Touch, Swipe or Click?: Understanding information exchange (eWOM) on Instagram and how it can be encouraged

(1)

Touch, Swipe or Click?

Understanding information exchange

(eWOM) on Instagram and how it can be

encouraged

Master’s Thesis 30 credits

Department of Business Studies

Uppsala University

Spring Semester of 2017

Date of Submission: 2017-05-30

Vanessa Auer

Evelina Bergström

(2)

ABSTRACT

Electronic word-of-mouth (eWOM) is known as one of the most influential factors impacting consumer behaviour. Yet, not all consumers are motivated to like, tag or comment on commercial posts online. Therefore, marketers need to capture consumers’ attention and encourage eWOM creation (interaction) through the content features (visual, text and audio) of their brand posts. However, currently as many as 91% of marketers lack an understanding of what content actually drives people to interact (Stelzner, 2015). Thus, this study’s aim is to explore, using a survey and an in-depth experimental study, which content forms engage and encourage users to create eWOM on Instagram. The results reveal that users seldom interact with commercial posts, which pinpoints the difficulties for brands on Instagram and thereby the need for a different content strategy. However, when posts connect to users’ intrinsic interests and display content forms that are in line with users’ characteristics, individuals find it more engaging and consequently create eWOM. Specifically, findings reveal that employees (27-29 year olds) are less negative towards commercial posts and more likely to create eWOM than students (18-26 year olds). In regards to the content forms, a post should include; short text/videos, emotion-driven emoji, maximum three hashtags, non-static and contrasting backgrounds, and verbal audio rather than music.

Keywords: Electronic-word-of-mouth (eWOM), Word-of-Mouth (WOM), Engagement, Content Features, Social Networking Sites (SNS), Instagram, Social Media Marketing

(3)

ACKNOWLEDGEMENTS

First and foremost, a sincere gratitude is directed towards our supervisor, Leon

Michael Caesarius, for continuous support and supervision during the process of this study. The guidance has provided useful insights into how to develop the study in new directions, as well as how to improve its quality.

Additionally, a sincere gratitude is directed to all respondents, whose time and commitment has enabled the study to achieve its purpose. Without your effort, it would not have been possible to complete this research.

Evelina Bergström & Vanessa Auer 30-05-2017, Uppsala

(6)

page 1

1. INTRODUCTION

1.1 Background

Dinner time – it used to be a time when people sat down to eat and talk about their days, because they had not seen each other since breakfast. Now, we are constantly connected via phones, social media and apps, which has led us to update friends and loved ones on our plans, activities and what has happened instantly rather than waiting until we see them. This change in information exchange connects to that in the past 20 years, the world has seen a whirlwind of technological innovations every year (Roser, 2016) and with every new tool, product or service, people’s behaviour changed (Cantallops & Salvi, 2014). Why though did we, as consumers, change our behaviour? Based on Cantallops and Savi’s (2014) argument that consumer behaviour has changed, due to changes in technology and information

exchange, it can be argued that understanding how people behave and touch, swipe and click on social media is crucial for marketers today. This is because otherwise marketers might miss the window of opportunity to connect with us, and it is important to note that we as consumers can love, hate or end you within minutes.

This change in information exchange can further be see in how people share their experiences with others about an advertising or marketing campaign. That is, whilst in the past, people mainly shared their consumption experiences through the act of word-of-mouth (WOM) (Hennig-Thurau et. al., 2004), today more than three billion people like, tag and comment about brands, products and services online and are engaged in what is known as

electronic-word-of-mouth (eWOM) (Rosario et. al., 2016). From a marketer's perspective,

eWOM is highly desirable (Hautz et. al., 2013; Hennig-Thurau et. al., 2004), since it is widely known that consumers´ information sharing has the potential to influence other consumers´ decision-making processes, including the process of purchasing intention

(Malthouse et. al., 2016), and the image and knowledge one might have of a brand (Hautz et. al., 2013). Yet, not all consumers are highly motivated to interact (create eWOM) and

marketers need to differentiate the content (i.e. visual, audio and text) in an ad, campaign or brand post online to encourage interaction and information exchange, as well as draw people in. The ability to be drawn in or be attracted to something, is in this study referred to as

engagement (Chapman, 1997), and can be argued to correlate with a person’s intrinsic

(7)

page 2 useful value in something) (O’Brien, 2010). Consequently, this can stimulate positive or negative feelings (O’Brien & Toms, 2008; Kim et. al., 2013) that a person wants to share with others, and thus can impact other’s behaviour and willingness to, for example, purchase a specific brand (Richins & Root-Shaffer, 1988; Engel et al, 1969; Gilly et al, 1998).

Acknowledging the wide impact eWOM has, and subsequently highlighting the significance of engagement and content features, one can argue that from a marketer’s perspective the understanding of these three concepts and how these can stimulate reactions that users want to share, is desirable.

However, as many as 91 percent of marketers currently lack an understanding of what content actually drives people to interact (Stelzner, 2015). This issue intensifies with the awareness of how information exchange has evolved in the online sphere. In this online environment, Social Networking Sites (SNS) such as Facebook, Instagram and Twitter, serve as important and powerful platforms for this form of information exchange (eWOM). One of the newest and fastest-growing SNS is Instagram (Sheldon & Bryant, 2016), which provides a unique mobile platform, where people can share pictures and videos with the general public (Zhu & Chen, 2015; Instagram, 2017). People can also follow other people or brands, and share their experiences and opinions on this platform (Carah & Shaul, 2016; Jaakonmäki et. al., 2017). Therefore, Instagram plays a key role for marketers today, due to the fact that

young users (i.e. individuals that use (Oxford Dictionaries, 2017) social media platforms) are

most active on this mobile platform (Salomon, 2013). This is evident in that the majority of Instagram’s 600 million users (Statista, 2017) are women and men between the ages 18 to 29 (Statista, 2016). Individuals between these ages, can also be categorized into Millennials and Generation Z (Prensky, 2001; Schneider, 2015; Salomon, 2013; Statista, 2016), and are not only Instagram’s biggest demographic, but also the Internet’s (Statista, 2015). Even though certain characteristics can be generalized for each generation (Tadajewski et. al., 2008), every individual is different and thus one generalized strategy will not work when wanting

everyone to interact. Nevertheless, Instagram offers the most effective mobile platform to investigate and understand these young individuals; what their various needs are and how these can be addressed in order to impact and encourage eWOM.

1.2 Problem

Yet, people’s needs and information exchange behaviour do not come with a clear road map, but are quite more complex. Even though previous researchers have investigated

(8)

page 3 why individuals interact with eWOM and how it captures their attention on well-established SNS, such as Facebook and Twitter (Bakhshi et. al., 2014; Shang et. al., 2016), few have analysed which forms of content features foster eWOM (Bakhshi et. al., 2015; Jaakonmäki et. al., 2017; Russman & Svensson, 2016; Carah & Shaul, 2016). Specifically, few have

investigated what form of visual (e.g. background and filter) (Jaakonmäki et. al., 2017; Bakhshi et. al., 2015), audio (e.g. music or verbal sound) (Ebie, 2004), or text (e.g. emoji and hashtags) (Jaakonmäki et. al., 2017; Jang et. al., 2015) draw people in or encourage them to create eWOM on Instagram.

