and influencer interviews is confirmed by the model and finally increases influencer followership, consumers' brand awareness and purchase intention. Consumers are motivated to follow due the influencers due to his or her personality (IN and IM) mainly but develop brand awareness and purchase intention due to the authenticity and real-life character of posts. All four factors interact in getting consumers involved with the brand and product.
The empirical study has combined a qualitative and a quantitative section, where the qualitative section focused on the supply side and the quantitative section on the demand side of the German consumer market.
Interviews with three influencers in the German fashion consumer segment have been conducted which led to a concretization of the research hypotheses. According to the experience of the influencers the major determiners of influencer effectiveness with consumers are IN, IM, IA and IR. Influencers admit that the fit IF as well as IP could moderate influencers' impact. The interviews further suggest that influencer followership, brand awareness and purchase intention are mutually interdependent objectives of influencer marketing activity.
The second part of the empirical study has been implemented in the form of a consumer survey. Specifically, consumers following the interviewed influencers have been addressed to control the influence of secondary moderators. The survey fully confirms the research model derived from the hypotheses. IN, IM, IA and IR determine consumers' influencer followership, consumers brand awareness and purchase intention. There are interrelationships between the target factors. More specifically, the survey shows that IN and IM have an effect on consumers' followership, while IA and IR impact brand awareness and purchase intention above all. The assumed moderators of IP and IF are of comparatively low relevance as compared to the dominant impact of the four determiners.
Thus, we provide a model and causal explanation of the impact of influencer activity on consumers. Our empirical study has contributed to identify the most important determiners for the German B2C fashion influencer segment. Moreover, we identified that influencers' personal engagement and the quality of their contributions interact to motivate consumers followership, brand awareness and purchase intentions.
Generally, we offer the various major contributions, which distinguish our study from earlier research in influencer marketing effectiveness.
As the review of empirical studies (chapter 2) has shown, to date no study on the impact of influencer marketing in the German B2C consumer segment has been conducted so far. To our best knowledge, this study is the first to conclusively explore the relevance of influencer on the formation of consumer attitudes and purchase intentions in German fashion marketing. Earlier studies were not conclusive with regard to the evaluated influencer marketing strategies (Lock, 2016;De Veirman et al., 2017;Belanche et al., 2019;Breves et al., 2019;Eroğlu and Bayraktar Köse, 2019). Thus, we collected all relevant determiners of influencer marketing impact in the form of a thorough review and selected the most important factors for the German B2C fashion influencer segment in a targeted way by conducting interviews with three experienced influencers.
Also, previous literature is not fully comprehensive concerning the impact of influencer marketing at the consumer level. While some assess purchase intention Gadalla et al., 2019;Hughes et al., 2019;Istania et al., 2019;Riedl and von Luckwald, 2019), other focus on the impact on branding (Dalstam et al., 2018;Kucharska, 2018;Xiao et al., 2018;Belanche et al., 2019;Jin et al., 2019;De Veirman et al., 2017;Lock, 2016). Hence, we contribute by interconnecting the effects of influencer marketing with the most important elements of the purchase funnel within one single study. In addition, this study has explicitly analyzed the potential effects of moderating factors in the interviews as well as in the survey, as prior studies have vaguely referred to potential moderators of influencer activity. In accordance with previous research (Chae et al., 2016;Istania et al., 2019) we also found moderate impacts.
Finally, the study is the first to comprehensively assess all major elements that determine the effectiveness influencer marketing strategies at the consumer level. Earlier studies have been incomprehensive concerning the relevant determiners and their interaction (Xiao et al., 2018;Belanche et al., 2019;Jin et al., 2019;De Veirman et al., 2017;Lock, 2016). Specifically, our study shows that the identified elements of influencer marketing design are independent and each factor is itself important to enhance the impact of influencer campaigns.
In sum, we postulated a model of determiners and moderators of influencer marketing effectiveness on the purchase funnel, in which we systematically extracted the key factors relevant to the German B2C fashion segment based on influencer interviews and finally confirmed the developed causal model successfully in an empirical consumer survey. The resulting model is thus validated and reliable and can guide further research in influencer marketing effectiveness in fashion and other sectors, in Germany and abroad.
