Factors influencing popularity of branded content in Facebookfan pagesFerran Sabate a Jasmina Berbegal-Mirabent b Antonio Cantildeabate c Philipp R Lebherz d

a Universitat Politegravecnica de Catalunya bull BarcelonaTech Department of Management C Jordi Girona 1-3 Edifici C5 08034 Barcelona Spainb Universitat Internacional de Catalunya Department of Economy and Business Organization C Inmaculada 22 08017 Barcelona Spainc Universitat Politegravecnica de Catalunya bull BarcelonaTech Department of Management C Jordi Girona 1-3 Edifici C5 08034 Barcelona Spaind Karlsruher Institut fuumlr Technologie (KIT) Kaiserstraszlige 12 76131 Karlsruhe Germany

A R T I C L E I N F O

Article historyReceived 22 October 2013Accepted 11 May 2014Available online

KeywordsContent marketingSocial networking siteConsumer engagementSocial media optimizationFacebook brand page

A B S T R A C T

Social media is achieving an increasing importance as a channel for gathering information about prod-ucts and services Brands are developing its presence in social networking sites to meet brand aware-ness engagement and word of mouth In this context the analysis of the factors that are conditioningconsumer interaction with branded content becomes a matter of interest This paper aims to shed lighton those factors that are expected to impact on Facebook branded post popularity A conceptual modelis developed to reflect the influence of the contentrsquos richness and time frame on the number of com-ments and likes An empirical analysis using multiple linear regressions is conducted based on 164 Face-book posts gathered from the fan pages of 5 Spanish travel agencies Results suggest that the richness ofthe content (inclusions of images and videos) raises the impact of the post in terms of likes On the otherhand using images and a proper publication time are significantly influencing the number of com-ments whereas the use of links may decrease this metric

This study empirically contributes to the existing literature on the management of marketing strat-egies for consumer engagement in social networking sites

copy 2014 Elsevier Ltd All rights reserved

1 Introduction

Social networking sites (henceforth SNS) have become verypopular and have been increasingly attracting the interest of mar-keters They account about 6 of all website visits done and 19 ofall time spent online (Radwanick Lipsman amp Aquino 2011 Tuten2008) Social media is achieving more and more importance as achannel for gathering information about products and services andto take profit of new opportunities (Verhoef amp Lemon 2013)

SNS have stimulated new ways of interacting shaping new formsin which people communicate make decisions socialize collabo-rate learn entertain themselves interact with each other or evendo their shopping (Constantinides amp Fountain 2008 Hanna Rohmamp Crittenden 2011 Hansen Schneiderman amp Smith 2011 Mangoldamp Faulds 2009) Consequently the study of social media and itseffects on consumers and organizations is increasingly attracting ac-ademic attention while it opens new research avenues for strate-gists and marketers (Bughin amp Manyika 2009 Constantinides ampFountain 2008 Fischer amp Reuber 2011 Urban 2003)

Among others benefits arising from a well-designed social mediamarketing strategy may materialize in a better grasp of consum-ersrsquo behaviors and preferences making consumers share the brandrsquosmessage as word of mouth to their peers connecting to consumerfor improvement and RampD processes increasing brand engage-ment and brand message exposure as well as driving traffic to cor-porate websites (Hettler 2010 Smith amp Zook 2011 Tuten 2008)

People are gradually shifting their trust to recommendations andexperiences from other consumers SNS allow users to publish andinterchange opinions and experiences about brands and their prod-ucts and services Several studies confirm the influence of this usergenerated content on purchase intention and that this influenceapplies for different kinds of products and services (Chevalier ampMayzlin 2006 Dhar amp Chang 2009 Duan Gu amp Whinston 2008Rehmani amp Khan 2011 Sierra Saacutenchez 2012 Ye Law amp Gu 2009)

This behavior known under the term of word-of-mouth (BrownBroderick amp Lee 2007 Kozinets de Valck Wojnicki amp Wilner 2010Moldovan Goldenberg amp Chattopadhyay 2011 Trusov Bucklin ampPauwels 2009) contrasts with more traditional marketing com-munications and seems to be gaining importance with the rise ofWeb 20 Social media marketing should be oriented first to un-derstand clients second to create custom-made online content andthird to define a fitting strategy in a way that strengthens the rep-utation of the brand (Hettler 2010 Heymann-Reder 2011 Kilian

Corresponding author Tel +34 93 401 56 31E-mail address ferransabateupcedu (F Sabate)

httpdxdoiorg101016jemj2014050010263-2373copy 2014 Elsevier Ltd All rights reserved

European Management Journal (2014) ndash

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

Contents lists available at ScienceDirect

European Management Journal

journal homepage wwwelseviercom locate emj

amp Langner 2010) It is therefore a matter of interest to know whichcharacteristics should have this online content in order to be spreadby consumers without any enforcement or cost just because peoplelike to share the content to their peers in a fashionable but directand personalized atmosphere (Agresta amp Bough 2011 Constantinidesamp Fountain 2008 Fournier amp Avery 2011 Hettler 2010)

Content is the instrument that stimulates interaction Brands mustpublish pieces of content trying to address customersrsquo motiva-tions delivering interesting content for them when and whereneeded Successful content is adopted by customers adding valueby sharing it and producing derivative or original content that spreadsthrough peer-to-peer interactions Thus content reaches popular-ity thanks to those customers who positively interact with it con-tributing to its spreading and becoming brand advocates who caninfluence purchase decisions of others Postmodern consumers bringthe challenge of addressing these incentives individually and col-lectively (Sashi 2012 Scott 2007 Simmons 2008 Smith amp Zook2011)

Within Facebook dissemination of branded content or posts canbe achieved through several mechanisms Users who are fans of thebrand will see in their walls this branded content and then theycan interact with it by liking sharing or commenting Each of theseactions potentially promotes the content to all the customerrsquos friendsrsquowalls Consequently friends of fans can also contribute to exponen-tially disseminate this content