Additionally, Instagram is a new form of mobile SNS, which differs from its bigger SNS brothers in terms of user characteristics (mainly reaches the younger generations between 18-29 vs. all ages) (Abbott et. al., 2013); consumer wants (sharing a moment vs. engaging in multiple ways); platform design (only mobile vs. traditional SNS); and content (mainly visuals with little text and clutter vs. a lot of text and clutter) (Sitkins, 2016). Thus, the same forms of content cannot be used on all SNS (Kietzmann et. al., 2011), as well as its effect on eWOM creation can be argued to be different than on established platforms. This highlights the aspect of that information exchange has changed and new SNS have to be regarded differently.

In connection, in order to understand and target consumers, previous researchers and marketers have for long relied on the segmentation of different consumers in accordance to predefined generational brackets (Tadajewski et. al., 2008). However, even though

Instagram’s and the world’s biggest generations, Millennials and Generation Z (i.e. people born between 1977-2012 (WJSchroer, n.a.)), have been displayed to be similar in respect to being described as Digital Natives (Prensky, 2001) and iGeneration (Schneider, 2015), individuals should be understood on an individual basis, rather than being categorized based on generational generalizations. This is due to that other aspects, such as an individual’s interests, occupation, age and gender are factors that should be considered in understanding Instagram users and their individual tendencies to create eWOM (Tadajewski et. al., 2008). Consequently, this study will take these characteristics into consideration when analysing users’ reactions and intentions to interact.

Moreover, previous content features, engagement and eWOM research has not yet investigated the effect these three aspects have on each other. Even though eWOM is said to derive on SNS from content that displays what the person wants to be associated with (Kim

(9)

page 4 et. al., 2015; Russman & Svensson, 2016; Goffman, 1959), it has not yet looked into which content forms are needed in order to encourage eWOM after it has attracted a person to the post. Nevertheless, an investigation into what forms are appealing and encourage eWOM on Instagram is a crucial topic to look into.

1.3 Purpose

Therefore, the overall purpose of this study is to examine what forms of content features (e.g. visual background or verbal audio) are important when wanting to encourage eWOM on Instagram. Specifically, the study aims to explore, with the use of an extensive pre-study and experimental in-depth study, which forms of content features appeal to individuals (i.e. engagement), as well as to explore which aspects encourage users on Instagram to interact (i.e. create eWOM). By observing which content forms grab users’ attention and which encourage them to interact with each other, this study cannot only provide marketers with insights into how to successfully cater to young consumers’ needs, but also add to existing literature.

First and foremost, this study is important to researchers, because it provides new insights into people’s behaviours on a rather limitedly-studied social media platform. Due to that Instagram does not have the same principal layout as other SNS, such as Facebook and Twitter, researchers can through this study understand how the behaviour of people has changed on SNS as well as how interaction and exchange of information has transformed. Therefore, this study intends to provide researchers with insights into these areas, as well as provide a new perspective on which theoretical concepts are involved on visual SNS, and how they are interlinked. By showing this interlinkage, this study expects to show what content forms encourage eWOM on Instagram, and therefore provide researchers with new understandings into how eWOM on SNS has changed and the direction of its evolution.

Secondly, due to that every individual is unique, this study aims to demonstrate how important it is to understand each individual’s behaviours and needs. Specifically, this study intends to highlight what content forms are of relevance to certain people, and thus show to researchers that new insights and approaches are needed when wanting to cater to consumers’ needs in the 21st century.

Thirdly, having a deeper understanding of what forms of content drive engagement and eWOM on Instagram, might be desirable and important from a marketer's perspective,

(10)

page 5 since greater eWOM can potentially influence other individuals ́ purchasing decisions, their perception of brands, companies, products and services (Hennig-Thurau et. al., 2004, Rosario et. al., 2016; Hinz et. al., 2011). Therefore, this study expects the outcomes, in regards to the various influences content forms have on engagement and eWOM, to expand the

understanding of these relationships on social media. Due to that both engagement and eWOM relate to hedonic and utilitarian behaviour (i.e. act of seeking respectively sensory arousal, fun and enjoyment versus efficiency, practicality and to solve problems (Hirschman & Holbrook, 1982)), marketers can gather new insights from this study, in terms of how eWOM is created by certain content, which forms attract people, and how they can encourage more eWOM. Grasping these relationships is of utmost importance, because, as previous research into eWOM’s effect on purchase intention (Shang et. al., 2016; Kim & Johnson, 2016; Rosario et. al., 2016; Chao & Chen, 2016; Kiran & Vasantha, 2016) has shown, understanding the content forms that lead to eWOM is advantageous. Moreover, to effectively understand how eWOM today is achieved, this study will analyse participants’ intention to interact on Instagram. This analysis therefore intends to provide marketers with insights on which content forms are needed for certain individuals to create eWOM.

Additionally, this study’s outcomes are expected to provide marketers with insights about how they can change their posts in regards to diverse user needs, and therefore adapt a different strategy than on established SNS, due to Instagram’s differentiations.

In order to answer these questions, the study will firstly present the literature review including previous findings within the research streams of eWOM, engagement and content features, and conclude with a depiction of the study´s analytical framework. Secondly, a concrete methodology, explaining the practices of the pre-study and the experimental in-depth study, its sampling techniques, measurement methods and administrational procedures will be described. Thirdly, the results will be presented, followed by the analysis, in which individuals´ propensity to react and interact in relation to different content forms will be analysed. At last, the study will present the conclusions, limitations and future implications, which will answer the research question exhibited below.

1.4 Research Question

Which forms of content features effect users eWOM creation and engagement on Instagram?

(11)

page 6

2. LITERATURE REVIEW

2.1 Electronic Word-of-Mouth (eWOM)

We all do it – we exchange information today via a touch, swipe or click on SNS, which depicts even how eWOM has changed since its beginning (Hennig-Thurau et. al., 2004). Based on Hennig-Thurau et. al.’s (2004) and Kietzmann and Canhoto’s (2013)

definition, this study defines eWOM as any statement, tag or like that is based on positive or

negative emotions made by potential, actual or former consumers about a brand or product, as well as created by SNS users about other users, and made available to a multitude of people and institutions online. eWOM’s significance in today’s corporate and academic

world, even though it is seen as less personal than traditional WOM (Hennig-Thurau et. al., 2004; Kietzmann & Canhoto, 2013), is visible in regards to that it is an effective marketing tool used by numerous organizations (Yeh & Choi, 2011), and a phenomenon of how people interact and exchange information in the 2010’s. The effectiveness of eWOM stems from its characteristic to overcome time and space (Gretzel et. al., 2006; Dwyer, 2007; Yeh & Choi, 2011) and quickly and effortlessly spreads information across the whole Internet at the speed of light (Yeh & Choi, 2011). This is often facilitated through interpersonal communications (Chu & Choi, 2011; Meuter et. al., 2013), which has with the help of SNS increased, because people’s networks have augmented in size (Wilson et. al., 2012;Kim et. al., 2015). This expansion of social networks can be argued to be the result of the Internet, where one can “friend” someone he/she has never met, and who lives on opposite sides of the world.

