El uso de Twitter en las marcas de moda. Marcas de lujo frente a marcas low-cost

In this paper, we made a comparative analysis about the dialogical capacity of luxury fashion brands and low-cost brands in the social network Twitter. For this, we used a quantitative methodology and conducted a content analysis of 1,135 tweets published by H&M, Zara, Ralph Lauren and Hugo Boss. We observed that there is no consensus among brands to address their audiences, and while low-cost brands make the greatest effort to maintain a dialogue, luxury brands receive more interactions.


55
Enviado: 03-09-2018 / Aceptado: 04-27-2018 The use of Twitter on fashion brands. A comparison between luxury and low-cost brands El uso de Twitter en las marcas de moda. Marcas de lujo frente a marcas low-cost

RESUMEN
En este artículo se analizó comparativamente la capacidad dialógica de las marcas de moda de lujo frente a las denominadas lowcost en la red social Twitter. Para ello, se recurrió a la metodología cuantitativa y se realizó un análisis de contenido de 1135 tuits publicados por H&M, Zara, Ralph Lauren y Hugo Boss. Se observó que no existe consenso entre las organizaciones para dirigirse a sus públicos, y si bien las marcas low-cost son las que hacen un mayor esfuerzo por mantener un diálogo, las firmas de lujo reciben más interacciones.

INTRODUCTION
The rise of digital technology has changed the way in which fashion brands promote themselves and social networks have become a platform that has transformed how this economic sector communicates. Brands, regardless of their position in the fashion pyramid, are present in social networks (Alonso, 2015;Manikonda, Venkatesan, Kambhampati, & Li, 2016). Facebook, Instagram and Twitter are used to promote products and fashion brands, but also, and above all, to establish relationships and create communities of followers. The capacity of bidirectionality, the immediacy or the possibility of having analytics that allow knowing the details of the audience, and even its microsegmentation, are qualities of social networks that fashion brands want to keep using.
The media and social networks offer the possibility of listening to markets and customers in a different, empathic and close way to optimize their dialogical potential. However, we must not forget that technology cannot create or destroy relationships, since it is the way in which it is used that determines the relations between the organization and its audience (Kent & Taylor, 1998). The digital environment requires the creation of spaces that allow the public to dialogue with the organizations and even to comment and share information about them (González Romo & Contreras Espinosa, 2012;Kamboj & Rahman, 2017;Sharma & Shani, 2015). To do so, brands that want to connect with their consumers must know both the multiple devices and platforms they use to communicate and the type of messages to which they react.
Thus, it seems that Twitter has attracted special attention because, according to a study by Metricool, in 2017 it was the social network most used by brands (Santos, 2018), perhaps because it is "an ideal means to manage corporate reputation" (Atarama & Cortés, 2015, p. 27). Therefore, the abundant use of this communication tool by companies justifies its choice to conduct this study.
In this context, it should be noted that how fashion brands communicate in the digital field has been analyzed from different points of view. There has been researches on both the use of social media by organizations (Del Olmo & Fondevila, 2014;Domingo, 2013;Geissinger & Laurell, 2016;Manikonda et al., 2016) and the new formats adapted to the digital environment (Alonso & García, 2016;Caerols & De la Horra, 2015;Del Pino & Castelló, 2017), as well as on the social networks used: Instagram (Segarra-Saavedra & Hidalgo-Marí, 2018), Facebook and Twitter (Alonso, 2015;Arvidsson & Caliandro, 2016;Domingo, 2013;Jin & Phua, 2014). Most of these investigations confirm that brands do not talk to users much and in most cases, limit themselves to share news from the company itself, without creating ad hoc content for the platform.
However, the fashion industry has not been studied in a homogeneous way, since one of the "characteristics that define the research conducted on fashion is the special attention paid to luxury brands" (Segarra-Saavedra & Hidalgo-Marí, 2018, p. 314). Therefore, and despite the fact that in recent years the low-cost fashion sector has grown abundantly (Capell Tuca, 2016), research has not focused on it. In turn, and although Twitter could be considered as an essential communication tool between fashion brands and their stakeholders, the academia has not regularly examined this topic (Navarro-Beltrá, Miquel-Segarra & García Medina, 2017). Given this lack of studies, there is a need to examine the communication made on Twitter by low-cost fashion brands; in addition, it seems especially interesting to compare its results with those of luxury brands, a well-studied industry.
The main objective of this work is to analyze the activity and communication disseminated through Twitter by the main brands of low-cost and luxury fashion. Specifically, we will try to describe the characteristics of the published tweets and analyze to what extent the accounts selected and categorized according to their typology comply with the dialogic principles (Kent & Taylor, 1998;Rybalko & Seltzer, 2010), i.e., the conservation of visitors, the generation of return visits, the provision of useful information and the maintenance of the dialogical loop.
To do so, in the first section of this paper we have referenced the theoretical foundations to address the argument of the investigation. In the methodology section, we describe the strategy used, as well as the selection of the sample and its justification. Then, we present the results obtained. In the analysis, we first expose the characteristics of the tweets disseminated by the selected brands and, subsequently, the result of the variables analyzed in each of the dialogic principles, always offering the comparative vision according to the type of brands. In the conclusion, we state the differences and concomitances in the communicative strategies carried out by the organizations examined. Finally, in the discussion, we question the value that can be derived from the messages disseminated by the brands and the possibilities of increasing the interaction of the participants based on the characteristics of the disseminated tweets.

