The effectiveness of transparency in generating online customer engagement

(1)

!

Abstract 

Consumers nowadays are more demanding of companies to be transparent, however little knowledge exists of the effects of transparency in a business to consumer environment. Therefore, this study examines the influence of transparency on customer engagement, by taking brand post popularity as indicator. To measure this, 349 brand posts of the social media output of Triodos Bank NL have been analyzed on brand post popularity. Partially in succession of previous studies, type of post, type of transparency, timing of posts and social media channel have also been included in the research model to test for any effects on engagement. The key finding of the research is a positive relation between transparency and engagement. No influence of the timing of a post could have been found, but the type of post (photo, video or status update) led to significantly different levels of engagement. Also, the channel on which posts are shared made a difference in the brand post popularity rates. Last, reach did explain some of the popularity of brand posts as well.

Tags 

Transparency, Customer Engagement, Social Media, Brand Post Popularity

!

(2)

Index

!

Introduction

3 Theoretical Framework

6

Engagement, Word of Mouth & Brand Post Popularity 6

Transparency 8 Dimensions of Transparency 9 Effects of Transparency 10 Types of Transparency 11 Control Variables 13

Study Design

18

Operationalization of Variables 18 Data 21 Methodology 22

Results

23 Conclusions and Discussion

26 Managerial Implications

28

Enhancing the level of engagement 28

Limitations and Further Research

29 References

30

(3)

Introduction

Social media has become an important part of marketing. Not only is it a great tool to increase salience and gain new customers through advertising, it also is a great way to connect and interact with (potential) customers. Its eﬀects speak for itself with more than 50% of the active social media users follow brands online. These followers deliver a higher dispersion of word of mouth (WOM) among customers as well as more sales as customers are more tended to buy from brands they follow online (van Belleghem, Eenhuizen, and Vernis 2011).

Although plenty is written about social media strategies, there is still a lot to discover about the influence of social media on customer engagement. Customer engagement is Preliminary studies have been conducted about the eﬀectiveness of social media as a marketing tool, but little theoretical research exists about what influences brand post popularity. Research related to social media strategies varies from how to create brand communities to what is the best timing and type of content in order to generate interactivity and a positive sentiment (Hennig‐Thurau, Gwinner, Walsh & Gremler, 2004; Hung & Li, 2007; Muniz & O’Guinn, 2009; Mollen & Wilson, 2010; Szabo & Huberman, 2010; Chu & Kim, 2011; Brodie, Ilic, Juric & Hollebeek, 2013).

Social media has many positives to oﬀer to companies, which have made it necessary to participate in social media activities to remain competitive. More than 80% of the small businesses is on social media of which 94% uses social network sites for marketing purposes (Atkinson, 2014). With an increased availability of information and the web 2.0 as a platform for sharing information, companies lost control of the information shared about the company and its products and services (Bronner & Hoog, 2010; Kaplan & Haenlein, 2010; Szabo & Huberman, 2010). WOM has a bigger reach than ever because consumers can reach a worldwide public within one click

(Henning-Thurau et al., 2004; Bronner & Hoog, 2010; Rejón-Guardia & Martínez-López, 2014). Additionally, for those companies familiar to highly controlled marketing communication, it might be a struggle to deal with possibly unfavorable, transparent information. To prevent for unpleasant surprises, companies might want to act on an old saying: honesty is the best policy. Being open and honest about all business practices can help

prevent a company for unfavorable information to be shared by third parties (Arpan & Roskos-Ewoldsen, 2005). Furthermore, consumers are more demanding for companies to deliberately reveal this kind of (unfavorable) information itself, which can be called being ‘transparent’ (Cohn & Wolfe, 2013). There are several examples of companies that tried to hide information that eventually hit them in the back. For example, in March 2010 Greenpeace attacked Nestlé’s Kit Kat brand about its unsustainable forest clearing in production of palm oil on social media through a video (“How Nestlé dealt with a social

“Honesty

is the

best

policy”.

(4)

media campaign against it”, 2012). Nestlé responded with the removal of the negative content on their social media pages and a court order for Greenpeace and YouTube to remove the video. Nestlé’s response led to great antipathy and a viral outbreak of negative comments both on twitter and Facebook. After YouTube removed the video, Greenpeace posted it on another social media site, Vimeo, where it had 78,500 views within a few hours only. The video also reappeared on Youtube thanks to customers who were angry about Nestlé’s attempt to hide this negative information from them. Word spread across social network sites and Facebook pages called “Your Nestlé comments won’t be deleted here” were made. This is an example of how not being transparent, gets social media masses mobilized (“Nestlé kerfuﬄe presentation”, 2011; “De 47 slechtste social media cases #fail”, 2010). Therefore we can conclude that being honest and transparent about the current palm oil usage could have prevented Nestlé from this nightmare.

Another example of a company that deals with the power of social media is Heineken. However, Heineken communicated open and honest in an attempt to keep its customers satisfied and involved. This resulted in totally diﬀerent results as seen in the Nestlé case. In 2012 Heineken was associated with dogfights an an old photo circulated on the internet of a dogfight game with banners of Heineken on the walls (The Heineken company, 2012). It appeared as if Heineken had sponsored dogfighting. When Heineken was confronted with the issue, they started actively to communicate to all their customers that this photo was available on the internet and told their story upfront: they have never sponsored this event, the banners were from a party the night before. The first post of a specially made banner with all information about this situation, got shared almost 400.000 times (Heineken Facebook page, retrieved on June 22nd, 2014). Heineken’s community embraced the honesty of Heineken in revealing this semi negative information. These examples show that transparency can be an important (online) marketing tool to enhance (online) engagement. Therefore transparency seems like a promising research area.

This research will examine the influence of transparent communication on customer engagement with the company. Several characteristics will be taken into account, such as the type of transparency and type of content post. It will provide insights into the eﬀect(s) of using a transparent communication strategy and knowledge of how to improve (online) customer engagement.

The aim of the research is to investigate if and how transparency improves customer engagement. Moreover, this study will add to the current knowledge insights in the effects of transparency on brand post popularity and thereby will show if transparency improves consumer engagement. We assume there is a possibility that the different types of transparency moderate the relation between transparency and engagement. Hultman & Axelsson (2007) defined four different types of transparency, which they used in a business

(5)

to business (b2b) environment. The application of this b2b theory in a business to consumer (b2c) environment will also add to the current knowledge. Additionally for the risk and reputation management field this research could provide new insights, as well on how to manage possible negative information. Furthermore it contributes to the research about social media, where it gives a deeper and more over theoretical insight on what kind of information transparency contributes to higher engagement. In order to do so, this study examines engagement, by the popularity of brand posts and measuring this popularity by the number of likes, comments and shares (retweets). In this study the brand posts of Triodos bank NL will be used.