Information sharing is a crucial component of today’s SNS (Yeh & Choi, 2011), yet most communication happens between individuals, and not between businesses or businesses and individuals (Kietzmann & Canhoto, 2013; Kim et. al., 2015). Why is this? Duhan et. al. (1997) argue that a personal connection, between two individuals makes information more authentic and trustworthy. Thus, people’s non-commercialized knowledge and opinions (Chu & Choi, 2011), can be a great source for others, when evaluating alternatives (Gretzel et. al., 2006). However, this interaction highlights the difficulty for firms on popular SNS, due to that users prefer not to interact with them but rather with other users. Yet, seeing other users’ input as a source, connects to one of Chu and Choi’s (2011) eWOM behaviours: opinion

seeker (Feick et. a., 1986). In addition to opinion seeker, Chu and Choi (2011) present two

more behaviours, namely opinion leader and pass-along behaviour. Opinion leaders generate information, whilst pass-along behaviour incorporates a user’s ability to forward information

(12)

page 7 to other people in his/her online network (Chu & Choi, 2011) with the ease of tagging or liking. Opinion leaders and the notion of passing-along information can be argued to connect to how a user wants to be portrayed on SNS (Kim et. al., 2015; Russman & Svensson, 2016; Goffman, 1959). The motive behind why people want to be portrayed in certain ways (Shang et. al., 2016), connects to their self-image, namely something that is in-line with how the user sees him- /herself (Kim et. al., 2013; Sirgy & Sun, 2000), and the act of self-enhancement, where something enhances a person’s self-esteem or appears to display more expertise than he/she has (Wu & Wang, 2011; Wojnicki & Godes, 2008). Both these aspects are said to be important drivers on SNS, such as Instagram. In order to understand what drives individuals on Instagram, one has to not only understand the platform, but also what content users want or need to interact (Laurent & Kapferer, 1985; Wu & Wang, 2011). This refers to if a user sees the content created to be useful or pleasurable (i.e. can gather hedonic or utilitarian value from it). Yet, if the content of a post does not provide any value to an individual, then he/she is unlikely to engage with it or create more eWOM (Wu & Wang, 2011; O’Brien, 2010).

2.1.1 Engagement

In connection, understanding what various needs and wants users on Instagram have and what content appeals to them, can be argued to be of significance. This is due to that not everyone finds the same content interesting. For example, a picture with a black background and L’Oréal Paris’ Matte Addiction Lipstick, with hashtags, a product description, and a heart emoji, will not attract the attention from everyone on Instagram. Thus, the importance of how to foster an engaging environment on Instagram, where individuals can freely express

themselves and connect with a post, in form of eWOM, has not yet been successfully done by many marketers (Stelzner, 2015). According to O’Brien and Toms (2008) successful

engagement, that is our ability to be drawn in, find something appealing or be attracted to something (Chapman, 1997; O’Brien & Toms, 2008), can only be achieved by looking more

closely at the concept itself. O’Brien and Toms’ (2008) model can be divided into four stages, namely (1) point of engagement, (2) engagement, (3) disengagement, and (4) re-engagement (O’Brien, 2010). The first stage, point of re-engagement, can be argued to be of value to this study, because in order to achieve an engaging environment, one first has to capture users’ attention on Instagram (Carah & Shaul, 2016; Jaakonmäki et. al., 2017). This O’Brien (2010) argues can be done with aesthetically-pleasing content, which people want to invest themselves in. This investment connects to the reaction and behaviour of people,

(13)

page 8 which are vital in understanding why this stage occurs (O’Brien, 2010) on Instagram.

O’Brien and Toms’ (2008; Kim et. al., 2013) state that the attributes, attention, curiosity, intrinsic interest and motivation, are key in the point of engagement, due to that these attributes can influence how people will interact with a post (O’Brien, 2010). If a post, for example, appeals to a person’s intrinsic interests (e.g. women’s rights), then he/she is likely to want to associate him/herself with it (Shang et. al., 2016) through eWOM, compared to a post that is of no interest to him/her. Therefore, these attributes can be argued to connect to the concept of hedonism and utilitarianism (O’Brien, 2010), which both relate to the various ways a new Instagram post can be appealing.

This connects to Katz’s (1959) User and Gratification (U&G) theory, where people engage differently depending on their gratification needs, which according to Levy and Windahl (1985), can be divided into social and psychological needs. Consequently, different needs are associated with different personalities, ages, backgrounds and social roles. Other researchers have further divided these gratification needs into cognitive, affective, personal integrative, tension release and social integrative (McQuail, 2010; Katz et. al., 1979). More precisely; people might engage because they have a cognitive need to acquire information for a better understanding, or because they have an affective need for entertainment. This

connects to O’Brien’s (2010) hedonic and utilitarian value argument. Furthermore, Shang, Wu and Sie (2016), use these arguments to explain the reasons behind consumer resonance on SNS and the impact these have on a consumer's intention to buy. Even though Shang, Wu and Sie (2016) define consumer resonance as only involving users ́ positive and supportive reactions to a particular post, they also highlight that users’ engagement and eWOM can be both positive and negative in nature. The results of their study show that consumers are more likely to respond, that is positively engage with a post, when: (1) the information is perceived as useful (utilitarian value); (2) the information is posted by someone the person has a

relation with (tie strength); (3) the person has a need of being accepted and seeks approval from other members (normative influence); (4) the person wants to obtain more useful information from others (informational influence); and (5) the person has an intensified need to emphasize his/her self-image (self-image). All of this occurs in a cognitive continuous state in every individual (Kim et. al., 2013), and thus highlights the difficulty of fostering an engaging environment for everyone on Instagram.

(14)

page 9

2.2 Social Networking Sites (SNS) and Content Features

All current online exchange of information embraces the actions of touching, swiping, clicking, tagging, liking and commenting, which we, as users, will continue to do every single day, and thus researchers and academics need to comprehend its significance. In order to simplify the various platforms that exist online, one can refer to Constantinides and Fountain’s (2008) five social media categories, which are namely blogs, SNS (e.g.

Facebook), content communities (e.g. YouTube), e-forums, and content aggregators. SNS are of utmost importance, since most consumers are present in these networks, as well as they communicate with friends and others with similar interests on these platforms. Thus, they can form communities, relationships and portray themselves how they want to be seen, based on the eWOM they create (Shang et. al., 2016; Kim et. al., 2013; Wu & Wang, 2011).