NETWORKS
The communicative context, and especially advertising, along with the saturation of messages, has transformed the consumer into skeptical, and now he/ she is critical and demanding (Del Pino, 2011), asks to be heard personally and wants to dialogue and interact with both the brand and with other consumers. Thanks to social networks, organizations have ceased to be the only source of information on the brand and the communication process has ceased to be unidirectional (Domingo, 2013) to start the search for a continuous dialogue (Sharma & Sahni, 2015;Simmons, 2007).
In this context, a new consumer profile emerges: the prosumer, a term coined by Toffler (1980) that refers to those users who consume content from social media at the same time that they generate and share it with other consumers (Aced, 2013). In this evolution, the user has ceased to be a passive audience to become a content creator (Alonso, 2015;Cuesta & Alonso, 2010).
At the same time, the new consumer profile is characterized by collecting information through the Internet before making a purchase decision (Del Olmo & Fontdevila, 2014), since he/she attributes great credibility to the opinions of other consumers and users and distrusts traditional communication conducted by organizations (Aced, 2013;Domingo, 2013;Gil & Romero, 2008).

BRANDING AND BRAND RELATIONS
The brand represents a set of functional and emotional values that promise a unique and welcoming experience for its stakeholders (De Chernatony, 2006) and its true success lies in the lasting relationships it creates with its loyal consumers (Ruane & Wallace, 2013). In the era of social networks, the brand becomes a fundamental element that generates value, which intervenes directly in the level of trust and loyalty of the consumer (García & Aguado, 2011). By creating and maintaining loyalty, a brand develops long-term and mutually beneficial relationships with its consumers. Thus, the brand becomes an essential intangible since, not for nothing, loyalty is the most lasting asset of a company and a primary objective of relationship marketing (Lynch & De Chernatony, 2004).
In the digital sphere, companies regularly use social networks to communicate with their clients (García-Fernández, Fernández-Gavira, & Durán-Muñoz, 2015). However, the success of the brand communities depends to a large extent on the continuous participation of the clients (Wu, Huang, Zhao, & Hua, 2015). Therefore, companies must take measures to generate content, interact with consumers and share product information in the community (Kamboj & Rahman, 2017), in order to offer more alternatives and more brand experiences to their consumers (Dru, 2002;Edelman, 2010;Rowley, 2009), since these are an essential starting point to create commitment (Brodie, Illic, Juriec, & Hollebeek, 2013).
Instead of singing its praises, the content disseminated by the brands should focus on convincing their followers that they are useful since they help them make better decisions, as these could be based on adequate information on matters of interest in which the organization has something to contribute (González, 2013). To do so, "it is necessary to offer the consumer an experience of value, which leads brands to establish themselves as referents in those topics of interest related to their activity" (Alonso, 2015, p. 78).