!

(6)

Theoretical Framework

In this section, the current knowledge of the concepts used in this paper will be widely examined. Thereby a definition of transparency in business to consumer marketing and hypotheses will be formulated.

!

Engagement, Word of Mouth & Brand Post Popularity

In social media strategy, engagement is often perceived as a valuable goal (Mollen & Wilson, 2010; Hollebeek, 2011; Hollebeek, Glynn & Brodie, 2014). Although diﬀerent meanings are given to engagement and it is used in a variety of contexts. However, in the marketing field, consumer engagement behavior (henceforth engagement) has been studied in relation to brands and brand communities, in order to describe the nature of consumers’ interactive brand relationships and degree of involvement (Hollebeek, Glynn & Brodie, 2014). Additionally, engagement is often mentioned together with constructs as interactivity, Word of Mouth (WOM) and involvement. Therefore it is important to clarify these terms. First interactivity is defined by Hollebeek (2011, p. 560) as “the level of a customer's motivational,

brand-related and context dependent state of mind characterized by specific levels of cognitive, emotional and behavioral activity in brand

interactions”. Second, Mollen and Wilson (2010, p. 12)

found involvement to be “an important dimension of

engagement” but they stipulate the diﬀerences between the terms as: “consumer involvement requires a consumption

object which is usually defined as a product category”

whereas “engagement goes beyond involvement in

encompassing an active relationship with the brand as personified by the website”. Van Doorn, Lemon, Mittal,

Nass, Pick, Pirner and Verhoef (2010, p2) move away from the idea of involvement and refer to engagement as “the

customers’ behavioral manifestation toward a brand or firm,

beyond purchase, resulting from motivational drivers”.

Additionally, according to Van Doorn et al, customer engagement behavior includes a vast array of behaviors including WOM activity, recommendations, helping other customers, blogging, writing reviews, and even engaging in

legal action. All the definitions have in common that it is about the behavior of consumers, based on motivation, to act in a certain way. In this research, engagement will be examined in an online context and therefore the definition of Mollen and Wilson (2010) used is most applicable: “Online engagement is a cognitive and aﬀective commitment to an active

“Engagement is

the customer

behavioral

commitment,

beyond purchase,

to an active

relationship with

the brand as

personified by

online (inter-)

activity and

involvement.

”

(7)

relationship with the brand as personified by the website or other computer-mediated entities designed to communicate brand value.” The definition interestingly lacks a mention of behavior, which all other definitions included. In this research, therefore, we will add behavior to the definition of Mollen and Wilson and refer to engagement as “The customer’s

behavioral, cognitive and aﬀective commitment, beyond purchase, to an active relationship with the brand as personified by online (inter-) activity and involvement”.

The majority of studies about engagement have been primarily exploratory in nature, and so there is a lack of empirical research about what influences engagement. Constructs such as interactivity and involvement are often mentioned and used when exploring engagement. Hollebeek et al. (2014) indeed found proof for involvement to be an antecedent for engagement, together with self-brand connectedness and brand usage intent.

DeVries, Gensler and Leeflang (2010) examined what influences engagement online by analyzing 335 brand posts of 11 international brands. They tested brand posts on vividness, interactivity, information, position, nature of comments and entertainment of brand posts on Facebook. The vividness, position, share of positive comments and interactivity of brand posts have been found to be influencers of engagement. However, entertainment and information did not seem to contribute to more popular brand posts. This research contributed highly to the literature of engagement but also more specifically to knowledge of how customer engagement can be achieved on social media. Social media is “a group of

Internet-based applications that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of User Generated Content” (Kaplan & Haenlein, 2010, p.61). Web 2.0 is described by the latter as the platform for the evolution of social media, whereby content and applications are created by all users in collaboration. WOM is recognized as an important influencer of attitudes (Lee, Rodger & Kim, 2009) and consumer purchase behavior, mainly because information comes from a source that is perceived as more trustworthy: fellow customers (Hung & Li, 2007; Luo, 2008; Chu & Kim, 2011; Rejón-Guardia & Martínez-López, 2014). Rejón-Guardia and Martínez-López (2014, p. 823) define WOM as “a strong behavior of information exchange between consumers, which

has influence on diverse issues such as: attitudes and consumption behaviors and brand awareness”. WOM has existed for decades but as a result of new technologies, WOM was

able to move to an online setting (Chu & Kim, 2011; Rejón-Guardia and Martínez-López, 2014). Online WOM, referred to as electronic WOM (eWOM), can occur on almost any online channel such as blogs, review websites, forums, virtual communities and social network sites (SNSs).

Due to the possible positive eﬀects of (e)WOM to a company’s public image, companies are trying to generate more (e)WOM through the use of social media. Attendance of companies and brands on SNSs enables consumers to engage by commenting, liking or sharing content with their connections (Chu & Kim, 2011). As more marketers start using

(8)

social media, greater understanding is needed of the determinants that influence consumers’ engagement in eWOM.

!

Transparency

As consumers declare to value transparency the most after price and quality in purchase decisions, it seems interesting if transparency can be such influencer for higher levels of engagement (Cohn & Wolfe, 2013). This paper examines how engagement is influenced by the level of information transparency (henceforth transparency). Transparency is widely used within the business industry, where corporate transparency can be defined as “a situation in

which business is done in an open way without secrets, so that people can trust that things are done in a fair and honest way” (Cambridge Business

Dictionary, 2014). Especially for consumers who are concerned about the environment and society, information transparency can be important for purchase and consumption choices (Bhaduri & Ha-Brookshire, 2011). The diversity of transparency matters comes to light in the definition of Bushman, Piotroski and Smith (2004): “The availability of relevant, reliable

information about the periodic performance, financial position, investment opportunities, governance, value, and risk of publicly traded firms”. This definition shows the demands that

concern transparency: information should be relevant and reliable. Therefore, when transparency is applied in the marketing field specifically, it is about disclosing marketing messages that are open and honest and most of all relevant to the consumer as well (Christensen 2002; Bushman, Piotroski and Smith, 2003; Bhaduri & Ha-Brookshire, 2011). Again relevance of information seems to be an important demand of transparency. Being honest and open also implies that sometimes information is disclosed that is not favorable for a company or brand. Hultman & Axelsson (2007) allude for this kind of transparency too as shown by the following quote: “it

is about sharing information that is not usually shared. It is information that should be shared more often though”.

