Introduction

Prominent developments in the online sphere have been embraced by the concept of Web 2.0 (Cooke and Buckley 2008; Lehmkuhl and Jung 2013). The social movement is a vital aspect of this evolution, leading to a consumer-driven online sphere (Labrecque et al. 2013). Consumers utilize various tools to create and share online content, so-called user-generated content (Goh et al. 2013; Moe and Schweidel 2012). Hoffman and Novak (2012) found these social media to be characterized primarily by collaboration, community, conversation, and sharing. Because of the ability to interact with other consumers and organizations, the consumer voice becomes increasingly important for both consumer and management decisions (Labrecque et al. 2013). Consumers find information provided by other consumers, such as comments and recommendations, to be more valuable than corporate information (Bae and Lee 2011; Bronner and de Hoog 2010; Goh et al. 2013; Trusov et al. 2009). Thus, these consumer interactions may increase trust in a company and consequently increase consumption intention (Hutter et al. 2013; Ng 2013; Phang et al. 2013) and revenue (Chevalier and Mayzlin 2006; Moe and Trusov 2011). As consumer interactions are subject to less control and can spread rapidly, these interactions also pose some challenges and need to be monitored continuously (Munnukka and Järvi 2014; Peters et al. 2013; Yadav and Pavlou 2014). This ongoing monitoring process provides insights into the need to react to consumer interactions, consumer opinions on offered goods and services, and potential trends (Peters et al. 2013; Schweidel and Moe 2014). The present study follows this line of research by analyzing consumer social interactions from a competing retailer brand perspective and by providing additional insights into the information potential of social networking sites (Wirtz et al. 2013). From a competing retailer brand perspective, this study contributes to the literature by structuring the analysis along three customer engagement levels as well as providing an approach to identify and to relate social interaction strategies.

Social media has attracted an emerging stream of research, addressing various types of social media, such as blogs (e.g., Gopinath et al. 2013; Gruhl et al. 2005), microblogs (e.g., Tirunillai and Tellis 2012; Zhang et al. 2011), user-generated product rating and review websites (e.g., Chevalier and Mayzlin 2006; Moe and Schweidel 2012; Phang et al. 2013), social shopping communities (Holsing and Schultz 2013; Olbrich and Holsing 2011), and social networking sites (e.g., de Vries et al. 2012; Katona et al. 2011). Researchers have also considered the combination of traditional and social media (e.g., Onishi and Manchanda 2012; Yu et al. 2013) and multiple types of social media (e.g., Dhar and Chang 2009; Schweidel and Moe 2014; Stephen and Galak 2012). The present study focuses on consumer interactions on a social networking site and adds to the body of research by analyzing the informational potential of such social interactions from a competitor perspective. Social networking sites can be described as online platforms for networks of individual and organizational users to share and consume various types of information. On so-called brand pages, users can, for example, create posts and interact with these posts by commenting, liking, and sharing (de Vries et al. 2012). However, users do not need to interact with the information on these pages, but can potentially do so on the entire social network site – for example, through one’s own or any connected user’s newsfeed (Lipsman et al. 2012). The present study draws on the social networking site Facebook. Facebook reported 936 million daily and 1441 million monthly active users in the first quarter of 2015 (Facebook 2015). Across the purchase funnel, Facebook and Twitter are positioned at the top of the funnel driving awareness and interest, while in the middle of the funnel, Pinterest is about consideration, and at the base, Amazon, TripAdvisor, and Yelp, are about purchase intention and driving sales (Fulgoni and Lipsman 2015).

The overall aim of this study is to reveal insights that a retail brand can infer from consumer social interactions on a social networking site about a competing retail brand, as well as general and industrial trends. Across six apparel retail brands, we use 2,573,620 consumer social interactions from the social networking site Facebook. Specifically, this study analyzes different social metrics across three consumer engagement levels, so as to contribute to the research on social networking sites by identifying potential insights from the analysis of these social interaction metrics. Firstly, we analyze the number of individuals connected to a brand page, so-called fans. Across the study period, the number of fans increases continuously for all six brands. Thus, if this number decreases, a brand faces a relevant (social) problem. We also find a significant correlation between fan numbers and brand sales. While no causality is implied, the number of fans may indicate the sales or brand value of a competing brand. Secondly, the posting and response behavior of the apparel retailer is analyzed. A (social strategy) framework is presented to identify the social strategies of the apparel retail brands, based on the posting and response behavior. The social strategy framework extends the existing literature and provides comparative insights. Thirdly, social metrics that represent more engaged consumers, such as commenting, liking, and sharing messages, are analyzed. Specifically, a case of negative interactions is presented, indicating the ability of consumer-initiated social interactions to reveal relevant topics and trends. The three levels of analysis contribute to the structured analysis of social interactions. The social strategy framework presents a novel approach to position and relate different social strategies.

The remainder of the paper is organized as follows. The next section discusses related research on social networking sites and the findings of social media research in the apparel industry. The research framework and the research questions are then derived, after which the data sample is accounted for and presented. The empirical results and limitations of this study are discussed subsequently as well as directions for future research.