Like most SNS Facebook allows brands to create profiles andinteract with users Fan pages are brand oriented profiles that provideadditional functionalities like detailed analytics and better contentand fans administration Facebook characteristics ldquoprovide uniqueand interesting conditions for investigating the interaction of mul-tiple selves and the incorporation of brands in consumer self-expressionrdquo (Hollenbeck amp Kaikati 2012 p 396) From brands pointof view and according to the classification of Dholakia Bagozziand Pearo (2004) Facebook combines characteristics of small-group communities based on pre-existing offline relationships withthose of network-based communities where a member withoutpre-existing relationships connects around the brand through itsfan page As members of a brand community participation incen-tives are based on personal benefits Nevertheless fans canalso act as brand evangelists spreading branded content throughtheir friendsrsquo network where motivations to interact came fromsocial benefits This double affiliation to personal and brand com-munities is essential to understand word-of-mouth and the pathsthrough which the branded content reach popularity and spreadsvirally

Online social communities are identified as one of the key newmedia phenomena (Hennig-Thurau et al 2010) with research im-plications for the successful management of customer interac-tions Specifically how can communities be used for brandmanagement and how can a brand acquire virtual consumer friendsare some of the research questions that arise Based on this ratio-nale the analysis of the characteristics that makes branded contentpopular as well as the study of those factors that are conditioningconsumer interaction become a matter of interest for firms inorder to address their marketing efforts in social media in the correctdirection

Accordingly this work contributes to extend the knowledge onthose characteristics that make branded content popular by iden-tifying how richness and time frame of content publication influ-ence customersrsquo interaction Moreover Agresta and Bough (2011)state that there is no simple formula that guides on how to publishin social media due to diversity of brandsrsquo goals and sectorial char-acteristics Consequently we focus our analysis on the Spanish travelagencies sector by carrying out an empirical study of the posts pub-lished by five Spanish travel agencies on their Facebook fan pagesand the usersrsquo interaction with them

We find that the inclusions of images and videos raise the impactof the post in terms of likes Likewise images and publication timeare significantly influencing the number of comments whereas theuse of links decreases this metric These results draw implicationsapplicable for companiesrsquo social media marketing activities whichare interesting for academics as well as for practitioners

The remainder of the paper is organized as follows Section 2 pres-ents the theoretical framework Section 3 describes the sample andthe methodological approach Empirical results are offered in Section4 The discussion of the results and their managerial implicationstogether with the limitations of this study and potential researchavenues are displayed in Section 5 The paper ends with the mainconclusions summarized in Section 6

2 Theoretical background

21 Drivers for brand post popularity

According to Singh Jain and Kankanhalli (2011) there are no the-oretical frameworks available yet that could be used to analyze whyand how users contribute to social media Yet giving a simpleformula that guides on how to publish in social media is not pos-sible due to the particular circumstances of each brand and becauseof the very distinct set of goals and possibilities every business has(Agresta amp Bough 2011)

In this study we categorize content attributes of SNS accordingto a simple classification whether they are qualitative based on se-mantic analysis (soft criterion) or whether they are hints that areproved in a quantitative and empirical way (hard criterion)

The soft criterion considers the semantics and the interpreta-tion of the message behind a post Both Scott (2007) and Sterne(2010) argue that before publishing in SNS businesses should adopta consumerrsquos perspective and publish only those posts that reallyprovide value-added information for the reader The works ofHeymann-Reder (2011) and Hettler (2010) corroborate this state-ment In their respective studies they found that those posts re-vealing funny things of the working environment news affectingthe business or information that may report direct economic ben-efits to the reader are more prone to capture userrsquos attention Thesefindings indicate that post category has a significant effect over theuser interaction and as such should be used for planning of the com-munications strategy (aDigital 2011 Pletikosa Cvijikj amp Michahelles2011)

The main problem of the soft criterion is the difficulty in cap-turing and processing relevant data for analysis On the one handsoft criterion requires a careful content analysis of texts images orvideos On the other hand this analysis can be stigmatized as sub-jective as it may be difficult to properly discriminate between thosepublications that contribute to enhance the brand from those thatare damaging it This leads to a very meticulous analysis which ifmade manually is very tedious and time-consuming

Elements of SNS that can be quantified without the need of a sub-jective interpretation process can be considered under the hard cri-terion By this approach it is possible to compute the frequency andtiming a phenomenon takes place Likewise it allows assessing therichness of the content associated to a post by simply looking atthe content type that complements the text (ie a picture a videoor a link to another website) These quantitative factors are also char-acteristics of the published content but are more easily to captureand process than the former ones Thus given the resource con-straints in the collection and processing of data for the purpose ofthis study we have only focused on those characteristics of poststhat respond to the hard criterion Accordingly we aim to identifythose structural features of posts that act as drivers for brand postpopularity The next section presents the theoretical framework usedand the hypotheses that will be tested

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

2 F Sabate et alEuropean Management Journal (2014) ndash

22 Conceptual model

Fig 1 schematizes the conceptual framework considered wherethe number of likes and comments represent the metrics to eval-uate the popularity of wall posts We argue that richness (the viv-idness of the content of the post) and time frame (related to timeand date of publication) have a significant influence on brand postpopularity (measured by the number of likes and the number ofcomments) Furthermore the model includes two additional vari-ables controlling for size differences (number of followers of theFacebook brand page) and the length of the post (see Fig 1)

221 RichnessDifferent media types entail different capacity for immediate feed-

back This is so because of the richness (breadth and depth) of amessage (Daft amp Lengel 1986) This idea of richness is also com-monly referred to as vividness of the online content (De VriesGensler amp Leeflang 2012 Pletikosa Cvijikj amp Michahelles 2013)Indeed previous studies looking at how to enhance positive atti-tudes towards a website found that the richness of the message mayplay a role (Fortin amp Dholakia 2005)

According to a study conducted by Brookes (2010) images receive22 more engagement than video posts and 54 more than textposts but videos receive 27 more engagement than text posts Thesefigures point out that both images and videos are superior to text-only posts but on the whole images are definitively more compel-ling and persuasive than videos