FASHION BRANDS AND SOCIAL NETWORKS
The fashion industry can be classified into three levels depending on the price, production technology and level of exclusivity, namely: haute couture, pret-aporter and mass market or large-scale market (Capell Tuca, 2016). The luxury clothes would be located at the top of the pyramid, while pret-a-porter would be on the second level, with more affordable prices, a higher market volume and a considerable influence on trends (Furukawa, Miura, Miyatake, Watanabe, & Hasegawa, 2017). Low-cost fashion, which is characterized by having the lowest prices (Capell Tuca, 2016), would be in the lower level of the pyramid. Despite this classification, it is increasingly difficult to clearly define these segments, "as a growing number of fashion companies are implementing strategies to expand their businesses or brands in an attempt to attract a broader range of customers" (Posner, 2016, p. 15).
In general, and regarding digital communication, unlike what happens in most retail industries where Facebook maintains its supremacy, in the fashion sector brands connect with different audiences in alternative platforms like Instagram, Twitter and Pinterest (TrackMaven, 2016). The fashion brands try to adapt the content to each of the platforms and to optimize the characteristics and differential advantages of each of them.
In this regard, the microblogging network Twitter has become -due to its high degree of interaction and speed when it comes to sharing information, as well as its ability to achieve greater reach-an indispensable social network. The institutions with Twitter presence have the capacity to achieve a more dynamic communication and a better dialogue between the consumer and the brand. Thus, companies increase their activity on Twitter with the aim of establishing a two-way conversation between the organization and the final recipient (Hwang & Kandmpulli, 2012) and amplify the rapid dissemination of their messages in real time (Castelló, Del Pino, & Ramos, 2014).
Various investigations show that fashion brands (Sharma & Sahni, 2015), and especially luxury brands (Kim & Ko, 2012;Phan, Thomas, & Heine, 2011), have understood that they must be present on Twitter to respond to the needs of a new generation subject to continuous changes. However, as stated, there is still a large field of study regarding this topic.