Related literature, research framework and research questions

Overview

Social media has become subject to increasing research interest. Corresponding overviews and agendas are, for example, provided for social media brand management (Gensler et al. 2013), online social networks (Kane et al. 2014), social media (Ngai et al. 2015), social metrics (Peters et al. 2013), and social commerce (Yadav et al. 2013). For social networking sites, researchers have focused on modeling (online) social networks (Ansari et al. 2011; Chatterjee 2011; Katona et al. 2011; Trusov et al. 2010; Xu et al. 2012a and 2012b), the privacy of social networking sites (Gan and Jenkins 2015; Jeong and Coyle 2014; Külcü and Henkoglu 2014; Li et al. 2015; Narayanaswamy and McGrath 2014; Min and Kim 2015; Nosko et al. 2010; Patsakis et al. 2014; Stutzman et al. 2012; Tucker 2014), and the opportunities and risks associated with social networking sites (Champoux et al. 2012; Corstjens and Umblijs 2012; Dekay 2012; Gomez-Arias and Genin 2009; Hansson et al. 2013; Munnukka and Järvi 2014). Before the research framework and the research questions are introduced, related literature on the drivers of consumer interaction, the informational value of social networking sites, and research on social media in the apparel industry, is presented.

Drivers of consumer interaction

Consumer interactions are considered to be an easily perceived response metric. For that reason, the literature analyzes drivers of consumer interactions on social networking sites (Chatterjee 2011; de Vries et al. 2012; Kabadayi and Price 2014). For actual data from eleven brand pages, de Vries et al. (2012) identify different drivers with regard to the number of likes and comments. Vivid and interactive posts, as well as the share of positive comments, influence the number of likes. The number of comments is, by comparison, affected by interactive brand posts, such as a question, and the share of positive and negative comments. Based on 269 questionnaires, Kabadayi and Price (2014) found that comment and liking behavior differs for consumers who either prefer a one-to-many (‘broadcaster’) or a one-to-one (‘communicator’) interaction. The preference is affected by different personality traits. Correspondingly, users who contribute more actively on social networking sites exhibit different information and purchase behavior to that of more passive users (Liao et al. 2014). Considering consumer-generated versus brand-generated posts, the results of a clickstream study of 2173 consumers of a social networking site suggest that consumer-generated brand posts are more likely to be recommended, but are not more likely to generate referrals for high-involvement products (Chatterjee 2011). This finding underlines the multiplier effect of social networks (Lipsman et al. 2012). Based on these findings on the drivers of consumer interactions on social networking sites, the present study focuses on the following question: What information do these consumer interactions provide, especially from a competitive perspective? The study explores this question by relating various measures of consumer interactions to retailer characteristics (sales and store number), brand social strategies, and a specific brand incident.

Informational value of social networking sites

Another related stream of research focuses on social networking sites as a potential data source. Using a matching system across the social network structure, the relationships in social networking sites have been proposed as enabling new business contacts (Kazienko et al. 2013). The social connections have also been considered in terms of predicting a consumer’s perceived appropriateness of the online channel for buying products and services (Verbraken et al. 2014). Social network predictions perform well when the products or services are rarely bought or reserved online. Social network data has been shown to be useful for selecting groups with a relatively high propensity to click on advertisements, register for recreational leagues, and make retail purchases (Goel and Goldstein 2013). The authors found that such data complement demographic and behavioral predictors. However, if transactional data is available, social network data is found to provide almost no marginal benefit (Goel and Goldstein 2013). Predicting stock prices is another use of data from social networking sites (Schweidel and Moe 2014). However, Schweidel and Moe (2014) find that sentiment may differ across various types of social media. This study is closely related to this stream of literature, as the focus is on the informational value of social interactions for managerial decisions. In order to extend the current state of research, the study adopts a comparative perspective for consumer social interactions with six apparel retail brands. Specifically, we analyze whether social interactions relate to brand sales, and whether these interactions signal social strategies and specific trends.

Social media in the apparel industry

Previous research has addressed social media in the apparel industry. In an undergraduate sample, young consumers are found to engage in word of mouth activities, based on brand excitement, brand image, and brand love (Ismail and Spinelli 2012). For a social shopping community that focuses on fashion, living, and lifestyle, researchers identified the effect of direct shopping features, such as filter and search mechanisms, and consumer social interactions, such as lists, profiles, styles, and tags, on conversions (Olbrich and Holsing 2011) and micro-conversions (Holsing and Schultz 2013). Using an experimental Facebook shop for accessories and clothing, trust in the social networking site is found to play a mediating role in the relationship between social interactions and purchase intentions (Ng 2013). From an industry perspective, the present study is most closely related to Park and Cho (2012) and Goh et al. (2013). Based on 186 questionnaires from a sample of female college students, Park and Cho (2012) study how social networking sites affect consumer information seeking behavior in the context of apparel. The authors find that commitment is determined by factors reflecting consumers’ psychological attachment to the online community. Commitment then positively relates to the information seeking behavior, potentially moderated by social pressure to conform to the peer group’s expectations. Goh et al. (2013) compare the effect of user-generated and marketer-generated content in the case of a brand page of a casual wear apparel retailer. The authors find that general activity in brand pages increases sales. In their study, the marginal effect of consumer-generated content is about 22 times higher than the effect of marketer-generated content. For driving sales, undirected messages are found to be more effective than directed ones for informative and persuasive consumer-generated content, whereas directed messages are more effective than undirected ones for persuasive marketer-generated content (Goh et al. 2013). The present study uses actual consumer social interactions to identify insights that can be inferred from this behavior. Specifically, consumer interactions on a social networking site are analyzed from an individual apparel retail brand perspective and from a comparative perspective. High consumer engagement with fashion and lifestyle, and the correspondingly high likelihood of consumer social interactions for apparel retail brands on a social networking site, provide ideal conditions for studying the insights consumer interactions give into a brand’s market development, social strategy, and specific trends.