The aforementioned findings suggest that the richness of the postleads to a more proactive attitude toward the wall post Particular-ly the inclusion of dynamic animations (videos) contrasting colorsand pictures (images) and interactive links to other websites (links)may enhance the salience of a brand post These mechanisms arefound to stimulate different senses that may increase the usersrsquo pro-pensity to look at the content of the message compared with thoseposts with only text Therefore post effectiveness may be condi-tioned by content type

In contrast to previous studies where a priori judgments aboutprogressive levels of richness (low medium and high) are assumed(Fortin amp Dholakia 2005 Pletikosa Cvijikj amp Michahelles 2013) inour approach we only differentiate according to content type (imagesvideos and links) This way we avoid any potential subjective biason how richness is perceived in the usersrsquo eyes Thus we hypoth-esize that

H1a Posts including images are more likely to generate higher levelsof brand post popularity

H1b Posts including videos are more likely to generate higher levelsof brand post popularity

H1c Posts including links are more likely to generate higher levelsof brand post popularity

222 Time frameIn a domain like Facebook where usersrsquo profile walls are con-

stantly overloaded with content coming from multiple sourcesposting time is a relevant aspect that should be taken into accountwhen designing marketing strategies

Concerning the day of publication previous studies reveal thatmost of the user activities on Facebook are undertaken duringworking days (Golder Wilkinson amp Huberman 2007) The BuddyMedia Inc (2011) study also reports that approximately 86 of allbrand postings are done from Monday through Friday and that cus-tomer engagement rates on Thursday and Friday are 18 higher thanon other days of the week Similarly the study of Rutz and Bucklin(2011) on online advertisement supports this thesis stating that theclick-through rate substantially decreases over the weekend There-fore we propose

H2a Posts created on weekdays may cause higher levels of brandpost popularity

It is also suggested that the effectiveness of a post is also influ-enced by the time of the day it is published Identifying the custo-merrsquos habits (peak hours of activity) when determining the postingschedule is crucial

According to Golder et al (2007) usersrsquo interaction increasestowards the evening maintaining a steady high level during the nightRegarding when the post is published Buddy Media Inc (2011) foundthat brands that post early morning and late at night had engage-ment rates approximately 20 higher than the average (and that 60of all brand posts are done during core business hours (from 10amto 4pm) Nevertheless Pletikosa Cvijikj and Michahelles (2013) arguethat if posts are created during those periods with low user activ-ity when fans will connect (in peak hours) the brand post will appearat the top of the wall therefore the probability for being liked orcommented is higher

As observed different temporal patterns are proposedhowever there is no clear agreement Also the sector in whichthe firm operates has its own rhythms and as such should betaken into account Considering this lack of consensus we decidedto look at our data and observed that most of the posts werepublished within business hours Aiming at contrasting whetherposts published in business hours are more effective weformulate

H2b Posts created during business hours may result in higher levelsof brand post popularity

Fig 1 Conceptual model

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

3F Sabate et alEuropean Management Journal (2014) ndash

223 Control variablesPrevious research on advertising effectiveness indicates that

message length may affect performance measures such as click-through rates (Baltas 2003) Similarly the Buddy Media Inc (2011)report shows that posts with 80 characters or less have a 27 higherengagement rate We therefore include message length as a controlvariable expecting a similar negative effect

Several studies (eg Hong Dan amp Davison 2011 Suh Hong Pirolliamp Chi 2010) have pointed out that social features such as the numberof friends followers or similar have an effect on retweetability andcommenting activity Particularly Sun Rosenn Marlow and Lento(2009) observed that diffusion of wall posts in Facebook reaches upto 82 levels signaling that in comparison to real world content Face-book is capable to spread it faster and involve much more peopleLikewise in the study of Zhang et al (2014) on post popularity inone of the most popular microblogging sites in China the authorsfound that the more followers a user has the greater potential au-dience messages posted by this user will have Aiming at control-ling this effect the number of followers has also been included asa control variable

3 Methodology

31 Sample

To test the abovementioned hypotheses we have focused thestudy on Spanish travel agencies with a Facebook fan page The ra-tionale for the scope considered is threefold First according to thereport published by Silverman (2012) the travel agency sector is oneof the 10 industries generating major Internet ad revenues This sectorhas also suffered a strong transformation due to the Internet rev-olution (Buhalis amp Law 2008) and is assumed to be one of the mostactively involved in the use of social media channels (Xiang amp Gretzel2010) Second we decided to focus on Spain for two main reasonsOn the one hand the aforementioned growing importance of socialmedia channels in the tourism industry also applies for Spain (IABSpain Research amp Elogia 2011 Sabate Canabate Velarde-Iturraldeamp Grinon-Barcelo 2010) On the other hand this country is rankedin the top ten ldquoTourism Competitiveness Index 2011rdquo published byBlanke and Chiesa (2012) These two arguments reinforce the in-terest for studying the Spanish case Finally we chose the Face-book platform as it is the largest and most used SNS in Spain (IABSpain Research amp Elogia 2011) Moreover previous literature ex-amining customer engagement in Facebook reinforces the suitabil-ity of this SNS (De Vries et al 2012 Dholakia amp Durham 2010 SmithFischer amp Yongjian 2012)

Firms were selected according to three different criteria As wewere interested in travel agencies actively involved in the use of SNSa first criterion considered the number of social media channels inwhich the firms have presence Specifically we checked their pres-ence on Facebook Twitter YouTube and Blogs as these are the mostrecurrent social networks (Lawrence Pownal Joumlrg amp Carmo 2011)Second and similar to previous studies (Sabate BerbegalConsolacioacuten amp Cantildeabate 2009 Sabate et al 2010) we used the Alexatraffic rank (httpwwwalexacom) to measure the popularity interms of visits to the firmrsquos website Third we included an econom-ic dimension accounting for the revenues obtained during 2008 and2009 looking for firms with an important weight in the travel age-ncyrsquos sector

Five Spanish travel agencies were finally chosen RumboesAtrapalocom eDreamses MuchoViajecom and Barceloviajescom

32 Data collection

The data collection was gathered manually over one month fromMarch 21 to April 21 2011 In order to obtain relevant information

we focused on the evolution of posts published during the previ-ous month that is from February to 21 to March 21 2011 This delaywas necessary in order to capture how users interact with the contentalready published We believe that the time span considered isenough for the purpose of this research SNS are characterized forbeing extremely fast and dynamic communication channels hencea content posted on the net for more than 30 days is not likely toreceive more interaction