METHODOLOGY
To conduct this work, we resorted to the quantitative methodology, because "the study of social networks through qualitative methods is still an undefined field" (Alonso, 2015, p. 86). However, the quantitative methodology is considered by academia as a useful method to examine social networks since, in general, research on these communication tools is usually more quantitative than qualitative (Viñarás & Cabezuelo, 2012). Specifically, it is worth noting that previous studies on the fashion industry use mainly content analysis (Segarra-Saavedra & Hidalgo-Marí, 2018). Thus, in order to achieve the stated objectives, we used a quantitative methodology and, specifically, content analysis.
We examined all the tweets published between January 1 and June 30, 2016 by four important clothing brands: two low-cost brands (H&M and Zara) and two luxury brands (Ralph Lauren and Hugo Boss). To conduct this study, we selected a period of six months since it adequately represents the activity of a company in social networks. The data collection took place in July 2016; therefore, we chose the six previous months to analyze the information.
We selected the brands mentioned above since they were the only four that appeared in the apparel category of Interbrand's (2015) The Best 100 brands 2015 ranking. After identifying the brands, it was necessary to choose the specific Twitter profile we were going to study: we discarded those accounts specific to certain countries and chose the international profiles that included the "verified account" identification. Table  1 shows the main characteristics of the profiles analyzed.
The computer program used to obtain the specific tweets to be examined was Twitonomy, a web tool that allows downloading tweets, retweets and responses from Twitter profiles. Thus, after selecting the profiles and the dates indicated above, we obtained a total of 1135 tweets. Of these, 529 were published by low-cost brands (125 by Zara and 404 by H&M) and 606 by luxury brands (346 by Hugo Boss and 260 by Ralph Lauren).
The totality of these tweets was coded and analyzed to conduct this study through a coding protocol created specifically for this work. However, and regarding the validity of this instrument for gathering information, it is important to indicate that the variables used were the result of an adaptation of previous investigations. Specifically, the study is based on the dialogic principles applicable to the Internet established by Kent and Taylor (1998) and later adapted for Twitter by Rybalko and Seltzer (2010). Thus, the dialogical principles analyzed in this paper are: • Usefulness of information: providing value information to stakeholders is essential to maintain a dialogical relationship, since the public relies on institutions that provide useful and reliable information (Kent & Taylor, 1998). Due to the shortness of the publications made on Twitter, the most appropriate way to include this type of information is through links. Thus, we analyzed the links and sites to which they refer.
• Conservation of visitors: in Twitter, there is a danger of including links that lead followers to browsing spaces not related to the organization (Kent & Taylor, 1998), hence the importance of using corporate links, i.e., those that redirect to other websites managed by the brand (Rybalko & Seltzer, 2010). In addition to the type of link, it is important to regularly publish tweets to try to maintain the number of visitors (Rybalko & Seltzer, 2010).
• Generation of return visits: institutions have the need to include information that is attractive to their audiences, able to encourage them to visit again (Kent & Taylor, 1998). Therefore, we analyzed the use of labels and mentions, since these elements favor engagement and enhance the engagement between the company and its customers.
• Maintenance of the dialogical loop: the existence of a dialogue between brand and follower is possible in social networks, since the public can ask questions to the organizations and these can answer their doubts, concerns and problems (Kent & Taylor, 1998). Thus, we analyzed if the brands talk with their stakeholders by posing questions in their tweets or responding to comments or questions made by their followers (Rybalko & Seltzer, 2010). Since they involve a direct interaction between the tweet and the public, we also considered retweets and "likes".
Finally, the coding protocol used in this study consisted of 48 variables that, in addition to numbering the tweet to facilitate its identification, collected information on the features of the brand 1 , the text, the URL, the distinctive features of the tweet 2 , mentions and hashtags 3 , links 4 , the number of retweets and "likes", and the existence and number of questions and comments made by the company and its followers. To conduct this research, we defined all these variables and the categories to be studied to the maximum, trained the coders and carried out a previous test so that the coding protocol had a high degree of reliability, according to the recommendations established by Wimmer and Dominick (1996).
The coding protocol was accompanied by an instruction manual and we held several sessions to put these documents into practice and learn to use them appropriately. The possible discrepancies related to the coding of the variables that arose in these practical sessions were resolved through discussion and consensus. After conducting these sessions, we conducted a trial in which we obtained satisfactory results. Thus, the researchers in charge of data collection independently analyzed the same subsample of 15% of the tweets to be studied -i.e., 170-and obtained an average Cohen's Kappa index of 0.975 5 . This data shows that intercoder reliability was adequate.
Finally, and to examine the information collected, we used the SPSS software version 21; specifically, we resorted to descriptive, frequency, multiple response tables, line figures and Pearson's chi-squared test.
Regarding this last statistic, when the asymptotic (bilateral) significance is lower than the established alpha level (0.05) 6 and the Pearson Chi-square value for the fixed probability level (0.05) and the degrees of corresponding freedoms are higher than the value indicated in the Chi-square distribution table (Wimmer & Dominick, 1996), it is considered that there is a statistically significant relationship between the variables examined (Pita & Pértega, 2004). Thus, the alternative hypothesis is accepted (Ha = yes there is an association between variables) and the null hypothesis is discarded (H0 = the variables are independent) (Pita & Pértega, 2004). However, we cannot forget that in some occasions they were tables of 2x2. In these cases, it must be considered that when more than 20% of the boxes have an expected frequency of less than five, Fisher's exact test (Pita & Pértega, 2004) must be used, in which the test value cannot be considered, but the value of the variable p can be considered.