Research framework

This study draws on customer engagement theory as a research framework. Based on relationship marketing theory and the service-dominant logic, and considering conceptualizations in various fields of research, customer engagement has been defined as “a psychological state that occurs by virtue of interactive, cocreative customer experience with a focal agent/object (e.g., a brand) in focal service relationships. It occurs under a specific set of context-dependent conditions generating differing CE [=customer engagement] levels; and exists as a dynamic, iterative process within service relationships that cocreate value. …” (Brodie et al. 2011, p. 260). In particular, the present study acknowledges that customer engagement is situation-specific and may result in different levels (Brodie et al. 2011; Sashi 2012; Tsai and Men 2013). The research framework follows these findings as a sequence for analyzing social interaction measures at the levels of first creating a relationship, secondly consuming content, and thirdly contributing content (Sashi 2012; Tsai and Men 2013). Before extending these levels and the corresponding measures, this study is related to previous research studying customer engagement in the domain of social media (Bitter et al. 2014; Brodie et al. 2013; Cheung et al. 2012; Dessart et al. 2015; Gummerus et al. 2012; Hollebeek et al. 2014; Kabadayi and Price 2014; Sashi 2012; Tsai and Men 2013; Wirtz et al. 2013).

A recent theoretical framework proposes that consumer engagement in an online brand community, such as brand pages on a social networking site, leads to community and brand commitment, satisfaction and loyalty (Wirtz et al. 2013). Further, Wirtz et al. (2013) point out that consumer engagement leads to consumer interactions with the online brand community and thus become a source of consumer data and market research. Beyond idea generation for products and services, this present study argues that consumer interactions, as a result of consumer engagement, provide additional brand-relevant insights – also from a competitive perspective. Recent findings support the propositions of the theoretical framework (Brodie et al. 2013; Cheung et al. 2012; Dessart et al. 2015; Gummerus et al. 2012). Gummerus et al. (2012). for example, find that engagement leads to consumer loyalty and satisfaction mediated by economic, entertainment, and social benefits, in the case of a game brand on Facebook. Furthermore, consumer engagement on a social networking site increases brand commitment and consequently recommendation and purchase intention (Cheung et al. 2012). Similarly, in a health and fitness setting, Brodie et al. (2013) find engaged consumers to exhibit consumer loyalty, satisfaction, empowerment, connection, trust, and commitment. Whereas these studies draw on questionnaires (Cheung et al. 2012; Gummerus et al. 2012) and interviews (Brodie et al. 2013; Dessart et al. 2015) to identify brand and consumer outcomes of consumer engagement, this study uses measures of actual consumer behavior and analyzes their applicability for market research. As some researchers suggest, the implication of consumer engagement may not only represent (individual) sales value, but also that engaged consumers create value as co-creators in business processes, such as for market research, product development, and selling process (Barnes 2014; Kabadayi and Price 2014; Sashi 2012). The present study contributes to this line of research by relating social interactions to brand sales, social strategies, and specific trends.

Our study analyzes different engagement levels of consumer interactions on social networking sites. We thus extend upon the customer engagement matrix (Sashi 2012) by analyzing potential insights from three engagement levels of consumers with high emotional bonds and high relational exchange (‘fans’). The corresponding social interaction metrics have already been used to identify highly engaged users on social networking sites (Dessart et al. 2015; Gummerus et al. 2012). The present study further extends this line of research and explores potential insights gained from these metrics from a comparative perspective. For the first engagement level, this study considers the fan number. Fans are the individuals who actively like a specific page of a brand, company, organization, or public figure in a social networking site (Champoux et al. 2012; Lipsman et al. 2012; Naylor et al. 2012). Consumers thus actively create a relationship with one specific page. Post and user posts are analyzed as a second level. Posts are messages by a page representative, that are shown on the specific page and in fans’ newsfeed; user posts are generally written by consumers on the specific page (Chatterjee 2011; de Vries et al. 2012; Dekay 2012; Goh et al. 2013). This study centers particularly on the posting and response behavior at this level. The latter is characterized by whether and how fast a brand representative replies to a user post. Thus, the analysis concentrates on the behavior of a retail brand on this level. On a third level, the number of comments, likes, and shares measure the engagement with posts and user posts (de Vries et al. 2012; Dekay 2012; Kabadayi and Price 2014). The analysis on this level primarily focuses on consumer-initiated social interactions referring to user posts, likes, comments, and shares. Table 1 summarizes these social interaction metrics and Fig. 1 displays the research framework.