Previous research has mostly focused on Facebook and Twitter, due to the size and popularity of the platforms (Hu et. al., 2014). However, another SNS, which keeps on growing in popularity, is frequently used by organizations and is one of the fastest-growing SNS at the moment, is Instagram (Sheldon & Bryant, 2016). Instagram is a visual mobile storytelling platform, where users can upload and share photos and videos instantly with friends or other users (Instagram, 2017). Even though it is growing in popularity, few academic researchers have analysed this social media channel (Carah & Shaul, 2016; Hu et. al., 2014; Jaakonmäki et. al., 2017). However, even though Instagram is more frequently utilized by young individuals, it is a platform with a focus on visuals (Instagram, 2017), rather than a communication tool where individuals exchange their thoughts and opinions (e.g. Facebook) (Rayport, 2011). This difference highlights that even though Facebook is more frequently used by all age groups, Instagram shows how young individuals exchange information today. Yet, even though Instagram is growing in popularity, not everyone can be reached through this SNS. The limited reach of this platform highlights the drawbacks of new technology and the new way of exchanging information. Even though people from all age groups can be found on Instagram, the majority of the whole world’s population is not active on mobile SNS such as Instagram (Statista, 2016), due to the differences in behaviours and needs. Understanding Instagram is important, due to that a lot of information can be portrayed in a photo or video post; which emphasizes the saying that a photo is worth a thousand words. Yet, the photo or video in itself does not contain all information. Any kind of content, that is visual, audio or text, contains information that is contributing to the overall

(15)

page 10 understanding of a post, but also identifies on different levels with different users (Chapman, 1997). Thus, if a post on Instagram grabs someone’s initial attention by being visually appealing (O’Brien, 2010), and provides additional content in order to satisfy his/her need, he/she will probably happily like, use hashtags or tagging, in order to facilitate the possibility to share content and information across various platforms (e.g. own Instagram page, friends’ message board, and explore page) (Hu et. al., 2014). This concept of sharing information, highlights the argument made by Yeh and Choi (2011), that eWOM on SNS spreads at the speed of light.

Yet, it is important to note that photos and videos on Instagram and other SNS are rarely posted without any text, audio or visual manipulation, due to that users want to show who they are, what their interests are, and display their knowledge, in one single moment (Kim et. al., 2013; Sirgy & Sun, 2000; Wu & Wang, 2011; Wojnicki & Godes, 2008). Moreover, Instagram users tend to follow their intrinsic interests, in regards to that they post and create eWOM around the same topics over and over again. This is evident, in Hu et. al.’s (2014) research, in regards to that they were able to divide Instagram users into five

categories depending on the main visual aspect in their posts.

2.2.1 Content Features

But what kind of features does a post have on Instagram? An Instagram post’s features can be divided into three main drivers, namely creator, context, and content (Jaakonmäki et. al., 2017). A post’s creator’s sex, age and number of followers can have a great impact on engagement and information exchange, according to Hu et. al. (2014) and Gilbert et. al. (2013). Furthermore, when wanting to publish a post it is also of crucial importance to understand when one can achieve the most interaction (Ellering, 2016; TrackMaven, 2014; Jaakonmäki et. al., 2017). According to Jaakonmäki et. al. (2017) a woman born before 1995, who has between 50’000 and 100’000 followers, and posts on a Friday morning (6-7am) or evening (8-9pm) is most likely to achieve higher levels of eWOM. The last feature, content feature, can be appealing and encourage interaction, by connecting to an intrinsic interest, motivation, attraction or curiosity, through the use of text, audio or visual. These three content features can be argued to provide information, that every individual needs to cognitively process, and thus emphasizes the multitude of personalities and needs, marketers face when creating content on Instagram.

(16)

page 11 A content’s text, which has been analysed by various researchers (Arguello et. al., 2006; Berger & Milkman, 2012; Gilbert et. al., 2013; Hu et. al., 2014), can display hashtags, emoji and/or descriptive text, in order to provide other users with background information about a post (Jaakonmäki et. al., 2017; Lee et. al., 2015). Jaakonmäki et. al. (2017)

investigated text in more detail and revealed that different emoji have different effects on Instagram. This means that more expressive emoji, such as hearts and monkeys, are more impactful compared to standard emoji, like smileys. Additionally, in the context of SNS, Lee et. al.’s (2015) results showed that the more hashtags that a post has, the more trustworthy and attractive the text is to other individuals. Therefore, it can be argued that emoji, hashtags, and text connect to hedonic and utilitarian values in individuals, in regards to that they find it pleasurable or useful (O’Brien, 2010).

In contrast to text, visual and audio features have not yet been widely studied on Instagram (Bakhshi et. al., 2014; Bakhshi et. al., 2015; Hu et. al., 2014). Yet, both features further convey the importance of achieving hedonic and utilitarian value in content features. Visual features cannot only be processed 60’000 times faster by an individual than text (Gangwer, 2015), but can also be an aesthetically appealing aspect that grabs users’ attention on Instagram, as well as it can be enjoyable and useful (O’Brien, 2010). Pictures influence how people view products, services, brands and companies (Fahmy et. al., 2014), and provide a gateway for users to experience positive or negative emotions (Bakhshi et. al., 2014). In order to enhance the experience with products or brands, Erkan (2015) suggests to make use of filtered pictures, due to that these are more appealing to people. This is in line with Jang et. al.’s (2015) findings that teens manipulate photos more often than adults, in order for the post to correctly self-portray the individual. Moreover, Jaakonmäki et. al.’s (2017) study revealed that if a visual contained animals, scenic landscapes or people, users on Instagram are easier drawn into the post and are more likely to tag, like or comment on it. In connection, Bakhshi et. al.’s (2014) findings showed that if a post includes a face, eWOM will increase. Moreover, their results indicated that the number of faces, the age and gender did not impact the level of eWOM that was achieved. However, even though these forms can be incorporated into a post, one has to note that not everyone might find the visual alone to be in line with an intrinsic interest, motivational, to spark curiosity, or to be attractive, which in the end will impact their intention to create eWOM (O’Brien, 2010; O’Brien & Toms, 2008).

(17)

page 12 The last content feature is audio. Audio content on Instagram does not only entail music but also what people say in their videos or live-streams (Jaakonmäki et. al., 2017). Here again the content of the audio can be sensually pleasing, euphoric or contain valuable information, which one can use or share (O’Brien, 2010; Laurent & Kapferer, 1985; Wu & Wang, 2011). Moreover, audio is an important aspect on Instagram, due to its direct impact on human emotions through music (Ebie, 2004). Whilst this might be true in general, one has to note that users have the opportunity to turn the sound off on their mobile devices. Hence, this means that video on Instagram can, depending on individual’s preferences, be viewed with or without sound. Furthermore, Stratten and Stratten (2016) highlight that audio is not always considered to be an important factor, due to the complication to measure its impact on SNS. With this the authors mean that a video on most SNS first becomes a “video” as soon as it is watched longer than 3 seconds, which often does not happen with the younger

generations where if the content does not grab their attention right away, they will continue swiping (Kiisel, 2012; Sweeney, 2006). Thus, previous studies lack of insights into audio can be a result of that this feature cannot be statistically measured if it is not viewed/listened to longer than 3 seconds.