TWEETS' FEATURES: LANGUAGE, PLATFORM USED AND TYPE OF TWEET
Before analyzing the dialogic principles, it is interesting to examine the particularities of the publications studied: English is the language most used by all types of brand to write their tweets; however, there are statistically significant differences regarding this issue (x 2 =19.771; p=0.003), since low-cost firms are the only ones that use other languages. Specifically, they published six tweets in French (1.1%), four in Spanish (0.8%), two in Italian (0.4%), two in Romanian (0.4%) and two in Greek (0.4%).
The platform used to publish the tweets also shows statistically significant disparities depending on the type of brand (x 2 =653.008; p=0.000). The lowcost ones mainly use Adobe Social (63.7% of their publications), while the luxury ones do not use it at all and opt for Twitter Web Client (63.2%), which is the second platform most frequently used by the cheapest firms (19.3%). On the other hand, brands with higher prices also use, as second option, Twitter for iPhone (23.9%), although this is only used by low-cost companies in 5.5% of their publications. In addition, it is worth noting that Hootsuite, Periscope and Twitter Ads are only used by luxury firms (7.9%, 0.3% and 0.3% respectively) (table 2).
When analyzing the nature of the publication, we can also observe statistically significant differences depending on the type of brand (x 2 =110.497; p=0.000). Thus, even though it is most common to resort to own tweets (90.9% in the case of luxury firms and 84.3% in low-cost firms), in certain occasions the cheapest brands publish answers (13.8%), but it is not usual to include retweets (1.9%). However, the opposite situation occurs in the case of companies with a higher price, since they do not resort to answers on any occasion, but 9.1% of their publications are retweets.

PRINCIPLE OF PROVIDING USEFUL INFORMATION FOR THE AUDIENCE
It is common for both types of brand to include links in their tweets; in fact, 96.7% of luxury brands and 97.7% of low-cost firms do so, without statistically significant differences.
However, such disparities do occur when considering the number of links contained in each tweet (x 2 =304.912; p=0.000): brands with a higher cost usually include only one link (84.2% of cases), while the cheapest prefer to incorporate two (62%), and they are the only ones that have posted three links, although only on one occasion (0.2%).
As can be seen in table 3, when the brands -both low cost and luxury-use links, they refer mainly to images (70.3% and 46.3% of the links, respectively). However, low-cost firms also link websites (44.9%). In this regard, only companies with higher prices post links to Periscope (0.6%), blogs (1.2%) and media (2.4%) (table 3).

PRINCIPLE OF CONSERVATION OF VISITORS
When examining the frequency of publication, we can observe that, although the differences are not striking, luxury brands publish more tweets than the low-cost ones (53.4% of the total of analyzed publications versus 46.6%). Thus, the former has a daily average of 3.3 tweets and the second ones, of 2.9.
However, there is no homogeneity when considering the publication date, since there are statistically significant differences depending on the months studied (x 2 =18.034; p=0.003). These differences occur even though February is the month in which both types of brand publish the most tweets (23.1% in the case of luxury brands and 21.9% in the low-cost ones) and that January is the month in which they do it the least (13.5% considering the most expensive organizations and 11.3%, the cheapest). Nevertheless, low-cost companies also tend to publish in April and May (20.2% and 19.5%, respectively) and not in March (11.5%), although luxury firms do post in that month (17.7%) (figure 1).
There are also statistically significant differences when analyzing the day of the most prolific month (x 2 =65.321; p=0.000). Thus, and as shown in figure 2, low-cost brands prefer to publish at the beginning of the month. Specifically, they usually prefer the first four days, since 7.9% of their tweets are posted on day 2, 6.2% on day 3, 5.7% on day 1 and 5.3% on day 4. However, luxury firms publish more frequently in the middle of the month, especially on day 18 (7.4%), although the 29 also has a high frequency of publication (5%) (figure 2).
When examining whether the links lead their followers to websites related to the company, it should be noted that, in general, most of them redirect to corporate sites. Thus, 96.9% of the links of the lowcost brands and 85.8% of the links of the luxury firms are corporate.