Table 1 Description of social interaction metrics
Full size table
Fig. 1
figure 1

Research framework

Full size image

Even though the three levels suggest an increase in consumer engagement, it is important to note that the research framework and the analysis in this study is first based on the sequence and then on the degree of these activities. For example, consumer activities on the third level refer to varying degrees of consumer engagement on the social networking site (Tsai and Men 2013). In general, these different engagement levels follow traditional user differences in activity – from consuming to contributing – and have been demonstrated for engaged consumers on social networking sites (Tsai and Men 2013). These levels thus follow different stages of the customer engagement cycle, from connection via consumption and interaction to commitment, advocacy, and engagement (Sashi 2012). The classification of user posts and comments may be ambiguous, depending on their content.

Research questions

Creating a relationship: fan number

Consumer engagement on a social networking site begins by establishing a connection between oneself and a brand page (Champoux et al. 2012; Lipsman et al. 2012; Naylor et al. 2012). This study focuses on market insights, such as development, sales and size, potentially derived from the number of such relationships on a social networking site.

Fans are individuals who consciously decide, by liking a brand page, to follow a brand on a social networking site (e.g., Naylor et al. 2012). In this manner, fans are a measure of potential reach. More importantly, in the consumer-driven online sphere, fans are intermediaries for reaching the second-order fans – the friends of fans (Lipsman et al. 2012). Managers need to leverage the ability to influence friends of fans, increase their depth of engagement and loyalty, and generate purchase behavior (Barnes 2014; Lipsman et al. 2012; Naylor et al. 2012). To utilize the potential of a brand’s fan base, managers have to identify the influential individuals on a social networking site (Chatterjee 2011; Trusov et al. 2010; Xu et al. 2012a and 2012b). However, influential individuals are not easily identified by the number of their connections (Katona et al. 2011; Trusov et al. 2010). Previous research suggests that consumer interactions on a social networking site lead to recommendations and purchase intentions (Cheung et al. 2012; Hutter et al. 2013; Ng 2013; Phang et al. 2013) and predict other business measures, such as retail purchases (Goel and Goldstein 2013; Goh et al. 2013) and stock prices (Schweidel and Moe 2014). The present study extends this research stream and investigates whether the number of fans provides insights into market developments, such as a brand expansion strategy.

RQ 1: Does the fan number provide insights into market developments?

Furthermore, the study extends on the notion that consumer social interactions lead to recommendations and purchase intentions (Cheung et al. 2012; Hutter et al. 2013; Ng 2013; Phang et al. 2013) and predict retail purchases (Goel and Goldstein 2013). In a study of female apparel purchasing behavior in Germany, Cao et al. found (2014) that roughly every third purchaser becomes a fan of the brand. We use correlation analysis to study whether the number of fans on a social networking site correlates with sales and store numbers.

RQ 2: Does the number of fans relate to the sales revenue?

RQ 3: Does the number of fans relate to the number of stores?

Brand activities: posting and response behavior

The consumption of brand messages constitutes another level of consumer engagement. As pointed out above, this study concentrates on brand activities at this level. By analyzing these brand activities, can we gain insights about the social strategies of a brand from an external perspective?

Of the Fortune 500, 348 companies,70 %, are on Facebook (Barnes 2014). The ability to directly engage consumers via brand messages (posts) is mentioned as one beneficial aspect of social networking sites (de Vries et al. 2012; Gomez-Arias and Genin 2009; Lipsman et al. 2012). In general, brand posts are less likely to be recommended, but are not less likely to generate referrals for high-involvement products than consumer-generated posts (Chatterjee 2011). Goh et al. (2013) find that brand posts primarily influence purchase behavior through persuasive communication. Furthermore, in the case of persuasive communication, directed brand posts are found to be more effective than undirected ones. In a study of 355 posts of 11 international brands on a social networking site, vivid and interactive brand posts positively affect the number of likes, and interactive characteristics, such as a question, increase the number of comments (de Vries et al. 2012).

As companies have less control of consumer interactions, negative interactions pose serious risks to companies (Champoux et al. 2012; Corstjens and Umblijs 2012; Dekay 2012). Nonetheless, some evidence exists that negative interactions also have positive effects (Berger et al. 2010). One potential managerial strategy for reducing the dangers of social networking sites is formulated as acting quickly, managing the dialogue, taking responsibility, solving the problem, or directing the problem to a different channel (Champoux et al. 2012). Thus, the response behavior assumes an important role in countering the potential negative effects of social networking sites. However, in a study of 40 brand pages, fewer than a quarter respond to negative consumer posts and nearly half of the companies are found to delete unfavorable comments (Dekay 2012). However, deleting negative comments has been found to further harm a brand if consumers become aware of this (Champoux et al. 2012).

Hence, posting and response behavior reflect central elements of brand strategies on social networking sites. This study proposes a framework for inferring different approaches by utilizing these two behavioral characteristics. The aim is to derive the approach to the social networking site adopted by different brands. In this way, this framework is inspired by traditional product and market positioning (Kotler and Armstrong 2014). The focus in this study is on analyzing whether social interactions provide information that is useful for deriving such brand approaches. Specifically, this study draws on the average number of brand posts per day, representing posting behavior as one dimension. The second dimension, the response behavior, is measured by the response rate, calculated as the number of user posts that include a comment by the retail brand, related to the total number of user posts. Thus, the response rate is [0;1]. Naturally, there are no clear values for differentiating between the resulting strategies. For this explorative study, an average post per day of 1 and a response rate of 0.5 are chosen to differentiate the resulting four different social interaction strategies. This study concentrates on two behavioral dimensions that are in accordance with the dominant strategic discussion on brand management (e.g., Gensler et al. 2013; Goh et al. 2013) and negative consumer interactions (e.g., Champoux et al. 2012; Corstjens and Umblijs 2012; Dekay 2012) in the literature. These measures are expected to reflect the importance of and investment by a brand in its social interaction strategy.