Considering these factors, transparency seems to refer to the level of information visibility and availability, which intentionally have been disclosed and can be used for the decision-making process (Vishnawath & Kaufman, 2001; Christensen 2002; Bushman, Piotroski and Smith, 2003; Turilli & Floridi, 2009; Bhaduri & Ha-Brookshire, 2011). Hence, in this research transparency is referred to as deliberately revealing information that is relevant for the consumer.

!

“Transparency

can be referred

to as

deliberately

revealing

information that

is relevant for

the consumer

”.

(9)

Dimensions of Transparency

In an attempt to measure transparency, Vishnawath and Kaufman (2001) described four constructs in a b2b setting that helped put a measure to transparency: 1) access, 2) comprehensiveness, 3) relevance and 4) quality and reliability.

Accessibility is described as the ease and possibility for accessing useful

information. Thus information should be made available, without any intervening costs and should be understandable as well. If language is incomprehensible, the information does not come across.

Information relevance (henceforth relevance) is context dependent and might diﬀer

from person to person (Zimmer, Arsal, Al-Marzouq & Grover, 2010). That relevance is person dependent is described by Schamber, Eisenberg and Nilan (1990): “Relevance is the dynamic

exchange of information and communication depending on the relationship between information and the needs of information users” (p. 770). Schamber et al. carefully outlined the meaning of information relevance because the nature of relevance back than was still unclear. They compare a traditional and alternative perspective regarding relevance in the information science. Traditionally relevance is used as a judgement of eﬀective contact between system and user, whereas the alternative explanation of relevance looks more at conventional definitions of transparency: the judgement of quality of relationship between information and user’s information need. Years later Zimmer et al. defined information relevance as “The information to be disclosed is perceived to be useful or legitimate to the

function of the website” (p.117). This shows a consensus of the importance to approach

relevance in terms of the need of information utilization from a consumers perspective. Thus to know what is relevant for the target audience, company’s should examine users needs. In regards to transparency, it seems that people value most information about their safety, in terms of health, costs and quality, and business practices, such as supply and product development (Bhaduri & Ha-Brookshire, 2011; Cohn & Wolfe 2013). Cohn and Wolfe (2013) conducted a survey among 3000 adults and these people were more likely to show positive attitudes when they even knew about a company’s attempt to be transparent about these kinds of practices.

Quality and reliability are the last dimension of a message in order to be considered

transparent. According to Vishnawath & Kaufman (2001) quality standards must be set and they stipulate that information should be reliable and consistent in order to be eﬀective. Coherence theory can explain why, as it shows that information, whether it is negative or positive, should be coherent with the already existing truth in the mind of consumers (Murphy & Medin, 1985; Teece, Rumelt, Dosi & Winter, 1994). For coherency, consistency is very important because for a message to be perceived coherent, it should be consistent and in line with what is expected. According to the coherence theory, when the message is consistent, trust increases and the message will be considered as reliable. Chou, Yen and

(10)

Kuo (2011) found that quality of information disclosed is highly important in order to verify the correctness and timeliness of abundant online information. This also stipulates the importance of comprehensiveness of messages.

Comprehensiveness is the degree in which information is complete. Research by

Cohn & Wolfe (2013) found that keeping back information can lead to serious anger among customers. The greater the access to information and a greater demand for transparency made customers more critical and therefore the presumption that a company withholds information arises quickly. The completeness of information thus can be an important determinant of perceived transparency and in turn keeps the customers satisfied.

!

Effects of Transparency

Previous research about transparency shows that being transparent leads to a higher level of perceived honesty and trust and decreases skepticism (Beulens, Broens, Folstar & Hofstede, 2005; Eisend, 2006). This research shows that guided by an increased level of trust, transparency can even lead to favorable consumer behavior. But what if transparency means a company should disclose negative information? Research mostly show that the more negative a message, the more likely the receiver of a message will conform to it. Not only does that lead to unfavorable attitudes, it is much better remembered as well (Lee, Park & Han, 2008; Bambauer-Sachse & Mangold, 2011). The information becomes more salient in the minds of consumer and is likely to result in negative word-of-mouth which in turn can damage the reputation of the company and lead to a decline of sales. Evolution theories confirm that people give greater weight to negative messages: they are signals for protection and thus should be remembered well. The dominance of this negativity can be called the negativity bias. Research shows that negative message often are more potent and dominant than positive messages (Rozin & Royzman, 2001; Baumeister et al., 2001). A possible explanation of this dominance of negative information is that it is often considered to be more informative than positive or neutral information (Herr, Kardes & Kim, 1991). Therefore transparency does not seem to be attractive when dealing with negative information.

However, several researches about information disclosure showed that the positive eﬀects of deliberately revealing information also hold when information is negative. The only exception is when information is highly unfavorable (Beulens, Broens, Folstar & Hofstede, 2005; Eisend, 2006). Sending out a negative message seems to be counterintuitive at first glance, but literature already showed that adding a negative message to a positive can even be an eﬀective way of advertising (Kamins & Marks, 1987; Eisend, 2006, 2007, 2010; Kao, 2011; Pierro, Giacomontonio, Giannini, Krugslanski & Higginslee, 2013). According to research of Arpan and Roskos-Ewoldsen (2005) dealing with negative information, is best done using a stealing thunder strategy: be the first to disclose the negative information, before someone else does.

(11)

Literature about two-sided messaging shows a way in which sharing negative information can work positively. Two-sided messaging is messaging wherein both negative and positive information is shared (Arpan & Roskos-Ewoldson, 2005; Ein-Gar, Shiv & Tormala, 2012). These studies showed that two-sided messaging can actually help a message to be received positively. This effect is called the blemishing effect. This positive effect of two-sided messaging is especially proven to hold when consumers already hold negative beliefs or attitudes about a brand, which is in line with the coherency principle (Teece, Rumelt, Dosi & Winter, 1994; Eisend, 2007). The most important reasons why adding negative information works well is because it leads to the consumer’s conclusion that the truth has been told which in turn contributes to perceived credibility. Thus it seems that transparency, even if it information is unfavorable, can still have a positive outcome, especially on trust. As learned before, trust is a good estimation of favorable attitudes and behaviors (Beulens, Broens, Folstar & Hofstede, 2005; Eisend, 2006; Lee, Park & Han, 2008; Bambauer-Sachse & Mangold, 2011).