PRINCIPLE OF GENERATION OF RETURN VISITS
Regarding the use of mentions, the differences that occur between the types of brands are statistically significant (x 2 =84.957; p=0.000): the luxury ones do not include mentions in 50% of their publications, compared to 69.4% of the low-cost. Thus, although the most usual is only one mention per tweet (30.4% in the case of high-value firms and 27.6% in the cheapest ones), we cannot omit that in three occasions (0.5%) luxury brands included up to five. In this regard, it should be noted that low-cost companies included a total of 181 mentions, a figure that increases to 463 in the case of firms with higher prices. Hashtags, on the other hand, are used more frequently. Luxury brands resort to them in 85.6% of tweets, while low-cost brands do so in 82.4%. The former most commonly include a hashtag in each publication, followed by two (35.5% and 34.8%, respectively). In the case of low-cost companies, the inverse occurs, i.e., it is more frequent to use two hashtags in each tweet (40.5%) than one (34.2%). Although these differences are statistically significant (x 2 =23.436; p=0.000), both types of brand post a maximum of four hashtags in a single publication (0.7% in the case of luxury organizations and 1.3% in the case of lowcost ones). Firms with higher prices included a total of 920 hashtags and the cheapest, 739.

PRINCIPLE OF MAINTENANCE OF THE DIALOGUE LOOP
Although it is not usual for brands to pose direct questions 7 in their tweets, there are statistically significant differences regarding this issue (x 2 =63.736; p=0.000), since luxury firms propose this type of interaction with their audiences in 2.5% of the occasions, while the low-cost ones do it in 15.9%.
Despite the above, it is common for followers to interact with the organizations studied. Specifically, there are audience comments in 80.3% of the tweets of the cheapest brands, although this figure drops in the case of companies with higher prices (66.5%), which leads to statistically significant differences depending on the type of brand (x 2 =27.412; p=0.000).
The number of followers' interactions also shows statistically significant differences depending on the type of brand (x 2 =68.717; p=0.000). The average of comments obtained in the publications made by the companies that have a higher price is 3.35, compared to 3.09 in the case of the cheapest ones.
Although usually there are not direct questions from the audience in these comments, the differences that occur are also statistically significant (x 2 =58.046; p=0.000). Thus, there are only questions made by followers in 9.6% of the tweets posted by luxury institutions, while this figure increases to 26.8% in the case of low-cost companies.
Organizations do not usually interact with their followers again as a result of a tweet. In fact, the brands that have a higher price did so only once (0.2%), although the cheapest ones resort to this type of interaction in 16.3% of cases. Thus, these differences are statistically significant (x 2 =103.340; p=0.000).
The number of retweets obtained by the publications analyzed also shows statistically significant differences (x 2 =350.738; p=0.000). The luxury firms are the ones that obtain the highest number of retweets, with an average of 96.24 per publication, while the low-cost ones obtain an average of 71.14. This last type of brand is the only one that has publications without any retweets, specifically 17 (3.2%) of their publications, while luxury companies have a tweet with 4,922 retweets (this is, in turn, a retweet about a publication by Kim Kardashian in which she talks about Ralph Lauren).
All analyzed tweets have been marked at least once with the "like" symbol. The Ralph Lauren's retweet to Kim Kardashian mentioned above is noteworthy, as it has a total of 24,698 "likes". In this regard, and contrary to what happens in the case of retweets, it is the publications made by low-cost brands that have a greater number of "likes", since their average is 288.28 per tweet, compared to 282.14 in the case of luxury companies. These differences are statistically significant (x 2 =785.042; p=0.000).