The four social interaction strategies are the low-interaction strategy, posting strategy, response strategy, and high-interaction strategy. A low number of posts and a low response rate characterize a low-interaction strategy. A brand that focuses on the number of brand posts, but rarely responds to user posts follows a posting strategy. A response strategy, by contrast, is centered on a high-response rate and few brand posts. A high number of posts and a high response rate finally characterize a high-interaction strategy. This study explores the applicability and use of this social interaction strategy framework.

RQ 4: Does posting and response behavior provide insights into social media strategies?

Consumer activities: consumer-initiated social interactions

Consumer activities, such as liking, commenting, and sharing, spread content through an indivual’s online social network (Chatterjee 2011; Lipsman et al. 2012; Trusov et al. 2010). If these consumer activities follow trend movements, consumer-initiated social interactions may predict such trends.

Consumers can engage brands on a social networking sites by posting directly to the site (Chatterjee 2011; de Vries et al. 2012; Dekay 2012; Goh et al. 2013) or interact with brand and user posts by commenting, liking, and sharing (de Vries et al. 2012; Dekay 2012; Kabadayi and Price 2014). As negative interactions are a considerable concern of brands using social networking sites (Champoux et al. 2012; Corstjens and Umblijs 2012; Dekay 2012). this study is specifically interested in whether consumer-initiated social interactions provide information on such negative interactions, addressing the question of when to and when not to react to consumer-initiated social interactions.

RQ 5: Do consumer-initiated social interactions provide insights into specific trends?

Methodology

Data sample

The present study is based on six apparel retail brands. The apparel retail industry is chosen in line with previous research, because both the brands and the consumers are highly engaged and exhibit a high likelihood of social interaction (Cao et al. 2014; Goh et al. 2013; Olbrich and Holsing 2011; Park and Cho 2012). The six apparel retail brands are C&A, Ernsting’s family, Esprit, H&M, Primark, and Zara. All six brands are present on the social networking site, have retail outlets, and are established brands. The six apparel retail brands represent a spread of various retail sizes. Esprit, H&M, and Zara are global represented brands; C&A and Primark are located primarily in Europe, and Ernsting’s family is a retailer represented in Austria and Germany. In addition, Primark presents a retail brand with an aggressive expansion strategy in recent years. Thus, the changes in social metrics provide a comparable case across different retailer sizes. Facebook was chosen as a social networking site, being a global and leading networking site including all six studied retailer brands. Facebook reports 1441 million monthly active users for the first quarter of 2015 and a revenue of $12,466 million for 2014 (Facebook 2015).

Data collection and data overview

For the six apparel retail brands, all user social interactions on the social networking site Facebook were recorded from April 14th, 2014 to May 31st, 2014. In accordance with the research framework, we recorded the number of fans, posts, user posts, comments, likes, and shares from the global pages of these retail brands as metrics for social interaction. A software tool extracted the activities on the global pages. On a daily basis, the total number of fans and changes in the number of posts, comments, likes and shares are registered. Additionally, all user posts to the six brand pages were retrieved at the end of the study period, including the total number of comments, likes, and shares of these user posts. The user post data also includes the creation date, whether a brand representative commented on a user post, and if applicable, the response time in seconds.

The six apparel retail brands account for a growth of 1,882,518 fans, 414 posts, 3035 user posts, 12,241 comments, 664,012 likes, and 11,400 shares over the study period of 48 days. Descriptive data for these social metrics is presented in Table 2, which contains aggregated values for fan growth, posts, user posts, comments, likes, and shares and averaged values for fans. We also collected sales data and the number of stores for the six retail brands from financial reports (Esprit 2013; H&M 2013; Primark 2013; Zara 2013) and press releases (C&A 2012; Ernsting’s family 2013).Footnote 1 For example, in the fiscal year 2012/2013, 1717 stores yielded €814 m in sales for Ernsting’s family. The average fan numbers on Facebook was 157,035 over the 48-day study period, a growth in fan number of 7042. 64 Ernsting’s family posts received 3499 comments, 5397 likes, and 549 shares. The number of comparatively high comments and likes is primarily attributed to two sweepstakes on April, 17th and May, 28th.

Table 2 Overview of retail and social metrics (April 14th, 2014 to May 31st, 2014)
Full size table

One point worth noting is that the brand with the most fans (Zara, 21 m fans) is the least active, measured by brand posts (11), over the study period. In comparison, the smallest retailer, measured by sales (Ernsting’ family, €814 m) and fan number (157, 035), yielded the third most posts (64). The brand retailer with the highest sales (H&M, €16,840 m) seemed overall the most active on the social networking site..

For the social interaction metrics, the mean, standard deviation, minimum, and maximum value are presented in Table 3. For example, for the C&A brand, we registered a daily average of 0.92 posts by C&A representatives over the study period and the standard deviation was 0.58. Over the 48 days, C&A representatives posted a minimum of 0 messages and a maximum of 2 messages per day. Data on comments, likes, and shares refer to the corresponding brand posts per day and per brand.