For the selected period of time 164 posts published by the fivetravel agencies considered were obtained and manually pro-cessed Content shared by other users on these firmsrsquo fan pages wasnot considered

33 Variables

Fig 2 illustrates the information (variables) gathered for each poston travel agenciesrsquo fan pages Detailed description of these vari-ables is provided in Table 1

Independent variables are those identified with numbers from1 to 6 (Table 1) Following the classification explained in Section 2these variables respond to the hard criterion as they are structuralcharacteristics related to the posts rather than capturing the meaningof the content itself

In order to fit with a linear regression model variables have beencodified as dichotomous or dummy ones Codification was really in-tuitive for links images and videos as they clearly respond to a yesno question when asking for their presence in a post We decidednot to measure them as a number of items after realizing that themaximum number of images and videos per post was 1 and thatfor the specific case of links only three observations include morethan 1 (2 links)

Concerning the variable that captures the time of the post (Time)we decided to differentiate between those posts published duringbusiness hours than those published beforeafter this schedule (seeTable 1) Although we are aware that working hours may vary fromone firm to another we believe that the two segments consideredare representative enough of the traditional Spanish workday Yetconsidering more than two categories would have led to a reducednumber of observations for each segment of time implying a po-tential decrease of the explanatory power of this variable

For the case of the day of publication (DateDay) we use a two-way segmentation differentiating between those posts publishedon weekends than those published on weekdays looking for po-tential differences in terms of interaction rates as a consequence ofthe day of publication

Table 2 shows the number of observations for each variable ac-cording to the categories abovementioned

To proxy for content popularity in Facebook fan pages two de-pendent variables were chosen Likes and Comments Both mea-sures have been widely used as measures of publication impact (DeVries et al 2012 Pletikosa Cvijikj amp Michahelles 2013) On the onehand a post with a high number of likes may indicate that its contentis of interest increasing its probability to be liked by someone andthus disseminating the brand message to a broader number of po-tential customers On the other hand a high number of commentsof a post also represent a kind of success or impact as it implieshaving people spending their time giving their opinions and thoughts

Given this specification two models are therefore tested a firstone explaining the number of likes and a second one aiming at shed-ding some light on those elements that enhance the number of com-ments on brand fan pages

34 Method

The empirical analysis is based on multiple OLS linear regres-sions for each dependent variable using the stepwise method with

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

4 F Sabate et alEuropean Management Journal (2014) ndash

Fig 2 Variables collected from Facebook

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

5F Sabate et alEuropean Management Journal (2014) ndash

the criteria ldquoProbability of F lt= 050rdquo for entering variables into theregression model and ldquoProbability of F gt= 100rdquo for removing themNo missing values were found Outliers those observations withstudentized residual that exceed minus3 or +3 were eliminated In orderto guarantee a normal distribution of the residuals we used naturallogarithms of all the dependent variables these being calculated asLN(Likes+1) and LN(Comments+1) Fig 3 indicates that the errors ob-tained for the different regression models are normally distrib-uted confirming the validity of this approach

Before proceeding with the empirical analysis further clarifica-tion for the Followers variable is required as the number of follow-ers is a characteristic of the travel agencyrsquos Facebook fan page andis collected at the very beginning of the study whereas the rest ofthe independent variables considered are specific features of eachpost In order to improve the explanatory power of this variable wetransformed it using the natural logarithm Therefore the modelstested are designed with LN(Followers) instead of Followers

Table 1Variables definition

Id Variable Explanation Codification and comments

1 Followers Number of users that follow the travel agency Facebook fan page Numerical ge 0Captured at the beginning of the data collection It has beentransformed applying natural logarithm function to better fit a normaldistribution and improve the explanatory power of the model

2 Characters Post length measured by the number of characters Numerical ge 0Characters of the links are also considered

3 Links Number of links within the post Nominal-dichotomic0 ldquono linksrdquo1 ldquo1 or more linksrdquo

4a Images Number of images within the post Nominal-dichotomic0 ldquono imagesrdquo1 ldquo1 or more imagesrdquo

4b Videos Number of videos within the post Nominal-dichotomic0 ldquono videosrdquo1 ldquo1 or more videosrdquo

5 Time Publication time of the post Nominal-dichotomic0 ldquonon-business hoursrdquo (000ndash759 and 1800ndash2359 on Monday toThursday 000ndash759 and 1500ndash2359 on Friday Saturday and Sundayat all hours)1 ldquobusiness hours (800ndash1759 on Monday to Thursday 800ndash1459on Friday)rdquo

6 DateDay Publication date of the post Nominal-dichotomic0 ldquoweekendrdquo (from Friday at 1500 to Sunday at 2359)1 ldquoweekdayrdquo (the remaining time)

7 Likes Number of likes that the post has got Numerical ge 0It has been transformed using natural logarithm to better fit a normaldistribution and improve the explanatory power of the model

8 Comments Number of comments that the post has Numerical ge 0It has been transformed using natural logarithm to better fit a normaldistribution and improve the explanatory power of the model

Table 2Sample size after filtering by categories of dummy variables

Variable Category1 Numberof elementsLikes Modela (CommentsModelb)

Category2 Numberof elementsLikes Modela (CommentsModelb)

Links ldquono linksrdquo 87 (88) ldquo1 or more linksrdquo 75 (76)Images ldquono imagesrdquo 113 (115) ldquo1 or more imagesrdquo 49 (49)Videos ldquono videosrdquo 144 (145) ldquo1 or more videosrdquo 18 (19)Time ldquonon-business hoursrdquo 25 (25) ldquobusiness hoursrdquo 137 (139)DateDay ldquoweekendrdquo 16 (16) ldquoweekdayrdquo 146 (148)

a From a total of 162 observations after having discarded two outliersb From a total of 164 observations Fig 3 QndashQ Plots of Standardized Residual