CONCLUSIONS
The data obtained indicate that the fashion brands examined differ in their Twitter strategies to address and interact with their followers. In fact, the differences found between luxury and low-cost brands are usually statistically significant. We can affirm, however, that both types of brands use Twitter to disseminate content unilaterally and do not take advantage of the dialogic potential of social networks.
In this regard, we can observe that the brands studied seem to comply with the principle of useful information, since they usually facilitate and disseminate links of interest to their audiences, an essential requirement to share information in a social network as Twitter. Thus, both types of brand redirect mainly to images, as these are of vital importance for the fashion sector. Nevertheless, low-cost firms are those that include a greater number of links per tweet, which could indicate their greater willingness to comply with this principle.  As these links are mostly corporate for both types of brand and have and appropriate frequency of publication -on average, three daily tweets-they also promote the conservation of visitors. However, it seems that firms with higher prices try to encourage with more assiduity than the cheapest ones the generation of return visits, since they include more mentions and hashtags.
Although the companies studied do not usually pay attention to the dialogic loop, it should be noted that low-cost brands seem to be more aware of this issue than luxury brands. Thus, the former interacts more with their followers, asking more questions in their tweets and answering the comments in a greater number of occasions than the latter, although this frequency is still very scarce. However, luxury firms have a higher average of retweets, although not of "likes".
Therefore, and despite the differences found between the types of brands examined, we can conclude that, although Twitter has a broad dialogical potential, fashion companies profit from it very little.

DISCUSSION
Although it is common for fashion brands to be present on social networks (Alonso, 2015;Manikonda et al., 2016), we cannot overlook the fact that their high exposure to platforms 2.0 makes it difficult to control the communication generated on them (Alonso, 2015). Indeed, usually organizations do not to respond to the comments made by their followers; therefore, they do not take full advantage of the potential offered by online communication to interact with their stakeholders, either through websites (Kent, Taylor, & White, 2003;Park & Reber, 2008;Taylor, Kent, & White, 2001) or social networks (Rybalko & Seltzer, 2010).
The luxury fashion sector is noteworthy. Although digital marketing is considered one of the most used tools by this industry to promote brand awareness, interaction with consumers and knowing their purchase motivations (González Romo & Plaza Romero, 2017), the results of this work demonstrate that low-cost brands are the ones most concerned about the dialogue with their followers. Considering this situation, we wonder if it is possible that, as it would be expected, the studied luxury brands connect emotionally with their followers, as there is no interaction.
On the other hand, it seems that publications on social networks involving celebrities and influencers are the more attractive and the ones with which users interact the most, since they obtain a greater number of "likes" (Manikonda et al., 2016). In fact, influencer engagement is considered as an especially effective tool to increase the notoriety of a brand (Del Pino & Castelló, 2017). Given this situation, organizations usually direct their communicative efforts to influential people to reach more potential consumers (Castelló & Del Pino, 2015), since one of the most performed activities on social networks is following influencers (IAB Spain, 2017).
The importance of this type of activity lies in the credibility of the influencers: the defense of a certain brand does not depend on an economic retribution but on a sincere recommendation towards certain products and services (Del Pino & Castelló, 2017). Thus, luxury fashion brands often resort to bloggers and instagramers to try to influence their audiences (González Romo & Plaza Romero, 2017).
In this regard, it should be noted that we can all be influencers, since any user satisfied with a product or service can share this information with his/her followers through social networks (Castelló & Del Pino, 2015). In addition, we cannot forget that on certain occasions it is the users themselves who may be interested in publishing information about a brand and mentioning it to gain reputation or recognition in social networks (Arvidsson & Caliandro, 2016).
Despite the importance of the information obtained with this study, its main limitation is having analyzed only the communication made by fashion brands through a single platform (Twitter). Therefore, these companies may or may not be taking better advantage of the dialogic potential offered by information and communication technologies. This opens a possible line of research focused on examining the communication of fashion brands on other digital platforms, such as Instagram or Facebook.