!

A rating of the four constructs of transparency together will define the degree of transparency. One can assume that the degree of transparency contributes to the level of engagement as these four constructs are important for an understanding and a willingness to consume the message. When the message is not directly interesting, incomprehensible or lacks quality, one is expected not to pay attention to the message and thus not engage in any form of interactivity. If the information is highly relevant, it is connected to the perceived needs of the customer and the customer is more likely to interact with the information. Thereby, these constructs all contribute to the perceived perception of trust, which leads to favorable behavior. Therefore we expect to find a positive eﬀect of the level of transparency and the level of engagement:

!

H₁: The higher level of transparency of a brand post, the higher the level of engagement with a brand post

!

Types of Transparency

Hultman and Axelsson (2007) conducted a research about transparency in marketing management. They defined four types of transparency in a business to business (b2b) context: cost transparency, supply transparency, organizational transparency and transparency (see figure 1). All definitions can be assessed again in degree, direction and distribution of transparency. However, in this research only the types of transparency will be used and will be applied in a new context: business to consumer (b2c). As the definitions of the types of transparency have been conducted in b2b, the definitions should be reviewed and redefined in applicability for b2c:

(12)

Cost transparency is defined as the information disclosure on costs as well as on

price, which is also applicable to the new context of b2c. In this research, cost transparency therefore will be practiced as all relevant information for consumers about prices. This can be with regards to the expense of consuming a product or service, the transparency about additional costs and prices.

Supply transparency is about the transparency in various flows of products and

materials between a buying and supplying firm. An example used is a track and trace service that heightens the degree of visibility. This type of transparency is also interesting in a b2c setting but not directly applicable to this social media research and therefore will not be used.

Organizational transparency’s fundamental idea is the expansion of relational horizons

in a business, or in this case, a consumer relationship: the horizon to which a consumer can see into the service provided by the firm is extended. This can be for example through information about the employees that are working for the company, information about where and how products or services are developed and projects a company is working on.

Technological transparency is the last type of transparency as described by Hultman

and Axelsson (2007). Within this type of transparency the extension of technological horizons is key, in terms of firms sharing the use of such technologies with other business partners. To make this applicable to the b2c market, technological transparency should be applied as the sharing of relevant technological developments of the company, which extends the horizon of the consumer.

When it comes to transparent matters, a survey of Cohn & Wolfe (2013) show that people are mostly interested in accurate reporting of earnings and pricing of companies. Subsequently consumers indicate that a company should be open about its safety standards, material/ingredients usage and their partners and suppliers in order to be deem ed transparent. Research in the business and finance field conducted by Chou, Yen and Kuo (2011) show that people are most interested in risk-related information, followed by stakeholder and financial items respectively. Both feed the presumption that cost transparency, followed by organization and technological information, have a diﬀerent influence on perceived transparency and the associated customer behavior. A trend towards the importance of cost transparency is visible, assuming that cost transparency leads to higher levels of engagement than technological and organizational transparency.

!

H₂a: Cost transparent brand posts are more popular than organizational and technological transparent brand posts

The rapport of Cohn & Wolfe (2013) shows that consumers find transparency about employee satisfaction and company facilities also important. Information about innovations does not belong to what people find important for the perceived transparency. Thus, in turn,

(13)

organizational transparency is expected to have a higher impact than technological transparency.

!

H₂b: Organizational transparent brand posts are more popular than technological transparent brand posts

!

Control Variables

Kaplan and Haenlein (2010) did exploratory research about the use of social media and found that reach is important for high levels of engagement. It is also logical that the higher the reach the more popular a post seems to be (although relatively this does not have to be the case). Therefore, reach is taken into account as a control variable in this research. According to research of Campbell (2013) the degree of reaching your audience is indeed important for the popularity of a post. To bump up your online presence as much as possible, the timing is extremely important (Campbell, 2013). The best timing for posts differ among social channels. According to Campbell, 80% of mobile users check their phones in the early morning after waking up. This trend is also visible in the results they found on how timing generates traffic. Facebook and LinkedIn show a clear patron of high traffic rates in the early morning and (late) afternoon. Worst are late-night posts and posts posted in the night. The latter is also applicable to Twitter, but the best timing leave a small question mark: between 1 and 3 pm which is contradicting within working hours . However, this is also most rewarding on weekends instead of weekdays which also backs up the theory that posts outside working hours work best.

In 2012 Buddy Media (the Facebook management system of choice for eight out of the ten top global advertisers) conducted extensive research among their clients, analyzing all Facebook posts during two weeks. Their sample size represented the world’s largest brands in the entertainment, media, retail, automotive, business and finance, fashion, food and beverage, healthcare and beauty, sports and travel and hospitality industries. The engagement rate they measured was a combination of the comment and like rate, factoring in fan base size. They found that for example that a short length post (80 characters or less) have 27% higher on engagement rates. Also, URL shorteners are negative predictors for engagement compared to full links. It seems that transparency plays a role in this, as full, long URL’s score three times higher on engagement rates as people like to know where they are directed to. A lot more diversity exists about more complex things such as which day of the week or on which time of the day should you do a post.

Advertising research shows that weekdays are far more popular on surfing the internet compared to the weekends (Rutz & Bucklin, 2011). This can result in a higher level of interactivity on posts during weekdays. However, the research of Buddy Media shows that there is a wide diversity between industry’s on weekdays, but they use one key rule: “Post

(14)

working hours as most of all posts are published between 10 am and 4 pm (see graphic 1). However the research of Buddy Media shows that engagement seems to be 20% higher outside office hours, which is backed by most of the other research. Therefore a difference in post popularity is expected dependent on if the post is shared within or outside office hours.

!

H₃: Brand posts outside oﬃce hours are more popular than posts within oﬃce hours.

!

Graphic 1

Timing and number of brand post versus engagement rate (Buddy Media Report, 2012)

!

Not only did timing of the day seem to have an influence on post popularity, but the days of the week that posts are shared also differ in effectiveness. According to the research of Buddy Media, 86% of the posts are published between Monday to Friday. They found that Facebook usage is highest when people have downtime, which is probably why user engagement is peaking in the weekends (and outside office hours). Exceptions are found, for example, in the business & finance and food & beverages industries where posts score best in the middle of the week or on sunday. The exact engagement rates by days might differ among industries, but this trend is found in the research over 18 different industries and thus can be used as a good estimator. Therefore is expected that posts during weekends provide for more popular posts.

!

H₄: Post shared in the weekends score higher on engagement than post shared at the during the week.