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

6 F Sabate et alEuropean Management Journal (2014) ndash

4 Results

An overview of the results is presented in Table 3 summariz-ing the main findings of the two models tested Likes model andComments model

41 Likes model

Following the abovementioned procedure we test all the hy-potheses through an OLS linear regression with LN(Followers) Char-acters Images Links Videos Time and DateDay as independentvariables and LN(Likes+1) as the dependent variable for brand postpopularity Two observations were eliminated Table 4 provides thefull description of the coefficients for the significant variables in theLikes model

The explanatory power of the model (R-square) is 553 and theANOVA test calculates a value of 48468 (p-value lt 0001) for F(4157)evidencing a significant and positive linear effect (p-value lt 0001)of certain factors (LN(followers) Characters Images and Videos) overthe number of Likes The formulation of the resulting model is ex-pressed in Equation 1

LN Likes Videos ImagesLN Followers

+( ) = ++ ( )+

1 0 929 0 6730 632

00 003 3 951 Characters minus + ε

(1)

In order to guarantee the statistical correctness of the model wetested the residuals behavior in terms of normality indepen-dence homoscedasticity and multicollinearity assumptions Al-though some authors suggest that the normal distribution of theresiduals is not a requirement of the linear regression model (Greene2003) we tested it to strengthen the robustness of the model Thenormal QndashQ plot of standardized residual (see Fig 3) as well as theKolmogorovndashSmirnov normality test (p-value = 0200) andthe ShapirondashWilk test (p-value = 0054) indicate that we cannot refusethe hypothesis of normality

The independence assumption is also accomplished accordingto DurbinndashWatsonrsquos test which calculates a value of 1665 withinthe interval [15 25] meaning that results appear not to be autocorrelated Fig 4 (see Likes model) also demonstrates that thehomoscedasticity assumption is fulfilled Likewise no collinearityproblems were observed as the maximum VIF index calculated was1370 for the Characters variable (Allison 1999 Belsey Kuh amp Welsch1980)

Additionally other assumptions of the regression model are ac-complished the expected value of the residuals is 0 there is no sig-nificant correlation between the residuals and the independentvariables and there are neither outlier observations nor critical valuessince the standardized residuals interval is [minus2109 2883] and themaximum value of the Cookrsquos distance of the residuals is lower than1 (0111)

All these characteristics corroborate the robustness and appro-priateness of the model tested where independent variables explain553 of the Likes variable

In terms of the results our findings for richness highlight the pos-itive impact of images and videos which contributes to attract usersrsquoattention and are likely to be transformed into likes These resultsvalidate hypotheses H1a and H1b However hypothesis H1c is notsupported as there is no evidence that the inclusion of links in a posthas any effect

Our model fails in establishing a connection between the pop-ularity (in terms of likes) and time frame variables As shown inEquation 1 neither Time nor DateDay variables were entered in the

Table 3Results overview

Models Likes model Comments model

Ra 0743 (high) 0595 (moderate)R square 0553 0355Adj R square 0541 0338Significant variablesb

Images (++) (+)Videos (++)Links (-)DateDayTime (+)Characters (++)LN(Followers) (++) (+)

(+) (ndash) Positive or Negative interrelationship at the level of 005a Codification for r Very low [0 2) Low [2 4) Moderate [4 6) High [6 8)

Very High [8 1]b (++) (ndashndash) Positive or Negative interrelationship at the level of 0001

Table 4Coefficients for significant variables in Likes Model

B Std error T VIF

Videos 0929 0215 431b 1250Images 0673 0145 465b 1205LN(Followers) 0632 0089 711b 1189Characters 0003 0001 509b 1370(Constant) minus3951 0798 minus495b

a p-value lt 005b p-value lt 0001

Fig 4 XY Plots of Predicted value and Residual

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

7F Sabate et alEuropean Management Journal (2014) ndash

model Consequently this prevents us to confirm hypotheses H2aand H2b

Looking at the control variables we observe that as expecteda higher number of followers imply a higher impact on the numberof likes Surprisingly the number of characters is found to exert asignificant positive influence on the dependent variable In Face-book where the length of the message has no restrictions a longertext may suggest a post offering more detailed information

42 Comments model

All the hypotheses are tested through an OLS linear regressionusing the same independent variables than those used in the pre-vious model but having the LN(Comments+1) as the dependent vari-able for popularity Neither missing values nor outliers were found

The resulting model (Equation 2) explains 355 of thevariance and the ANOVA test calculates a value of 21828 (p-value lt 0001) for F(4159) confirming a moderate linear relation-ship between the dependent and some independent variables

LN Comments Images TimeLN Followers

+( ) = ++ ( )

1 0 813 0 6510 293

minusminus minus +0 627 1 320 Links ε

(2)

As shown in Table 5 there is a positive significant effect betweenLN(Comments+1) and Images (p-value lt 005) Time (p-value lt 005)and LN(Followers) (p-value lt 005) variables On the contrary Linkshas negative effect (p-value lt 005)

The statistical correctness of the model has also been tested fol-lowing the same procedure as for the Likes model Here we ob-served that the normal QndashQ plot of standardized residual (see Fig 3)corroborates the normal distribution of the residuals The indepen-dence assumption is also accomplished obtaining a value of 2115within the interval [15 25] in the DurbinndashWatsonrsquos test Al-though Fig 4 for the Comments model is not as clear as in the formerwe believe that the homoscedasticity assumption is validated asCameron and Trivedirsquos IM-test shows (p-value = 0127) Collinear-ity assumption is also fulfilled since the maximum VIF index cal-culated is 1721 for Images variable (Allison 1999 Belsey et al 1980)We also control for the non-existence of outliers and critical valuesobtaining an interval for standardized residuals of [minus2447 2717]and all Cookrsquos distances of the residuals being lower than 1 Addi-tional analyses (the expected value of the residuals is 0 and thereis no significant correlation of the residuals and independent vari-ables) further corroborate the validity of the model despite its mod-erate power of explanation

For richness our results prove that images help increase thenumber of comments a post gets supporting hypothesis H1aHowever there is no evidence that videos influence the number ofcomments meaning that hypothesis H1b is not supported Con-trary to what we expected it is shown that links have a negativeeffect on post popularity in terms of comments therefore hypoth-esis H1c is rejected