Therefore, by considering that each content feature is comprised of several forms, which each can have a different impact on every individual, this study understands the significance each one has in terms of attracting users or encouraging them to create eWOM. Each content feature, that needs to be cognitively processed by each individual, can identify with the users and provide them with a pleasurable or useful experience. Even though Jaakonmäki et. al. (2017) describes that visual, audio and text are part of the same concept, one has to understand each one’s separate effect.

2.3 Analytical Model

In order to analyse the depicted research question, a theoretical analytical model has been formulated based on the presented literature (see figure 1). The three content features, which are presented by Jaakonmäki et. al. (2017) to be visual, audio and text, can further be analysed in terms of different forms. As previous researchers have shown each separate form can be appealing in different ways (i.e. engagement) and stimulate users to exchange

information (i.e. create eWOM) on Instagram. Therefore, the eight presented forms in the figure below can be argued to not only have an impact on engagement but also on eWOM. As the literature shows, engagement and eWOM are linked on SNS, and therefore play an

(18)

page 13 important role in this study and in answering the presented research question and purpose. Moreover, the various terms used in this study are defined in Table 1, in order to provide readers with a clear understanding of the aspects investigated.

(19)

page 14

3. METHODOLOGY

3.1 Research Design

How was the analytical model though used and measured in order to achieve a deeper understanding of Instagram users ́ appeal towards different forms of content, as well as in understanding why people tended to interact and exchange information differently depending on the content shown? This was achieved by understanding that individuals ́ attitudes (i.e. appeal and attraction to something) and behaviours (i.e. intent to interact or create eWOM) are complex and often difficult to measure with only one data collection technique

(Tashakkori & Teddlie, 2003; Saunders et. al., 2009). Thus, in order to effectively meet the research objective, an exploratory quantitative pre-study was combined with an in-depth qualitative experimental study.

The pre-study was conducted through an online survey, which aimed to explore and explain (Saunders et. al., 2009), on a general basis, how the three content features (visual, audio and text) were related to individuals´ appeal and intention to create eWOM. Therefore, the survey was not related to any specific consumption context or brand accounts, since the aim was to seek new insights and clarify the understanding of how the aforementioned concepts were related. Although in-depth interviews are common in exploratory studies (Adams & Schvaneveldt, 1991), an online survey was believed to be more appropriate, since it statistically made it possible to analyse how the variables were generally interconnected (Saunders et. al., 2009; Lefever et. al., 2007). The survey also made it possible to collect a vast amount of data within a limited time frame, which according to Yin (2003) is more difficult when using in-depth interviews (Marshall & Rossman, 1999). Consequently, the survey resulted in new insights about users ́ attitudes on a general level, however, deeper insights could not be gathered in regards to why people tended to behave or react the way they did.

To understand the actual reasons behind the participants´ stated behaviour, the experimental-in depth study aimed towards being more of a qualitative character (Saunders et. al., 2009), in which the respondents were exposed to a variety of different visual, audio and text forms online. The posts shown were related to two specific brand accounts on Instagram (H&M or Nike) and each participant had to clearly justify their reactions to each post. As such, this part of the study can be perceived as an in-depth study with experimental

(20)

page 15 aspects (Hakim, 2000), since changes in the outcomes, engagement and eWOM intentions, were analysed in relation to changes in particular content forms (e.g. static vs. coloured background and long vs. short text). However, studies with experimental aspects should not be understood as being synonymous to classical experiments (Hakim, 2000), since the participants in this study were not randomly assigned to experimental and control groups (Saunders et. al., 2009). Nevertheless, the experimental aspects included resulted in new and deeper insights into how and why people touch, swipe and click on Instagram. A more detailed description of the pre-study and experimental in-depth study is presented in the sections below, as well as the full-length studies can be viewed in Appendix 1(pre-study) – Appendix 3 (in-depth study).

3.2 Pre-Study

In order receive a better understanding about the interconnectedness between content forms, engagement and eWOM in the context of Instagram, an online survey was conducted with 111 respondents. The majority of the questions included were adopted from previous researchers’ findings (see Appendix 1), which allowed the study to assess the findings reliability (Bourque & Clark, 1994) on a relatively new SNS (Instagram). Each theoretical concept was measured by six to eight questions, which can be inspected in Appendix 1 table 1. These questions, in combination with five background questions, yielded in a total amount of thirty-eight questions. This can, according to some findings (Edwards et. al., 2002), be perceived as too many questions for a self-administered questionnaire, since longer surveys are believed to have lower response rates. However, the research findings are mixed (deVaus, 2002), and some researchers (Saunders et. al., 2009) state that a survey with thirty-eight questions indeed is an acceptable length for a pre-study. Moreover, the exploratory purpose of this phase also required a great number of questions, due to its aim of finding new insights on the studied SNS. Consequently, the thirty-eight questions were needed in order to meet the research objective.

The majority of the questions were designed as close-ended questions (category and rating), whilst the introductory questions were designed as open-ended. Closed questions, according to Dillman (2007), enable the possibility to collect and compare data about attitudes and opinions in a time-efficient manner. Thus, closed questions were used since it allowed the study, on a general basis, to compare how changes in content impacted

(21)

page 16

3.2.1 Scale Measurement

The responses received were measured on a seven-point Likert scale, in which the individuals were asked how strongly he or she agreed/disagreed with a certain statement. Although a variety of rating scales exist (e.g. graphic rating scale and semantic scale), the Likert Scale is the most widely used one within social sciences (Carifio et. al., 2007), due to its effective ability in measuring attitudes. In accordance, this study also utilized the Likert Scale to measure the components of content features, engagement and eWOM. Despite its effectiveness, it is crucial to remember that the points on the scale might not represent equal changes in attitudes for all respondents. A movement from e.g. disagree to slightly disagree on the question “The use of emoji appeals to me more than words” (Appendix1), might not have the same meaning for all respondents. However, even though questions and answers could be perceived differently and thus negatively impact the validity, the seven-point Likert scale was chosen, since previous research (DeCoster, 2000) shows that seven-scored scales are more reliable than scales with greater or fewer options.

Yet, too many answering options might, according to Jamieson (2004), cause a central/outlier bias. This outlier bias can connect to that respondents who do not fully

understand the question or do not want to reveal their true opinions, tend to respond by giving a midline response or by concentrating on one response side. This tendency was assessed as being a potential problem for the pre-study since the questionnaire was constructed in English, whilst the majority of the respondents were Swedish. As a result, the respondents might have found it difficult to understand the true meaning of some questions and thus responded by relying on one response side (strongly disagree/strongly agree). However, to reduce this risk, the wordings of the questions were carefully considered and words with double meanings, abbreviations and double negatives were avoided. For instance, instead of using the term ‘engage’, which can be easily misunderstood, the questions included easier terms like ‘attract’ or ‘appeal’. In addition to paying attention to the wordings, the order of the questions was also randomized and included both positive and negative statements. By combining questions with positive and negative statements, respondents were required to think more thoroughly before responding, which was also believed to be an effective method for reducing the response bias (Sauro, 2011).