Table 3 Descriptive data on social interaction metrics (April 14th, 2014 to May 31st, 2014)
Full size table

Results

Analysis of fan numbers

Addressing RQ 1, fan development is displayed in Fig. 2 for the six apparel retail brands. For comparative purposes, Fig. 2 displays the percental change according to the index value at April 13th, 2014.

Fig. 2
figure 2

Fan development

Full size image

Over the seven-week time frame, Primark registered the highest growth in fan percentages (8.79 %). The average of the indexed percentage fan increase is 0.1831 and almost linear (β ≈ 0.1768 %; ɛ = 0; R2 = 0.998; p < 0.01). This development reflects the recent market expansion by Primark. Figure 2 also displays a constant period of comparatively low fan growth for Esprit from April 22nd to April 29th (β ≈ 0.0465 %; ɛ = 0; R2 = 0.988; p < 0.01). No decrease in fan numbers was registered over the time period studied. Consumers rarely bother to ‘unlike’ a certain brand. Consequently, a decrease in the number of fans is a strong signal that a brand is facing some challenges. As reported in previous research, consumers however engage directly with the brand by commenting negatively (Champoux et al. 2012; Dekay 2012), even rapidly in a 24 h time frame on a weekend (Hoffman and Fodor 2010). Thus, the number of fans is a strong signal, but not an indicator of negative sentiment. In a later section, a negative incident is presented for the brand Zara. However, the fan number for the brand Zara shows no significant change around April 22nd – the day of the news article. In contrast, Zara’s fan number follows a constant linear trend (β ≈ 0.0592 %; ɛ = 0; R2 = 0.999; p < 0.01). Overall, the fan number does not indicate a single local incident (Australia) for the global brand Zara, but the fan number reflects the expansion strategy of Primark. Thus, fan numbers provide some insights into market developments (RQ 1).

Further, the fan number is analyzed according to the sales volume (RQ 2) and the store number (RQ 3). Based on the six apparel retail brands, the Pearson product-moment correlation coefficient was calculated. The correlation coefficient for total number of fans and retail sales is 0.893, p = 0.017 and −0.194, p = 0.712 for the total number of fans and number of stores. If the number of stores accounts for the different distribution strategies of Esprit, by only considering the 1024 owned Esprit stores, the correlation coefficient is 0.711, p = 0.113, highly correlated with retail sales 0.727, p = 0.102. The number of fans positively relates to retail sales. From a comparative perspective, the ratio of fan number and retail sales provides a simple measure for socially comparing a brand. The results provide some support for RQ 2. The adjusted correlation between fans and store number was 0.711, but not significant (RQ 3). Even considering the small sample, retail space may be a better indicator of a brand’s sales activities. Retail space data was however not available for all six brands, so that future studies could extend usefully these results.

Analysis of social networking site strategies

According to the apparel retail brands and per day of the week, Table 4 provides an overview of the average posting behavior and Table 5 displays the response behavior. To calculate the response behavior, the number of user posts that include a comment by a the retail brand was related to the total number of user posts on the brand page. The time difference was measured in seconds until the first brand response. Table 5 displays the time difference in hours.

Table 4 Average posts per day and per apparel retail brand (April 14th, 2014 to May 31st, 2014)
Full size table
Table 5 Average user posts, response rate and response time per weekday and per apparel retail brand (April 14th, 2014 to May 31st, 2014)
Full size table

The posting behavior across all six apparel retail brands concentrates on the middle of the week, Tuesday till Friday, so that weekends record fewer brand posts overall. However, the apparel brands vary in their posting behavior. The posting strategy on weekdays is adopted by Ernsting’s family, H&M, and Zara. C&A and Primark, by contrast, increase on average, the posting behavior towards the weekend. The posting behavior of Esprit appears to be mixed, combining the middle days of the week and Saturdays for postings. An extended observation period is needed to clarify this approach, again, something future studies can build on.

Social strategies also differ with regard to the volume of brand posts. H&M provides on average the most brand posts during the time period studied. H&M adopts a strategy of posting multiple pictures relating to the post theme, thus avoiding potential information overload of the audience. The posting behavior of Primark can be described as being present and maintaining a certain level of social activity. In a less extensive manner, this behavior is also adopted by C&A, Ernsting’s family, and Esprit. By contrast, the posting behavior by Zara indicates that the brand is not engaged in the social networking site.

This evidence is supported by the company response behavior. Zara responds to 6 % of all user posts and takes about 15 h to respond. Similarly, Primark only comments on one of 465 user posts over the study period. This contrasts with posting behavior aimed at being socially present. C&A, Esprit, and H&M follow a more selective approach to responding to user posts. Besides the timing effect of weekends, the smallest retail brand of Ernsting’s family displays the most engaged social behavior strategy, indicated by posting and response behavior. The comparatively high number of comments is based on two sweepstakes on April 18th (1591 comments) and May 28th (1515 comments). The brand is also the fastest to respond to any user posts, although, still taking about 12 h on average.

The range for responding to user posts extends from half a day to more than a day, in the cases of C&A, Esprit, and H&M. Overall, this finding indicates that all six apparel retailers rely partly on the capabilities of the social networking site and thus its fan base, to respond to some user posts. All 3035 user posts accounted for 2568 comments, 799 from brand responses and 1769 from users.