Regarding the variables representing the time frame our find-ings indicate that the hour of publication (Time) also plays a key role

Particularly our results corroborate the hypothesis that those postspublished during business hours are more likely to be commentedthan those published outside this schedule Consequently we canassume that H2b is supported Nevertheless the effect of the dayof the week dilutes as this variable does not enter the model speci-fications with a significant coefficient signaling that H2a is notsupported

As for the control variables we find that the variable Followersexhibits a similar behavior as in the Likes model meaning that havinga large amount of followers positively influences the number of com-ments a post may get indicating that more people is expected tohave access to it

An overview of finals results through hypotheses testing is shownin Table 6

5 Discussion and future research avenues

In this paper we have analyzed hard criterion factors that in-fluence the popularity of brand posts published on Facebook andtested them for the Spanish travel agency sector These factors arenot related to the meaning of the content but represent structuralcharacteristics of posts

Following the conceptual model the structural characteristics ofa post have been classified according whether they refer to the viv-idness of its content (richness) or indicate time frame (time and dateof publication)

With respect to the richness our results point towards the im-portance of the use of images which are proven to cause the great-est level of engagement attracting more easily usersrsquo attention andturning this attention into likes and comments This result is in ac-cordance with previous studies suggesting that images are an im-portant element of the posting strategy which significantly increasebrand post popularity

Creative endeavors in the form of videos to enrich the contentof a post only apply when post popularity is measured through thenumber of likes In our interpretation this result could signal thatimages are easier to digest and in a few seconds users can write ashort comment about the feelingsopinions that the picture hasinvoked on them However the process of commenting requires usersto dedicate more time to first assimilate the content and second topublicly assess it by writing an opinion Undoubtedly comment-ing requires an additional effort in comparison with liking (only oneclick is needed)

Results also show that links are negatively influencingthe number of comments When publishing a link Facebook showsa small summary of the content of the destination page This outlinemay be evocative enough to motivate likes (although no relation-ship has been found) whereas to be able to comment users needto visit an external page and consume its content Nevertheless click-ing on the link implies navigating away from Facebook to the des-tination page increasing the risk of users not coming back andcommenting

At this point it worth noting that the choice of avoiding a priorijudgments about progressive levels of richness (low medium andhigh) has been effective Operating as in this paper we have been

Table 5Coefficients for significant variables in Comments Model

B Std error T VIF

Images 0813 0230 353a 1721Time 0651 0227 287a 1033LN(Followers) 0293 0112 262a 1076Links minus0627 0205 minus305a 1625(Constant) minus1320 1032 minus128

a p-value lt 005b p-value lt 0001

Table 6Expected and obtained results by hypotheses

Hypothesis Expected Likes Model Comments Model

H1a (Images) (+) Supported SupportedH1b (Videos) (+) Supported Not supportedH1c (Links) (+) Not supported Not supported (negative effect)H2a (DateDay) (+) Not supported Not supportedH2b (Time) (+) Not supported Supported

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

8 F Sabate et alEuropean Management Journal (2014) ndash

able to individually investigate different behaviors according tocontent type In this respect our findings support our argument thatthe use of different types of richness does not necessarily drive tothe same effect on brand post popularity

Regarding the potential effect that time frame factors have overthe popularity of a post our findings are limited We only found ev-idence to support the hypothesis that relates the time of publica-tion (differentiating between business and non-business hours) andthe number of comments This result suggests that people get noticeof new posts during business hours because they are connected andprobably are in front of a computer This could also be related withthe use of different devices Writing a comment using desktopdevices is undoubtedly easier than doing it from mobile ones It issupposed that outside business hours more users are connectingto Facebook through their mobile terminals which could make dif-ficult to write comments Although data gathered do not include thissort of information if posts are published during business hours theyare more likely to be commented On the contrary liking activityis not influenced by posting time No effect occurred in regard tothe day of the week (weekdays vs weekends) neither for likes norfor comments

Finally the control variables also provide remarkable informa-tion When the content published in the fan page (by the brand oreven by fans) meets the characteristics to become popular it is virallydisseminated through the network of fans fansrsquo friends friends offriends and so on Consequently the larger the number of follow-ers the easier it will be for the company to spread their messageand reach sizeable audiences

Concerning the length of the post our results supportthe convenience of writing larger posts for increasing the numberof likes On the contrary the length of the message is not signifi-cant for the number of comments This finding contrasts with theone reported by Buddy Media Inc (2011) that posts shorter than80 characters have on average 27 higher engagement This dis-agreement could be grounded in very different causes such as meth-odological differences (the report defines engagement as acombination of likes and comments) cultural differences betweenSpanish and north American audiences (those used in theaforementioned report) the language idiosyncrasy or industryspecificities

To conclude our findings reinforce the abovementioned notionthat liking and commenting actions are of different nature and needto be examined separately Not all determinants which are foundto positively enhance one form of brand post popularity (likescomments) also have a positive effect on other forms (commentslikes) Writing a comment is a much more time-consuming processthan liking and is related to different motivations triggeredby the meaning of the content For instance a short question likeldquoWhat are your plans for this weekendrdquo seems to be morelikely to motivate comments than likes People will comment whenthe content is really meaningful for them or request them to actWriting a comment seems more dependent on the emotions andfeelings Also users may be willing to comment if they perceivespecific benefits (eg discounts special offers) or when the contentof the post causes an emotional impact or a feeling that over-whelms the reader (Hettler 2010) Given the increasing impor-tance that SNS are gaining as marketing tools and in the light ofour findings we believe that further research efforts in this direc-tion are necessary

Given the aforementioned considerations a recommendation forfurther studies relates to the effect that soft criteria factors have overlikes and comments as this is one of the main limitations of thepresent work More sophisticated models of brand post populari-ty can be developed by including both criteria (soft and hard) Thesemodels can be enhanced through structural equation modelling(SEM) to benefit from the ability to construct latent variables and

to reflect indirect casual relationships that may arise between factorsAlso the use of text mining and sentiment mining methods such asthose reported by Barbier and Liu (2011) and Aggarwal and Wang(2011) would enable capturing bigger data samples and the incor-poration of new variables in the models tested Notwithstandingwe are aware of the difficulties in obtaining reliable data as thereare some concerns about the consistency of these automatic algo-rithms when capturing and analyzing the meaning of content drawnor recorded in posts and the potential emotions that may ariseamong users