!

(15)

Graphic 2

Engagement rate for days of the week (Buddy Media report, 2012)

Research of de Vries et al. (2010) proved the importance of vividness of posts on brand post popularity. On the other hand, the relation between informative posts and brand post popularity could not be found. This seems to imply that posts should be triggering, which would be in favor of visual content such as images or video’s. The research of Buddy Media points at photo’s as the biggest driver for interactivity. However, videos’ come third after status, which is a plain text with or without a url. Other sources also argue the power of visual media sharing “Eight eﬀective ways to engage your customers with social media marketing”, 2014; “Photo Posts Spark the Most Engagement on Facebook”, 2014). Therefore one can expect that photo’s and video’s are much more likely to generate interactivity and thus engagement compared to a plain status update.

!

H₅: The sharing of visual content in a brand post generates higher popularity rates than a

textual status update.

!

There is a lot of diﬀerent social media platforms to engage in. A few obvious ones are Facebook, Twitter and LinkedIn which are used by a lot of company’s. However, there are a lot of other possibility’s for companies such as engaging on Instagram, Pinterest, Tumbler, Google+, Myspace or starting a blog of your own. It might not always be most profitable or rewarding to engage in the biggest platform. For some companies it can be much more rewarding to engage in smaller ones where their main target group is most active (Kaplan & Haenlein, 2010;

Wexler & Kepner, 2013; “Which social media platform is best for your business?”, 2014). When diﬀerent platforms are used, highly important is that the message is congruent among

“Nothing is more

confusing than

contradicting

messages

across different

channels

”

(16)

all channels. A big part of the reason of engaging in social media content is to communicate a clear message of the companies activities/values and reduce uncertainty. As Kaplan and Haenlein (2010, p.65) put it: “Nothing is more confusing than contradicting messages across

diﬀerent channels”.

Different social platforms also attract different kinds of people. On LinkedIn for example, the average age is 44 years. whereas on Facebook and Twitter the most active users are women between 18 and 29 years old. The difference between these platforms can also be found in the description: “Twitter is a social micro-blogging network that limits 140

characters to openly message and communicate with anyone or a brand”; “Facebook makes the world open by letting people stay connected to friends and family, discover what is going on, and to share what matters to them”; “LinkedIn connects the world’s professionals to

make them more productive and successful” (Wexler & Kepner, 2013, p. 6,7 &9). Twitter is a micro-blogging site that forces you to be straightforward and convey your message in only 140 characters. Facebook is built to share your life with friends and family: not necessarily with brands, thus Wexler and Kepner. In comparison with Twitter, Facebook might be more useful to explain the products and services or the company’s values widely. LinkedIn is more professional than both Twitter and Facebook and therefore is expected to be more suitable for b2b expressions rather than b2c (“How to increase your social media reach: Tailor content to different platforms”, received on June 22nd, 2014). This might lead to a difference in expectations from platform users and as learned from the coherence theory, the b2c expressions thus might be better in place on Twitter and Facebook and therefore score higher on engagement (Murphy & Medin, 1985; Teece, Rumelt, Dosi & Winter, 1994). Because Facebook has more space to disclose information, it is plausible that consumers perceive Facebook messages to be more transparent and thus lead to a higher level of engagement. Therefore a difference between the platforms are expected to give different results on brand post popularity, where Facebook will have the highest post popularity rates and LinkedIn the lowest.

!

H₆: Brand posts on Facebook provides higher levels of brand posts popularity than brand posts on Twitter and LinkedIn

!

(17)

Figure 1

Summary of the research model

!

(18)

Study Design

Operationalization of Variables

T

his study investigates if transparency explains engagement. Following the study of de Vries, Gensler and Leeflang (2010) this study measures engagement by means of brand post popularity. In contrast to de Vries et al., this research makes use of brand posts on three diﬀerent social media channels instead of only one: Facebook, Twitter and LinkedIn. All analyzed brand posts are from shared by Triodos Bank NL. Triodos bank is a specialized bank with a diﬀerent approach than an average bank. It has three main pillars: transparency, sustainability and ethics and then specifically with regard to people, planet and profits. For the measurement of transparency, the posts are carefully studied and given a score based on the four dimensions of transparency as described in the literature review: 1) accessibility, 2) relevance, 3) quality and reliability and 4) comprehensiveness (Vishnawath & Kaufman, 2001).

Accessibility in this research is measured as comprehensible. In operationalizing what is understandable, the average education level of the people reached with the message should be taken into account. In the Netherlands the education level is fairly average and thus one can assume that the understanding level of messages is average as well. (Minsterie van Onderwijs, Cultuur en Wetenschap, 2013). Regarding the data of Triodos, posts that contained jargon has been coded as low comprehensibility as the target is the average customer, not a business partner.

As seen in the literature review, relevance is based on consumer needs. Research of Cohn and Wolfe (2012) ordered the most important indicators for a organization to be perceived transparent in the minds of consumers from most to less important. First mentioned as an important determinant is information regarding where a company sources its materials/ingredients. Second in row is the funding/ownership of the company. Third, the earnings (profits, losses) and fourth information about who the company does business with (suppliers). One year later, Cohn and Wolfe (2013) found that the first two have switched, but still the importance to the consumer of both safety and sources of materials and ingredients is clear. This is also found by Bhaduri and Ha-Brookshire (2011), who showed that a company’s attempt to be transparent about these factors, influences the purchase decision of consumers. As this research uses data of a business in the finance industry, the need for safety and quality can be translated into information that directly impacts the consumer. Thus, when applied to Triodos’ data, a relevant post should contain information that directly by safety and quality, or indirectly influences the customer by information of how the company does business with, for example, the savings and investments of its customers. For example: “Via Triodos Bank you can invest in green projects that protect and extend

(19)

Drenthe" (free translation from a Dutch post). This post is coded as highly relevant as Triodos

is open about the business practices they are investing in, with the customers money. Less relevant are posts about the company’s earnings, as these seem of less relevance to the consumer according to Cohn and Wolfe (2012), for example: “The annual numbers over 2013

are out. Last Friday Peter Blom presented the numbers on the milk/cheese event at cheese farmer ‘De Groene griﬃoen”. Do you want to know how Triodos bank performed in 2013? Read the press release here: http://bit.ly/1jMaigx” (free translation from a Dutch post).