The insights of the empirical study also offer valuable insights to marketing management practice. Influencers have become an important instrument in consumer marketing. So far, their impact on consumers' decision processes has found less attention in Germany than in other countries. In the age of digitalization, German consumers however follow similar trends as consumers in other industrialized countries, have become equally susceptible to social media contributions and in result, to influencers. The consumer survey illustrates that many participants are still loosely interested in the contributions of the influencers only and do not base their brand and product decisions on influencer contributions alone. Comparing these results to previous insights from other national context, influencer marketing is just about to gain ground in Germany and by now should not be applied as a standalone marketing strategy. Marketing experts in the consumer segment should certainly consider targeted influencer campaigns as part of their marketing strategy. This advance could endow companies with a first mover advantage in a just emerging marketing domain. Businesses should acquire influencers in a targeted way and choose personalities adequate for the communication of the respective product, based on their personal interest with regard to the target audience.
The study has assessed determiners of the impact of influencer activity on consumers in the German fashion segment based on a comprehensive qualitative and quantitative study that considers the supplier (influencer) and consumer perspective. The insights of the study are conclusive. The theory-founded model has been refined and adjusted to the German fashion sector in the qualitative section and has been confirmed in the course of a consumer survey of 385 participants. Still the study has got some limitations, which invite further empirical research.
To begin with, the qualitative study is based on three interviews with influencers in the German fashion sector. These are not representative for the German influencer business and not even for the German fashion influencer business as a whole since several thousands of influencers are listed online. The interviews rather represent the individual opinions of the three participants. These statements have been compared and adjusted to differentiate the research model. This strategy necessarily leads to exemplary categories, which would not necessarily occur, if other influencers were interviewed. The comprehensive review-based model design has thus been refined in a partly arbitrary way.
The second limitation refers to the consumer survey with 385 participants. All of them are followers of the previously interviewed influencers. This approach ensures the internal validity of the study but prevents external validity. The results of the survey concern the contributions of the selected three influencers only. Other influencers might make completely different statements and contributions, dispose of other followers who would probably develop different opinions on the determiners of influencer effectiveness. Hence, the results are not generalizable beyond the reach of the three influencers. The extension of the interview section to further influencers could amend on this problem but would not fully solve the issue of representativeness. Still, the selected influencers are typical for the German fashion market, which indicates that somewhat similar results should be expected for other German fashion influencer networks.
Furthermore, regression analysis was applied to assess the postulated hypotheses. This kind of analysis assess linear impacts of several determiners on a single target only. It neglects interactions between the determiners and moderators and does not directly include cross-relationships between several targets. The complementary correlation analysis used to determine interactions between the targets is possibly biased since it does not consider the mutual interdependencies between the input factors. A structural equation model could amend on this problem and should possibly be applied in follow-up studies to illustrate the full complexity of the impact of influencer strategies on consumer attitudes and behavior.

CONCLUSIONS
The study has shown that the choice of influencers is essential to the success of the campaign. First, influencers' personal engagement is decisive to motivate consumers to follow the influencer. The quantitative consumer survey has confirmed that the readiness to subscribe to the influencers' contributions depends significantly on IN and IM. The interviewees, however, assert that intensity of IN is more important to marketing success compared to a broad network reach. Marketing experts recruiting influencers should be aware that the influencers' contributions are decisive to influencers' impact on consumers brand and product choice, in so far that perceived IA and IR of influencers determine whether consumes develop brand awareness and form purchase intentions towards the advertised products.
Finally, the review section has shown that influencer marketing strategies should fit with the brand image the company pursues as a whole. The interviewed influencers have pointed out that self-reliance is essential to provide authentic and real-life type contributions. If businesses relegate influencers' freedom of contents and style too much, they risk that influencers loose in authenticity and in result in consumer effectiveness. Businesses intending to include influencer marketing into their strategy should balance between the development of a comprehensive marketing guideline and the admission of self-defined advertisement styles.