In accordance with the social interaction framework, the six apparel retail brands are displayed in Fig. 3. On the x-axis, the average posts per day represent the posting behavior of the apparel retail brands. For presentation purposes, the x-axis is in logarithmic scale, so as to reflect the high posting value by H&M. The average response rate by brand per day reflects the response behavior on the y-axis. For the six apparel retail brands, we identified a prototypical representative for three of four social interaction strategies: low-interaction strategy, posting strategy, and high-interaction strategy. As the critical values are selected arbitrarily, the retail brands C&A and Esprit could thus be attributed to different strategies. Across the seven-week study period.

Fig. 3
figure 3

Social Interaction Strategies

Full size image

The smallest apparel retail brand of Ernsting’s family showed a comparatively constant posting behavior (except for weekends) and strong engagement through responding to user posts. Primark clearly represents an extreme case of the posting strategy. Based on the comparative change in the fan base (see Fig. 2), this strategy is successful according to this measure. Another reason to follow this strategy may be that brand posts from the company override user posts by attracting more consumer attention, which in turn leads to more consumer interactions. During the time period studied, the retailer Zara is representative of a low-interaction strategy, with relatively few posts and a low response rate. Potential dangers of such a social interaction strategy are discussed in the next section.

For the six apparel retail brands, Ernsting’s family, Primark, and Zara are prototypical representatives of a low-interaction strategy, posting strategy, and high-interaction strategy respectively. The two behavioral dimensions naturally reflect part of the investment by a brand in its social interaction strategy. However, similar to traditional product and market positioning (Kotler and Armstrong 2014), brands need to integrate a value perspective when applying the framework. To facilitate the integration of a value perspective, research must further clarify the value-creation process of social media and social activities. Researchers generally attribute value to social media by directly creating purchase intention and purchase behavior (Hutter et al. 2013; Moe and Trusov 2011; Phang et al. 2013) and indirectly by consumer co-creation, brand effects, and market research (Barnes 2014; Kabadayi and Price 2014; Wirtz et al. 2013). However, the total contribution and return on investment of different approaches to social media needs further assessment.

Analysis of consumer-initiated social interactions

On April 22nd, the number of user posts showed a significant increase on the retailer brand page of Zara (t = −35.01; p < 0.01). A total of 56 user posts were recorded, in contrast to an average of just below 15 posts per day. Of 56 user posts, 45 refer to a potential trademark infringement by a small company named Zarabumba. The owner of Zarabumba, Ms. Neda Luketic, has consequently acknowledged her intention to rename her business (Belot 2014). For the seven-week time period, this incident accounts for 829 of 1342 (62 %) likes and 213 of 560 (38 %) comments for April 22nd. In contrast to its average response behavior, Zara responded in 11 cases (24 %) with an average response time of 6 h – compared to an overall 0.06 response rate and 15 h overall average response time (see Table 5). Negative user posts and comments continued for days, but subsided in the following weeks.

In the case of Zarabumba, the three social interactions, user posts and the corresponding number of likes and comments, signal a significant deviation. Consumers used the social networking site to express their (negative) opinion on the topic. In the case of a retail brand that is characterized by low social interactivity, such an incident can spread throughout the brand’s network – thus potentially reaching over 20 million Zara fans worldwide. Whereas social interactions on brand posts can provide valuable information on controllable aspects, consumer-initiated social interactions can represent information on topics that affect a brand’s target group. Moreover, by interacting with the brand page, consumers become involved and activated, indicating the relevance of the topic in question. Thus, consumer-initiated social interactions are not only an indicator of consumer displeasure, but also provide insights for market researchers into important topics that affect a brand’s target group.

Discussion

The number of fans is a multiplier of potential reach – and thus the effect – of a brand’s social media strategy. The fan number correspondingly has been described as leveraging the ability to influence friends of fans, increasing their depth of engagement and loyalty, and generating purchase behavior (Barnes 2014; Lipsman et al. 2012). The fan number is not the most suitable metric for identifying potential (social media) problems, as demonstrated by the continuous increase in Zara’s number of fans. If companies record a fan decrease, then the company faces a serious (social) threat. As demonstrated, other consumer social interactions more effectively determine whether social media managers need to react. Besides its importance for leveraging the social media strategy, the fan number potentially provides insights into sales or brand value. A potential explanation may be that as the usage of social networking sites spreads, social interactions represent the behavior of the target groups and thus reflect purchasing behavior or perceived brand value.

The posting and response behavior of brand pages provides insights into social interaction strategies. Posting and response behavior is reflected by the average number of posts per day and the average response rate per day, respectively. Based on these two values, we constructed a framework for comparing different social interaction approaches. For the six apparel retail brands, we identified a prototypical representative for three of four social interaction strategies across the seven-week study period. These three strategies are labeled the low-interaction strategy, the posting strategy, and the high-interaction strategy. No brand adopted a prototypical response strategy focused primarily on responding to user posts. This systematization provides comparative insights for companies, beyond classical interaction measures, such as post engagement and ‘people talk about this (ptat)’, identifying the social interaction strategy of competing brands and the corresponding positioning. Whereas the social interaction framework provides insights into the strategies employed by companies on the social networking site, the framework does not reflect any direct normative instructions, besides a comparison with competing brands. Researchers and managers alike need to add a value perspective to the strategic decision process, thus extending the proposed framework.