(22)

page 17

3.2.2 Administration of the pre-study

The layout of the questionnaire was created in Google Forms and administered with the use of an online link that was posted on student community pages (Master Students Uppsala University and Kurslitteratur, Bostad och Allt Annat med Ekonomer), as well as distributed to the authors ́ friends to further allocate throughout their Facebook networks. This form and place of distribution was assessed as the most effective channel for reaching a high response rate, in contrast to other social networking sites (e.g. Twitter and Instagram), since the desired young population (18-29), according to Bolton et. al., (2013), is most likely to be active and read postings on Facebook.

When a respondent accessed the questionnaire through the link, they were welcomed with an introduction letter, designed in warm-pastel colours of green, brown and orange (see Appendix1). Warm colours such as these, in contrast to bright or fluorescent colours such as turquoise, was chosen since previous research (Edwards et. al., 2002) shows that warm colours generally generate more responses. The introduction letter informed about the purpose of the study, time required (five minutes), anonymity and lastly fair treatment of the responses received. Assuring anonymity is an important aspect of receiving high response rates (Healey & Rawlinson, 1994). When people did not have to reveal their identities, they were likely to indeed show their “true attitudes” in relation to different content features, engagement and eWOM. It is crucial to assure anonymity since previous research shows that individuals are more likely to give inaccurate responses when their anonymity is not assured (Kozinets, 2002).

The questionnaire was reposted three times, in order to remind potential participants to take part of the study. The link was in function for two weeks in the end of March and beginning of April 2017. In total, the questionnaire yielded in 129 responses, of which 111 were valid. For being valid, the respondent had to fulfil the criteria of age (18-29) and usage (Instagram user). Consequently, the study utilized the sampling technique of what Black (2010) refers to as purposive sampling, in which all respondents who fulfilled the previously stated criteria were included for further analysis. Although non-probability sampling

techniques, like purposive sampling, makes it difficult to generalize findings on statistical grounds (Pallant, 2010), it was believed to be the most time efficient method for achieving the purpose of studying young Instagram users’ attitudes and propensity to create eWOM. The limitations of the chosen sampling technique could, however, be partly managed when

(23)

page 18 the findings from the pre-study were combined with the deeper insights received from the experimental in-depth study. A more concrete view of the respondents included in the pre-study is exhibited in Table 2 below.

As can be seen in Table 2, there were mostly women (N=81) who participated in the pre-study. Thus, the results received might have been biased towards how women, rather than users in general, reacted and interacted to different content forms. That more women

participated in the study was not, however, assessed as being a potential problem, due to the fact that Instagram´s biggest gender demographic is female and not male (Seligson, 2016). Consequently, the sample was believed to be an accurate representation of Swedish

Instagram users.

3.2.3 Analysis of theoretical constructs and collected data

The data collected from the pre-study was in a final stage analysed with the SPSS Software tool. SPSS is a widely-used program for statistical analysis within market research (Pallant, 2010; KD Annual Software, 2013) and allowed the study to analyse how content features affected individuals’ engagement and eWOM. However, prior to conducting the Multiple-Regression Analysis, preliminary analyses were performed to ensure that the assumptions underlying Multiple Regression were satisfied. Moreover, the Content Validity was assessed in an Exploratory Factor Analysis (EFA), whilst the reliability was evaluated by one of the most commonly used reliability-statistics, Cronbach´s Alpha (Pallant, 2010). The results from the pre-study´s initial data analysis are further described in section 1.2 in

(24)

page 19 Appendix 1. However, the most crucial findings in regards to the development of the

experimental in-depth study are presented in Table 3 below. According to the pre-study, Audio had the strongest impact on individuals´ propensity to create eWOM (Like, Tag or Comment) (0.356), whilst Visual had the strongest impact on Engagement (0.231).

3.3 Development of experimental in-depth study

The findings from the exploratory part of the study was used to further develop the experimental aspects in relation to two different brands, in which the overall aim was to investigate the proposed relationships presented in the analytical model exhibited in Figure 2 below. The experimental study can be accessed in Appendix 2 (Full-Length) and Appendix 3 (Summarized Version).

(25)

page 20

3.3.1 Selection of brands

The experimental in-depth study was in an initial phase developed based on the selection of two of the most followed brands on Instagram, namely H&M and Nike (Statista, 2017). The number of followers served as an important selection criterion, since the number of followers generally indicates that an account is displaying content that is engaging for the population at large (Brenner, 2017). If more narrowly focused accounts had been included (e.g. directed towards makeup or dog-lovers), there would be limited possibilities to

generalize the findings to broader user groups, which this study aimed to do. Thus, the brands H&M and Nike were chosen mainly due to its general appealing characteristics (Hu et. al., 2014). However, these chosen brands were not only chosen based on this criterion.

Additionally, H&M, a well-known retailer, and Nike, a sports brand, were selected since previous researchrelated to experimental studies highlight the importance of including study objects (e.g. brands) that respondents have prior knowledge of and/or connections to (Hakim, 2000). Prior knowledge is generally perceived as advantageous, since it is often easier for respondents to describe their mental reactions and intentions to interact in an accurate manner (which in itself is a complex task in experimental studies) (Saunders et. al., 2009; Hakim, 2000). However, it is important to note that there are potential disadvantages related to the

(26)

page 21 inclusion of well-known brands, since respondents´ prior connections might bias their

responses. If a participant, for instance, would have had a previous negative experience with the Nike brand (e.g. due to poor quality of the latest product bought), he/she might have shown negative reactions and been less inclined to create eWOM, due to his/her personal opinions. Consequently, the inclusion of well-known brands might have impacted the validity negatively. However, to overcome this potential risk, the participants were specifically asked to focus on, for example, the visual or text phrase alone, without taking other factors into consideration (see Appendix 2).

Although it can be argued that the external validity of the findings could have been improved by including additional brands, with different degrees of consumer awareness, in different industries and product categories, both time constraints and respondents´

willingness to participate made it difficult to include additional brands. Thus, to meet the research objective, two brands in two different consumption contexts (fashion and sports) were assessed as being satisfactory.

3.3.2 Selection of Instagram Posts

The choice of brands subsequently aided in the selection of the different Instagram posts (Appendix 3), where previous content forms´ research, availability and a maximum variation of content, served as selection criteria. In regards to previous content forms research (Erkan, 2015; Jaakonmäki et. al., 2017; Bakshi et. al., 2014; Ebie, 2004; Lee et. al., 2015), Table 4 presents a closer view of the selection and evaluation criteria for each post. As can be seen for visual forms, filtered pictures (Erkan, 2015), modified backgrounds, landscapes and people (Jaakonmäki et. al., 2017; Bakshi et. al., 2014) served as a basis for the choice of the different visual posts, whilst verbal and musical (Ebie, 2004) audio forms were used as selection criteria for the audio posts. Finally, expressive emoji (Jaakonmäki et. al., 2017), number of hashtags (Lee et. al., 2015) and text positions (Jaakonmäki et. al., 2017) were used as selection criteria for the text posts. More precisely, for example audio posts 1, 4 and 5 were chosen due to that the main audio form in these videos were verbal, whilst different types of musical forms occurred in audio posts 2, 3 and 6.