Indeed a model using variables following both the soft and thehard criteria may lead to models with the highest predictive andexplanatory power shedding some light on those factors that helpfirms to engage more efficiently with their customers improvingtheir current communication channels Also investigation compar-ing results in different SNS could reveal interesting facts formarketers

6 Conclusions

This work empirically contributes to a better understanding ofthe use of social media marketing strategies Particularly we haveidentified those structural factors of posts published on Facebookbrand pages that are observed to influence brand post popularitymeasured through the number of likes and comments To do thiswe have focused on a sample of Spanish travel agencies with a Face-book fan page

Results obtained point to some guidelines for improving the likingof posts published on Facebook brand pages Community manag-ers should include images and videos which seem to better attractcustomersrsquo attention especially in the case of images As for thelength of the post moderators should not be worried about writingto many characters if this is essential for a good understanding ofthe content In this sense we found that the number of charactersemployed is positively correlated with the number of likes Finallythe low statistical significance of time frame factors prevents us fromformulating recommendations related with the day and the timethat best helps increase the number of likes

Guidelines for improving the number of comments differ fromthose suggested to increase the number of likes In this case com-munity managers should look for posts that include images becausethis is the only richness factor that is positively related with thenumber of comments Furthermore the avoidance of links wouldreport more comments Links can act as barriers driving users toexternal websites that make them forget returning to the Face-book fan page and leave a comment Another advice that modera-tors should follow consists in publishing during business hours asit seems to improve usersrsquo willingness to comment However thisstrategy should be taken with a grain of salt because as shown inprevious studies there is some controversy in the exact definitionof the most effective time period

Another remarkable conclusion emerging from this study is thatimages are more powerful than videos in increasing consumersrsquo en-gagement The variable Image is positively significant in both models(likes and comments) whereas Videos is only significant when pop-ularity is expressed in terms of likes

It is expected that both the conceptual model and the resultsobtained through this empirical analysis provide meaningfultheoretical and managerial implications for firms and marketersand especially for those professionals working on the travel agencysector Acknowledging the effects that structural characteristics ofposts have on usersrsquo involvement (in terms of likes and com-ments) may help community managers to effectively exploit socialnetworking sites within the integrated marketing communica-tions of the brand

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

9F Sabate et alEuropean Management Journal (2014) ndash

References

aDigital (2011) Uso de Facebook por parte de las empresas espantildeolas AsociacioacutenEspantildeola de la Economiacutea Digital lthttpwwwadigitalorgserviciosuso-de-facebook-por-parte-de-las-empresas-espanolasgt Accessed 052113

Aggarwal C C amp Wang H (2011) Text mining in social networks In C C Aggarwal(Ed) Social network data analytics (pp 353ndash378) Springer US Hawthorned NYlthttpwwwspringerlinkcomcontentn612222780l37460abstractgt Accessed101012

Agresta S amp Bough B B (2011) Perspectives on social media marketing the agencyperspective the brand perspective Course Technology Boston MA

Allison P D (1999) Logistic regression using the SAS system theory and applicationSAS Institute Cary (NC)

Baltas G (2003) Determinants of Internet advertising effectiveness an empiricalstudy International Journal of Market Research 45(4) 505ndash513

Barbier G amp Liu H (2011) Data mining in social media In C C Aggarwal (Ed) Socialnetwork data analytics (pp 327ndash352) Springer US lthttpwwwspringerlinkcomcontentv2h43358124qu862abstractgt Accessed 101012

Belsey D A Kuh E amp Welsch R E (1980) Regression diagnostics identifying influentialdata and sources of collinearity John Wiley New York NY

Blanke J amp Chiesa T (2012) Travel amp tourism competitiveness report 2011Beyond the downturn lthttpwww3weforumorgdocsWEF_TravelTourismCompetitiveness_Report_2011pdfgt Accessed 052113

Brookes E J (2010) The anatomy of a facebook post Study on post performance bytype day of week and time of day Vitrue-Oracle lthttpwwwvitruecomwp-contentthemesVitrue-30white-papersanatomy_of_fb_wppdfgt Accessed090511

Brown J Broderick A J amp Lee N (2007) Word of mouth communication withinonline communities conceptualizing the online social network Journal ofInteractive Marketing 21 2ndash20 doi101002dir20082

Buddy Media Inc (2011) Strategies for effective facebook wall posts a statistical reviewlthttpwwwbuddymediacomnewsroom201104introducing-our-latest-research-strategies-for-effective-facebook-wall-posts-a-statistical-reviewgtAccessed 073011

Bughin J amp Manyika J (2009) How businesses are using Web 20 a McKinsey GlobalSurvey McKinsey Quarterly (09) lthttpwwwmckinseycominsightsbusiness_technologyhow_companies_are_benefiting_from_web_20_mckinsey_global_survey_resultsgt Accessed 061713

Buhalis D amp Law R (2008) Progress in information technology and tourismmanagement 20 years on and 10 years after the internet ndash the state of eTourismresearch Tourism Management 29 609ndash623 doi101016jtourman200801005

Chevalier J A amp Mayzlin D (2006) The effect of word of mouth on sales onlinebook reviews Journal of Marketing Research 43 345ndash354 doi101509jmkr433345

Constantinides E amp Fountain S J (2008) Web 20 conceptual foundations andmarketing issues Journal of Direct Data and Digital Marketing Practice 9 231ndash244doi101057palgravedddmp4350098

Daft R L amp Lengel R H (1986) Organizational information requirements mediarichness and structural design Management Science 32 554ndash571 doi101287mnsc325554

De Vries L Gensler S amp Leeflang P S H (2012) Popularity of brand posts on brandfan pages an investigation of the effects of social media marketing Journal ofInteractive Marketing 26 83ndash91 doi101016jintmar201201003