For quality and reliability it is important that a post is in line with what is expected of the company by its customers. For example a poorly explained message, a message that is not in line with the core concepts of Triodos or a message containing orthography errors can be coded as not qualitative or reliable. An example of a high qualitative and reliable message is “Would you like to support small entrepreneurs in a developing country? With Triodos Bank

its possible to invest in institutions for micro financing. This makes it possible to make a big chance with a small amount of money. Read more on: http://bit.ly/W11fA7” (free translation from a Dutch post).

When comprehensiveness is applied to social media research, it can be described in terms of the presentation of direct, free information without intervening parties or actions. Thus comprehensiveness means that information not only should be made available, but there also should be easy access to complete information. As it is hard to retrieve which information is out there which is not shared by Triodos, a measure only can be made of how accessible the available information is. An example of a low scoring post on comprehensiveness is “Growth is more than a number. But what is growth? And is growth

necessary for prosperity? http://www.triodos.nl/nl/over-triodos-bank/nieuws/actueel/wat-is-groei/ ” (free translation from a Dutch post). This post does not give answers in the post at all, if you want to know the answer you are directed to first watch a movie, on another website. Thus the information has not been made easily available.

For each dimension the brand post could score 1 (low transparent) to 3 (high transparent). In total a score of 4 to 12 could be obtained, which is divided by 4 for a final code of 1(low transparent) to 3 (high transparent) (see table 1). Thus, transparency in this research is calculated using a continuous 3 point scale.

The level of engagement in this research is measured by the number of likes (or favorites), comments and shares (or retweets). A new variable has been created called total interaction: this is the sum of the amount of likes, shares and comments for one post.

!

The possible moderation variable type of transparency is dispersed in cost, organizational and technological transparency. Applied to Triodos’ data cost transparency is operationalized as information about mortgages, interest on loans and annual costs of

(20)

having an account at Triodos. Also annual reports can be scheduled under cost transparency as this is an important parameter for understanding the costs and prices of Triodos. Organizational transparency is information about the employees that are working for the company, information about where and how products or services are developed and projects a company is working on. For Triodos this can imply transparency about investments, projects they are supporting, press releases and employees’ activities. Technological transparency for Triodos is interpreted as information about technological developments that improves the ease or understanding of using Triodos products or services, for example information about mobile banking and jammings in the systems.

The control variable time of the day is categorized in time zones which are quite similar, for example early morning, around lunch, afternoon and night. See table 1 for a summarized overview of how the research variables are operationalized and the appendix for a full overview.

!

Table 1

Overview of the research variables (for extended value coding information, see appendix 1: coding scheme)

!

Variable Type Values

Transparency Independent variable Scale

Low transparent (1) to high transparent (3)

Engagement (in dataset “interactivity”) Dependent variable Continuous

Type of transparency Moderator Categorical

Cost (1), organizational (2), technological (3)

Total reach of post Control variable Continuous

Type of post Control variable Categorical

Status (1), photo (2), video (3)

Part of the day Control variable Categorial

Early morning (1) to late evening (6)

Part of the week Control variable Categorical

(21)

Data

The dataset consists of computer-generated output of Triodos social media channels Facebook, Twitter and LinkedIn over a period of 15 months starting on January 1st in 2013. After a manual selection to distinguish all

useful content of the big data file, a dataset remained of 349 brand posts (see graphic 1). Some of the remaining data is coded into scaled codes so the data can be compared and analyzed (see appendix 1 “coding scheme”). The average number of interactions (likes, comments and shares) with the posts diﬀer extremely among the channels with a rounded mean of 161 (SD=179.634) for Facebook, 11 (SD=11.385) for Twitter and for LinkedIn 57 (SD=111.04). Even wider diﬀerences are found in the average reach of the posts: 12179

(SD=58839.416), 1997 (SD=655.751) & 22854 (SD=62253.713). This shows a large amount of variation within the data.

!

Table 2

Partial Correlations dependent and independent variables

!

Bold figures: p-value <0.05, Italic figures: p-value <0.1

!

Control variables Comments Likes (favorites) Shares (retweets) Total Engagement Transparency Type of post, day of the week, part of the day, total

reach Comments 1,000 0,846 0,806 0,856 0,098 Likes (favorites) 0,846 1,000 0,851 0,931 0,196 Shares (retweets) 0,806 0,851 1,000 0,883 0,066 Total Engagement 0,856 0,931 0,883 1,000 0,155 Transparency 0,098 0,122 0,066 0,155 1,000 Graphic 1

(22)

Methodology

In the case of Twitter and Facebook, the three dependent variables for consumer engagement are likes (favorites), comments and shares (retweets) are measured separately. The LinkedIn data output makes no distinction between likes, comments or shares and only gives a total number of interactivity. As can be seen in Table 2, the correlations between the diﬀerent engagement measurements (comments, likes and shares), are of a subsequent high level, that engagement will be measured as one construct: the sum of all likes, comments and shares. This means that the data of all social media channels can be used in the same way, testing for one dependent variable: total interactivity as measurement for brand post popularity, hence engagement. The independent variable in this research is the level of transparency. Independent control variables are the type of post, day of the week, time of the day and total reach (see table 1). The influence of the level of transparency on the level of engagement, is measured using a multiple OLS regression, while controlling for total reach, part of the day and time of the day. To gain understanding of the eﬀects found in the regression analyses, a follow-up one-way anova for each variable will be conducted.

The relation between transparency and engagement will also be tested if it is influenced by moderation eﬀects of type of transparency and social media channel. The moderation model can be expressed as: Y= b₁X+b₄M+b₅W (see figure 2).

!

Figure 2

Moderation model

!

Because the moderators are nominal variables (not continuous), the moderation model is tested using the Process macro developed by Preacher and Hayes (Preacher and Hayes, 2008; Preacher, Rucker and Hayes, 2007; Hayes, 2012;). This macro uses and ordinary least square or logistic regression analysis in the estimation of eﬀects of both mediator and moderation models.

Y = the sum of likes, comments and shares as measurement for engagement,

X = the degree of transparency,

M = type of transparency,

(23)

Results

Most posts were shared during week days (Monday - Friday) and between 07.00 am - 11.00 pm Looking into the eﬀect of transparency on engagement, noticeable is the fact that 35% of the posts were coded as not transparent (see graphic 2). The remaining transparent posts showed a positive significant correlation with engagement (see table 2). This confirms the possibility of a causal relationship between transparency and engagement.

The model as a whole was found to be significant (F-value = 23.616, p-value < 0.01), confirming the independent variable and the control variables as an explanation of the variance found in the dependent variable level of engagement (R² = 26%, adj. R² = 25%). The level of transparency is significant and positively related to the level of engagement (β-transparency = 0.138, p-value < 0.05), in support of hypothesis 1.