The value of consumer social interactions is illustrated further by the case of Zarabumba. For three measures, user posts and corresponding comments and likes, the brand page registered a significant increase. Thus, managers should incorporate social media measures into their monitoring process to identify trends. In cases of negative incidents, managers can enable corresponding reaction strategies (Champoux et al. 2012; Corstjens and Umblijs 2012; Dekay 2012; Hansson et al. 2013). This case exemplifies the fact that the right approach to analyzing social interactions is multi-step. At the first level, social networking sites provide a variety of measures that are readily available, such as the number of fans, comments, and likes. At the second level, these measures need to be considered. As all measures increased in the case presented, an analysis must reflect the sentiment of, for example, user posts and comments. This is in line with previous research (e.g., de Vries et al. 2012; Schweidel and Moe 2014; Tirunillai and Tellis 2012). This study extends this research by establishing insights from a comparative perspective. Thus, consumer social interactions provide information about competitor brand pages, as well as general and industry trends. For example, on April 23rd, 2013, the collapse of the factory Rana Plaza in Dhaka, Bangladesh, unsettled the apparel industry. Several movements, such as Clean Clothes and Fashion Revolution, are constantly aiming at improved labor conditions. The topic still presently affects several apparel retail brands. Another aspect of this second level analysis is the identification of total reach through a social networking site and across different social media. In this study, we also demonstrated the potential danger of such connectivity. Especially in the case of a low-interaction brand, negative user posts can spread rapidly through the brand’s social network. The third level of analysis thus needs to integrate company data, as pointed out by Yadav and Pavlou (2014) in their synthesis for future research. This third level is exemplified by our finding regarding predictive value. In summary, this study found evidence that social interactions have both predictive and comparative value.

Conclusion

Implications

Social interactions on a social networking site provide valuable insights for the own and competing brands. This study analyzed the potential insights gained from observable social interaction metrics across three levels of consumer engagement. The number of fans needs to be observed on a brand’s own page and for competing brands. The fan number may indicate market developments, such as expansion strategies. A decreasing fan number signals an existing problem in the consumer-brand relationship. Firstly, brands should incorporate strategies for reacting to negative social interactions. Some strategies have been proposed in previous research (Champoux et al. 2012; Corstjens and Umblijs 2012; Dekay 2012; Hansson et al. 2013). Secondly, managers need to be aware of how competitors utilize the social networking site to identify consumer expectations. The proposed social interaction strategy framework provides insights into such strategies.

Managers further need to identify individuals who spread brand posts across their (digital) social networks, in order to tap into the full potential of the multiplier effect of a brand’s fan number (Chatterjee 2011; Lipsman et al. 2012; Trusov et al. 2010). Katona et al. (2011) demonstrate that the number of friends is not a good indicator of these influencers; sharing behavior may provide a better measure. A corresponding question for managers and researchers is how to stimulate sharing behavior without undermining the credibility of such brand advocates, as for example, in the case of monetary incentives.

Following consumer expectations of multichannel integration, employing a social networking site as another communication channel further necessitates the integration of business knowledge, supported for example by an extensive customer relationship management system. Brand representatives may not be able to resolve all issues directly, but need to redirect certain problems to another channel, that from a consumer perspective, satisfactorily resolves the issue. From an organizational perspective, this leads more to the internalization than externalization of social networking activities. Consequently, managers need to resolve the resulting organizational challenges.

Limitations and future research

Even though the study was conducted carefully in accordance with and appropriately for its aim, there are inevitably some limitations. For the chosen explorative approach, the data sample across six different-sized apparel retail brands and a 48-day time period certainly provides some insights that can be inferred from consumer social interactions on a social networking site. However, longitudinal and cross-industry approaches are warranted to broaden the findings. The data of the six apparel retail brands yielded a significant correlation between the number of fans and sales volume. We plan to further extend this finding by increasing the number of brand pages and including brand value measures. Similarly, the social interaction strategy framework needs to be expanded.

Future research should also adopt a business value perspective with regard to the four social interaction strategies. For example, Primark represented a prototypical posting strategy across the study period and the smallest retailer, Ernsting’s family, revealed a high-interaction strategy. The fan base of Primark, however, grew comparatively more than for all other apparel retailers, even though this is in line with Primark’s recent expansion in Europe. For research and management, a relevant question is how these social strategies affect and contribute to the economical development. Future research is needed to clarify the drivers and effects of social interaction strategies and to derive guidelines for strategic management decisions.

This study registered data from global brand pages, but no country-specific comparisons were performed. Future research can thus extend the study results by inspecting potential differences across countries and their contribution to a brand’s social interaction strategy.

The focus of the present study was on a single social networking site. Social media is, however, decentralized. Researchers should thus extend their approaches across different social media channels in the future. As social media reaches consumers at different phases of the purchase funnel (Fulgoni and Lipsman 2015). future research could address the interaction between multiple social media across one or more levels of the purchase funnel. Also, as social media drive both online and offline sales (Cao et al. 2014). future research should account for these cross-channel effects.

Consumer-initiated social interactions were found to indicate potential displeasure on the part of consumers. A common saying is that even negative publicity is good publicity. However, researchers need to determine whether this philosophy also applies to social media.