In addition to the research selection criteria, the availability of the posts also assisted in the selection. More precisely, the two chosen brands in the initial stage limited the

(27)

page 22 H&M or Nike to fit the sample. Consequently, there was a potential risk that the content displayed in the different posts included too similar elements, since brands often try to communicate a consistent message throughout all posts (Baines & Fill, 2014). This was assessed as being a potential problem for this study, since the aim of the experimental phase was to analyse how different forms of content impacted engagement and eWOM. As such, a maximum variation technique was used to ensure a wide variety of different content forms. More precisely, instead of choosing, for example, two audio posts with similar content forms, the study aimed to display various adaptation of a content form. This means that both hyped music as well as calm music, and informative verbal audio and storytelling verbal audio were chosen.

3.3.3 Development of questions

The selection of brands and Instagram posts aided in the creation of the questions, where the respondents were required to firstly answer questions in relation to their

background and secondly share their feelings, perceptions and information about how and why they reacted to a certain post. Each question intended to measure the theoretical concepts of content forms, engagement and eWOM (see concept definitions in Table 1). The

(28)

(29)

page 24 The majority of the questions were designed as open questions since open questions, compared to closed ones, are more appropriate when the aim is to explore and get a deeper understanding behind individuals´ reasoning (Dillman, 2007). As can be seen in Table 5 and Appendix 2, all questions were framed with simple wordings and structured in a similar way in order to simplify the answering procedure for the respondents, as well as to reduce the risk of misinterpretations. Each section (Visual, Audio and Text) was ended with four statements (I will click Like on this post”, “I will write a comment on this post”, “I will send this as a private message to someone”, “I will tag someone in this post”) that intended to measure individuals´ behavioural intentions to create eWOM. These statements were used since previous studies (Hennig-Thurau et. al., 2004); Kietzmann & Canhoto, 2013) have followed a similar approach. Although these four statements are commonly used when measuring

behavioural intentions in relation to eWOM,one can argue that only limited insights could be gathered in regards to how individuals indeed would react and behave in real life. The true behaviour could possibly be captured by a classical experiment (Hakim, 2000), in which the behaviour in relation to participants´ reactions (engagement) and interactions (eWOM) could be physically observed. However, classical experiments might not either be the optimal method to measure behaviour, since it is widely known that individuals tend to modify their behaviour when they know they are being observed (McCarney et. al., 2007). This is

generally not considered as being a problem for online-administered experimental studies (Lefever et. al., 2007; Dillman, 2007), since participants have the time and quiet to answer more truthfully without the feeling that someone is observing them (Lefever et. al., 2007). This aspect, in combination with limited time and geographical constraints, made the online-administered experiment more suitable for this study compared to a classical experiment.

3.3.4 Administration of the experimental in-depth study

The in-depth experimental study was, after the selection of brands, Instagram posts and questions, created in Google Forms and administered through an online link that was distributed to fourteen carefully selected individuals. The first part was similar to the introductory section of the pre-study (Appendix 2), in regards to that information about the purpose, time, anonymity and fair treatment of the responses was given. In contrast to the questionnaire, the in-depth study required more time and effort (1.5 hours), since each

respondent was required to look at various photos, videos, and texts, and successively explain their reactions thoroughly. The length of the questionnaire and thus the associated time

(30)

page 25 requirements were believed to have a negative impact on respondents ́ willingness to

participate. As such, a purposive sampling technique (Black, 2010) was used to ensure that only individuals who indeed would participate and had the willingness and effort to do so, were chosen. However, the individuals included were not only selected based on that

criterion. As Patton (2002) emphasizes, purposive sampling techniques are especially suitable when one wants to select individuals that are particularly informative in relation to the

research question. Thus, a set of sample criteria was set up to ensure that insights could be gathered in relation to how different users intended to react and exchange information (i.e. interact). More precisely, age (fair distribution between 18-29 year olds), gender (fair distribution between men and women), occupation (different types of occupations), interests (different types of interests e.g. sports and fashion) and Instagram usage (different degrees of usage), were used as selection criteria. These specific categories were chosen since previous research generally has followed a similar categorization (Tadajewski et. al., 2008).

Although this sampling technique resulted in a wide range of attributes, behaviours and experiences, it is important to have in mind that those who indeed participated were selected based on the judgment made by the authors. Hence, the personal connection between the respondents and the authors could, according to Robson (2002), have biased the way the respondents answered. For instance, some individuals might have chosen not to reveal and discuss certain aspects of their opinions or behaviour (e.g. why they preferred the picture with a male and not a female), because they wanted to put themselves in a “socially desirable role”. To reduce this risk, all individuals were assured anonymity and it was clearly stated that there were no right or wrong answers to any questions (see introduction part Appendix 2). By contrast, Ghauri andGrønhaug (2005) state that personal connections can have a positive impact on individuals´ propensity to answer in a truthful way. When a respondent knows the author, he or she often has a higher confidence in revealing detailed and sensitive information. Thus, the personal connections between the authors and the respondents were assessed as being particularly advantageous in this case, due to the respondents´ potential willingness to share detailed information about their reactions and propensities to create eWOM. A further description of the individuals included is exhibited in Table 6 below.

(31)

page 26 The selected individuals represent a relatively fair spread in regards to the variables studied. Yet, one has to note that the majority of the participants were Swedish, which might have impacted the generalizability of the findings negatively. Thus, the insights and results received cannot necessarily be applied to users´ behaviour across different nationalities, but rather to Swedish users´ behaviour. This limitation is further discussed in section 7.2.

A sample size of fourteen respondents is, according to Guest et. al., (2006), a

sufficient number of cases for qualitative research, although Patton (2002) argues the actual sample size to be of minor importance for purposive sampling techniques. What matters is not the sample size per se, but rather how information rich each participant is in relation to the subject studied. The fourteen individuals included were all chosen based on the selection criteria and thus believed to possess the knowledge and experiences needed for the study to meet its research objective. Consequently, the fourteen individuals included in the participant pool were assessed as being a satisfactory sample size for the experimental in-depth study.

(32)

page 27

3.3.5 Analysis of data

The responses received from the experimental in-depth study was in a final stage evaluated in relation to previous research findings, where the proposed relationships in Figure 2 were deeper evaluated. More specifically, the impact of each content form in the different posts was analysed in relation to individuals´ 1) engagement and 2) intention to interact and create eWOM. Thus, the selection criteria for the posts presented in Table 4 also formed the evaluation criteria for the analysis, where the applicability of previous content form,

engagement and eWOM research was assessed on Instagram. Through this analysis, the specific aspects which stimulate information exchange on Instagram were identified.