!

Graphic 2

Brand posts characteristics of when it is posted and what is posted

!

(24)

The moderation model was also found to be significant (F-value = 18.096, p-value < 0.01), confirming the eﬀect of the moderators ‘type of social media channel’ and ‘type of transparency’ as an explanation of the variance found in the relation between transparency and engagement (R² = 33%, adj. R² = 11%). However, no significant influence (p-value > 0.05) was found for type of transparency as influencer of the level of engagement. Also type of transparency does not moderate the relation between transparency and engagement (p-value > 0.05). Therefore hypothesis 2 should be rejected.

Although brand posts seemed to be most popular with highest interactivity rates in the early morning and the evening, a significant diﬀerence cannot be found (β-time of the day= 0.062 p-value > 0.05). Therefore hypothesis 3 should be rejected. Also for day of the week no influence can be found, although engagement seemed to be highest on weekends (β-day of the week = 0.018 p-value > 0.05).

Hypothesis 5 however can be supported, as a positive significant relation has been found for the type of post and total interactivity (β-type of post= 0.176 p-value < 0.05). Thus the type of post explains diﬀerences in the level of engagement. Posts containing photo’s are most popular, second video’s and of least influence are simple status updates.

The type of social media channel was found to be of significant influence on brand post popularity ( p-value < 0.05). A follow-up one-way Anova shows that this eﬀect is highest for Facebook content, followed by LinkedIn and Twitter respectively. However, examining the interaction eﬀect, the results show that type of channel does not necessarily interact in the relation between transparency and engagement ( p-value > 0.05). This implies that hypothesis 6 also should be rejected.

!

Table 3

Summary of Results

!

Hypotheses Expected Result

H₁: level of transparency + Supported

H₂𝖺: type of transparency (cost vs organizational) + Not supported H₂𝖻: type of transparency (organizational vs technological) + Not supported

H₃: time of the day + Not Supported

H₄: day of the week + Not Supported

H₅: type of post + Supported

(25)

!

Table 4

Regression analyses results of Time of the day, Day of the week, Type of post, Reach, Transparency and Engagement (dependent variable)

!

Table 5

Process moderation analyses results of Type of Transparency, Social Media Channel, Transparency and

Engagement (dependent variable)

!

B SE(B) β t Sig. (p)

Time of the day 7.035 5.290 0.062 1.1330 0.184

Day of the week 1.942 5.032 0.018 .386 0.700

Type of post 55.091 14.796 0.176 3,723 0.000

Reach 0.001 0.000 0.449 9.431 0.000

Transparency 30.115 10.370 0.138 2.904 0.004

B SE(B) t Sig. (p)

Type of Transparency (M) -17.465 15.998 -1.0917 0.276

Social media Channel (W) -60.275 8.133 -7.411 0.000

Transparency (X) 12.050 16.902 0.713 4.956

Interaction XM -1.061 21.457 -0.050 0.961

(26)

Conclusions and Discussion

The aim of the research was to investigate if and how transparency improves customer engagement. De Vries, Gensler and Leeflang (2010) already added to the literature that vividness, position and share of positive comments on and interactivity of brand post are predictors of customer engagement. However, an eﬀect of the information shared in brand posts could not be found. This is contradictory to the expectations and findings of this research, where the information of brand posts did seem to matter. The higher the level of transparency, the more popular we found a brand post to be. Thus, this research added to the knowledge that transparency is a predictor for engagement. This is in line with the survey research of Cohn and Wolfe (2013) where people indicated to increasingly appreciate transparent companies. The findings of this study confirm the fact that consumers are more demanding about transparency but added prove to the previous knowledge that consumers also act on it. In accordance to Cohn & Wolfe (2013) the results are very generalizable: the levels of engagement rises after a companies attempt to be transparent. This outcome is a big contribution to the transparency literature as we know now that transparency not only leads to higher levels of trust and possibly favorable behavior, but also that transparency leads to higher levels of engagement (Bhaduri & Ha-Brookshire, 2011).

The four types of transparency of Hultman and Axelsson (2007), of which we translated three to a b2c environment, could not have been recognized. The type of transparency did not seem to be a factor of importance, as no significant effect could be found in explaining any variance in the relation to transparency on engagement. It is a confirmation that as long as people recognize an attempt of a company to be transparent, people will value it. This means that the types of transparency could not be applied in a b2c setting, with this interpretation. However, the lack of effect might be caused by the fact that only 8,6% of the posts were cost and technological transparent and thus a comparison between the constructs is not quite proportional and the chance to find an effect decreases. In confirmation of earlier findings from both scientific and non-scientific research, total reach was found to be of significant influence on the brand post popularity (Kaplan & Haenlein, 2010; Campbell, 2013). The results of the other control variables were mainly surprising. Although (non-)scientific research found clear differences in engagement levels on different times during the week or the day, no effect could be found for the timing of when posts were shared (Rutz & Bucklin 2011; Buddy Media, 2012). Thus, it does not seem to matter which day of the week, or which time of the day a company shares a post. Therefore the findings of previous research cannot be confirmed for the scientific social media literature. However, interesting for the latter are the findings that the type of post and channel did matter (see figure 3).

We found that brand posts with a photo were by far the most popular ones (see figure 3). Following pictures, video’s were the best predictors of engagement, and least

(27)

popular among social media users were simple status updates containing plain texts or links. This is partially in confirmation of the study of de Vries

et al. (2010) who found that the higher the level of vividness of a brand posts, the more popular a brand post. Indeed vivid posts did best, but video’s can be considered as even more vivid than photo’s but scored less. This can be possibly explained by some of the determinants of transparency: accessibility and comprehensiveness are extremely important (Vishnawath & Kaufman, 2001). Video’s might be perceived as harder to access as you first have to watch a movie before full information can be consumed, whereas photos are pretty straightforward and less time consuming as well. All comprehensive information is available at once.

To my best knowledge, no other study ever tested the level of engagement measured by brand posts popularity using multiple social media channels. By using this measurement method, a difference have been discovered between Facebook, Twitter and LinkedIn in its effect on brand post popularity. Although no effect has been found for type of channel as a moderation variable in the relation between transparency and engagement, it did has a direct effect on the levels of engagement. The research learns that Facebook leads to the highest levels of engagement. However, the reason why does not seem to come from the fact that Facebook offers a large amount of space to communicate a message and thus come across as very transparent, as discussed in the literature review. Also the coherency theory cannot be an explanation of this result (Murphy & Medin, 1985; Teece, Rumelt, Dosi & Winter, 1994) According to the coherency theory was assumed that the brand posts were less of influence on LinkedIn, as brand posts are created as b2c expressions and LinkedIn is a b2b platform, wherefore the messages might not be accordance people’s expectations. This research showed however, that LinkedIn is second in row in providing high levels of engagement. A possible explanation of why type of channel cannot be recognized as a moderator, might yet again be the cause of people valuing any attempt of transparency and therefore it does not matter where this attempt is shown (Bhaduri & Ha-Brookshire, 2011; Cohn & Wolfe, 2013). As long as companies listen to the increasing demand for transparency, people seem to be happy about it.

!

Figure 3

Result of ‘type of post’ on engagement* 

(28)

Managerial Implications

Enhancing the level of engagement

For those managers responsible for digital customer interactions and engagement in all shapes and forms, the results of this study are highly valuable in deciding on the brands’ b2c social media and/or communication strategy. The main result is the proven eﬀectiveness of transparency in increasing customer engagement rates. In order for a post to be perceived as transparent, a messages should be, most of all, relevant and useful for the receiver. Thereby messages should be as comprehensive, reliable and easily accessible as possible and in line with what consumers expect from the brand. However, no eﬀect is found for the types of transparency defined by Hultman and Axelsson (2007). That means that to enhance the level of engagement, any attempt of being transparent will be valued and thus the subject of transparency does not matter. A company or brand can choose to be transparent about, for example, cost structures or the supply chain, both will lead to higher levels of engagement.

When a company would like to bump the engagement levels, adding vidid content such as a photo or a video also contributes, as long as it is still relevant for the customer. The control variable ‘reach’ has been examined exploratory and found to be of significant, positive influence on customer engagement. Managers should be seriously considering strategies and methods to increase their levels of reach in order to maximize customer engagement, by, for example, buying reach on the social media platform.

This research found proof that engaging on Facebook leads to highest brand post popularity rates. This implies that the channel can be important, and according to this research Facebook is the most engaging one. However, as learned from Wexler and Kepner (2013), social media platforms diﬀer in function and users. Therefore, a manager should really think through which social media channels to involve in their social media strategy as this can be of significant influence in the level of engagement. Thereby one should think about who is the target group and what is the goal of sharing posts.

For the risk and reputation management field this research provides new insights as well, as transparency has proved to be an eﬀective in terms of favorable behavior. This sheds the light on the theory of stealing thunder, to examine this as well in a digital, social environment (Arpan and Roskos- Ewoldsen, 2005).

!

(29)

Limitations and Further Research

To test all hypotheses, several tests have been conducted. Separating the analyses increase the possibility of alpha inflation and the risk of a type I error. Conducting several tests at α= 0.05 leads to a higher chance of an accidental eﬀect found within the 95% confidence interval. This implies that the chance grows that a result in this research was found to be significant when in fact the H₀ hypothesis should have been rejected.

Although the results of this study can be fairly generalized and applied to all b2c activities, one must consider that Triodos is a special company that tends to do business in an ethical and transparent way. Therefore Triodos’ customers, which is probably a big part of the people that are reached with their brand posts, might be consumers that value transparent messages. The coherence theory confirms this possibility: people value information that is in line with an already existing truth and thus with what is expected (Murphy & Medin, 1985; Teece, Rumelt, Dosi & Winter, 1994). For Triodos transparency might be something that is expected and therefore leads to a higher level of engagement than it would do for a company without transparency as a pillar. On the other hand, it might also work the other way around: the fact that transparency is paramount for Triodos, can also be detrimental as consumers might be more critical and more demanding in even higher levels of transparency. This could imply that for other companies the transparent expressions might be valued higher and the relation found could be even greater. Therefore this study should be repeated on a diﬀerent company to rule out the possibility that the nature of a company is of any influence in the relation of transparency and engagement.

As seen in the research of Buddy Media (2012), results might also diﬀer for the variety of industries within the b2c market. Therefore future research in a diﬀerent b2c setting should prove the sustainability of the findings for all b2c industries.

!

(30)

References

!

“De 47 slechtste social media cases #fail”, 2010. Received on June 22nd_{, 2014 from}

http://www.dutchcowboys.nl/socialmedia/20977

!

“Eight eﬀective ways to engage your customers with social media marketing”, 2014. Received on June 23rd_{, 2014 from http://www.businessbee.com/resources/marketing/}

social-media-marketing/8-eﬀective-ways-to-engage-your-customers-with-social-media- marketing/

!

“How Nestlé dealt with a social media campaign against it” , 2012. Received on June 22nd_,

2014 from http://www.ft.com/cms/s0/90dbﬀ8a-3aea-11e2b3f0-00144feabdc0.html# axzz367tJMzEf

!

“How to increase your social media reach: Tailor content to diﬀerent platforms”, received on June 22nd _{2014 from www.insightpool.com/how-to-increase-your-social-media-}

reach-tailor-content-to-diﬀerent-platforms/.

!

“Nestlé kerfuﬄe presentation”, 2011. Received on June 22nd_{, 2014 from http://prezi.com/}

kmrh4fmlzsen/nestle-kerfuﬄe/

“Photo Posts Spark the Most Engagement on Facebook”, 2014. Received on June 23rd_,

2014 from http://www.marketingprofs.com/charts/2014/24924/photo-posts-spark- most-engagement-on-facebook#ixzz36E2kAg1X

!

“Which social media platform is best for your business?”, 2014. Received on June 15th_{, 2014}

from http://thenextweb.com/socialmedia/2014/03/05/social-media-platform-best- business/

!

Arpan, L.M., Roskos-Ewoldsen, D.R. (2005). Stealing thunder: Analysis of the eﬀects of proactive disclosure of crisis information. Public Relations Review, 31(3), 425-433.

!

Atkinson, D. (2014). Received on June 15th from http://www.entrepreneur.com/article/ 232645.

!

Barakat, C. (2014). Received on June 15th_from

(http://socialtimes.com/best-worst-times-post- social-media-infographic_b141745

Buddy Media (2012). Strategies for Eﬀective Wall Posts: A timeline Analysis, 1-32. salesforce.com, inc.

!

Beulens, A. J., Broens, D., Folstar, P., & Hofstede, G. J. (2005). Food safety and transparency in food chains and networks relationships and challenges. Food Control, 16(6), 481-486.

!