(33)

page 28

4. RESULTS

4.1 General Insights

Yet, what were the specific results in relation to how appealing content was to individuals and their propensity to interact with a post, based on the content forms they were presented with? Before looking at the specific results, certain general findings have to be presented, which relate to how information exchange has transformed and is different on Instagram than on other SNS. The exploratory in-depth study revealed some general results in regards to individuals’ Instagram habits, which include the findings that individuals primarily use Instagram to follow friends and family, rather than official brands or accounts (see Table 7). This following-tendency is stated by participants to relate to their preference to get updates on friends and loved ones’ lives and activities. Moreover, looking closer at visual, audio and text content features, general findings have been identified in regards to each feature.

Firstly, even though a person’s interests play a key role in the propensity to act towards a post on Instagram, the results showed that some of the pictures were likely to appear on participants’ newsfeed, such as Visual 1 and 3. The other pictures, according to the results, would not be seen by the majority of the participants, simply because participants neither followed such accounts nor were they interested in the aspects they were displaying. Furthermore, it is uncommon to start following an account only based on the visual seen, since almost all of the respondents claimed to mainly follow friends or companies they find interesting. This is further supported by the fact that almost no person used the explore page or watched Instagram stories of official accounts.

Secondly, the in-depth analysis of audio on Instagram portrayed the varied reactions towards the content. Not every audio included on Instagram is of value or importance to users, which can be seen in Table 8 and 9, where the various components of audio (i.e. music, verbal sound and message) have been commented on by the participants. No single audio investigated revealed one and the same reaction to it.

Thirdly, the in-depth analysis revealed that users, generally, read the text below a visual, although some stated that they only do so if the picture in itself ”…is not clear

(34)

page 29

4.2 Engagement - Content Appeal

When looking closer at how appealing brand content forms are on Instagram, several interesting findings can be highlighted, in regards to visual, audio and text. Yet, one has to firstly note that in order for a person to start following an account on Instagram, the content needs to be continuously updated and in line with individuals´ stated interests. The majority stated as well that they seek entertaining content that simply makes them smile or laugh (Respondent 12, 2017).

(35)

(36)

page 31 Furthermore, consistent messages or themes throughout an account were appreciated by many. Yet, it was highlighted to depend on if it would be on an official account or a friend´s account. Inconsistent messages were said to be more appreciated on private accounts, since it gives “…a true picture of the person owning the account” (Respondent 4, 2017). Official accounts, by contrast, need consistent messages to a higher degree, but run the risk of being too monotone and could be seen as too commercial.

4.2.1 Visual

When looking closer into an account´s pictures and videos, the majority mentioned visually appealing pictures with bright colours, landscapes and vacation spots as being particularly captivating. Consequently, edited pictures that use filters and colouring functions were appreciated, since

“…a photo is interesting when it displays a nice setting, mainly related to a nice landscape or in a cool spot. “Normal" pictures as you see on Facebook are not that appealing” (Respondent 6, 2017).

Moreover, the results showed that almost all pictures presented were assumed to be to some degree manipulated (i.e. used filters or photoshoped), which according to the results made, for example, Visual 2 “brighter”, more “appropriate” and more “appealing”

(Participant 3, 2017). For Visual 1 participants were highlighting that they liked the black and white filter, however that if the picture was shown in a coloured filter, the happiness of it would have come across even more. Nevertheless, for Visual 5 no one thought a filter was used and thus it was mentioned that something was missing or not highlighted enough to enhance the picture’s meaning. Hence, missing or single coloured backgrounds, with no contrasts, were less attractive and said to be boring and without a context or theme. On the other hand, the amount of people/faces that should be shown in a post, depended on the picture itself and the message/theme that was related to it. For example, due to that Visual 1 was mostly associated with happiness, participants explained that if the picture would display two women instead of men, it would not have made a significant difference.

The investigation of visuals in relation to videos on Instagram, have revealed

interesting information in regards to visual as well as audio forms. Table 8 shows the general video outcomes when audio was not considered. In regards to the likelihood to watch the six

(37)

page 32 videos on their own Instagram, most answers revealed that participants would not, due to it being commercial, uninteresting, too long or not fun. However, videos, by contrast to

pictures, do not necessarily need to be edited, since the informative message and storytelling aspects is what mainly captures the attention.

4.2.2 Audio

However, even though a person’s interests again played an important role in the appeal of the videos, participants seemed to agree on that most videos needed a sound in order to understand the context. Yet, they also agreed that if a video is short and the visuals are well done, no sound is needed and the likelihood of watching the video on their own channel is higher than for the other videos (e.g. Video 3) (see Appendix3). In regards to Video 6, the majority of the participants (those not interested in fashion) thought the video without sound to be annoying and uninteresting and did not find it to give any value, which impacted the overall negative viewpoint on the post. Yet, the results about audio content revealed that verbal sound was preferred over music, which is enforced by participants stating that the music in Video 6 was “not important [and it] did not increase my interest”

(Participant 10, 2017), whilst in Video 1, where a verbal aspect was used, participants were stating that it is “very important! […]. The video without the visuals was not appealing to

(38)

page 33 These arguments were further highlighted throughout the other videos, based on that the verbal videos (Video 1, 4, and 5) (videos where speaking was the main form) were described with words such as story, message and value, which is in contrast to the more mixed and negative words used for videos containing music (music is the main audio form in these videos) (Video 2, 3, and 6), that were namely unnecessary, bland and annoying (see Table 8). The spoken aspect was also preferred, due to that it provided a feeling to the video, whilst with music most argued that it depended on the music chosen (see Table 9).

Participants pinpointed that some did not add any value, and did not match the video well. However, they also highlighted that if music and speaking were combined an even more powerful connection could be built, if they both connected to one message and emotion. Overall, the two most preferred videos were the video about Serena Williams and the H&M’s environmental ad, due to that they conveyed important messages:

“The H&M environmental video, it had a clear message that added to the video” (Participant 3, 2017).

“The Serena Williams one, I loved the strength in it” (Participant 12, 2017).

4.2.3 Text

In terms of text, some partakers explicitly mentioned that “…the text below the video

needs to be short and inform about what the video is all about…” (Respondent 2, 2017). That

is, it should preferably only consist of 4-5 clear words and not become overly generic or commercial. This argument is further emphasized in Text 1 and 4, in which some respondents believed that the commercial purpose takes away the positive emotions created from the picture alone.

Nevertheless, the results indicated that when an unclear picture is accompanied by a short text (see Table 9), which shows a connection to the picture, it can have a positive impact on users on Instagram (Participant 3, 2017). Moreover, a happy tone, in combination with the use of emoji is described as “…evoking positive feelings” (Respondent 7, 2017) and is further stated to attract users on Instagram (see Table 8 and 9). By contrast, few appreciate the use of hashtags due to it giving an ”…unserious feeling” (Respondent 4, 2017), where it is apparent that the owner of the post is only looking for attention (see Table 9). This is, however, dependent upon how many hashtags that are used. If a post includes less than three

Touch, Swipe or Click?: Understanding information exchange (eWOM) on Instagram and how it can be encouraged