Dhar V amp Chang E A (2009) Does chatter matter The impact of user-generatedcontent on music sales Journal of Interactive Marketing 23 300ndash307 doi101016jintmar200907004

Dholakia U M Bagozzi R P amp Pearo L K (2004) A social influence model ofconsumer participation in network- and small-group-based virtual communitiesInternational Journal of Research in Marketing 21(3) 241ndash263 doi101016jijresmar200312004

Dholakia U M amp Durham E (2010) One Cafe Chainrsquos Facebook experiment HarvardBusiness Review 88 26

Duan W Gu B amp Whinston A B (2008) The dynamics of online word-of-mouthand product sales ndash an empirical investigation of the movie industry Journal ofRetailing 84 233ndash242 doi101016jjretai200804005

Fischer E amp Reuber A R (2011) Social interaction via new social media (How) caninteractions on Twitter affect effectual thinking and behavior Journal of BusinessVenturing 26 1ndash18 doi101016jjbusvent201009002

Fortin D R amp Dholakia R R (2005) Interactivity and vividness effects on socialpresence and involvement with a web-based advertisement Journal of BusinessResearch 58 387ndash396 doi101016S0148-2963(03)00106-1

Fournier S amp Avery J (2011) The uninvited brand Business Horizons 54 193ndash207doi101016jbushor201101001

Golder S A Wilkinson D M amp Huberman B A (2007) Rhythms of social interactionmessaging within a massive online network In C Steinfield B T Pentland MAckerman amp N Contractor (Eds) Communities and technologies (pp 41ndash66)Springer London lthttplinkspringercomchapter101007978-1-84628-905-7_3gt Accessed 040114

Greene W H (2003) Econometric analysis Prentice Hall Upper Saddle River NJHanna R Rohm A amp Crittenden V L (2011) Wersquore all connected the power of the

social media ecosystem Business Horizons 54 265ndash273 doi101016jbushor201101007

Hansen D L Schneiderman B amp Smith M A (2011) Analyzing social media networkswith NodeXL insights from a connected world M Kaufmann Amsterdam Boston

Hennig-Thurau T Malthouse E C Friege C Gensler S Lobschat L RangaswamyA et al (2010) The impact of new media on customer relationships Journal ofService Research 13(3) 311ndash330 doi1011771094670510375460

Hettler U (2010) Social media marketing marketing mit Blogs sozialen Netzwerkenund weiteren Anwendungen des Web 20 Oldenbourg Muumlnchen

Heymann-Reder D (2011) Social media marketing strategien fuumlr Sie undIhr Unternehmen Addison Wesley in Pearson Education DeutschlandMuumlnchen

Hollenbeck C R amp Kaikati A M (2012) Consumersrsquo use of brands to reflect theiractual and ideal selves on Facebook International Journal of Research in Marketing29(4) 395ndash405 doi101016jijresmar201206002

Hong L Dan O amp Davison B D (2011) Predicting popular messages in Twitter InProceedings of the 20th International Conference Companion on World Wide Web(pp 57ndash58) ACM New York NY USA doi10114519631921963222

IAB Spain Research amp Elogia (2011) III estudio sobre redes sociales en internet IABSpain Research and Elogia lthttpwwwiabspainnetwp-contentpluginsdownload-monitordownloadphpid=73gt Accessed 052113

Kilian T amp Langner S (2010) Online-Kommunikation Kunden zielsicher verfuumlhren undbeeinflussen Gabler Wiesbaden

Kozinets R V de Valck K Wojnicki A C amp Wilner S J S (2010) Networkednarratives understanding word-of-mouth marketing in online communitiesJournal of Marketing 74 71ndash89

Lawrence D Pownal C Joumlrg D amp Carmo C (2011) 2011 Fortune Global 100 socialmedia study The Burson-Marsteller Blog February 15 lthttpwwwburson-marstellercomInnovation_and_insightsblogs_and_podcastsBM_BlogListsPostsPostaspxID=254gt Accessed 052113

Mangold W G amp Faulds D J (2009) Social media the new hybrid element of thepromotion mix Business Horizons 52 357ndash365 doi101016jbushor200903002

Moldovan S Goldenberg J amp Chattopadhyay A (2011) The different roles of productoriginality and usefulness in generating word-of-mouth International Journal ofResearch in Marketing 28(2) 109ndash119 doi101016jijresmar201011003

Pletikosa Cvijikj I amp Michahelles F (2011) A case study of the effects of moderatorposts within a Facebook brand page Social Informatics 6984 161ndash170

Pletikosa Cvijikj I amp Michahelles F (2013) Online engagement factors on Facebookbrand pages Social Network Analysis and Mining 3 843ndash861 doi101007s13278-013-0098-8

Radwanick S Lipsman A amp Aquino C (2011) Itrsquos a social world top 10 need-to-knowsabout social networking and where itrsquos headed comScore lthttpwwwcomscorecomPress_EventsPresentations_Whitepapers2011it_is_a_social_world_top_10_need-to-knows_about_social_networkinggt Accessed052113

Rehmani M amp Khan M I (2011) The impact of E-media on customer purchaseintention International Journal of Advanced Computer Science and Applications 2100ndash103

Rutz O J amp Bucklin R E (2011) From generic to branded a model of spillover inpaid search advertising Journal of Marketing Research 48 87ndash102 doi101509jmkr48187

Sabate F Berbegal J Consolacioacuten C amp Cantildeabate A (2009) SEO strategies inbooksellers sector Intangible Capital 5 321 doi103926ic2009v5n3p321-346

Sabate F Canabate A Velarde-Iturralde M-A amp Grinon-Barcelo R (2010) Use ofinternet promotion strategies by the Spanish travel agencies Profesional De LaInformacion 19 149ndash159 doi103145epi2010mar05

Sashi C M (2012) Customer engagement buyer-seller relationships and social mediaManagement Decision 50(1ndash2) 253ndash272 doi10110800251741211203551

Scott D M (2007) The new rules of marketing and PR how to use news releases blogspodcasting viral marketing amp online media to reach buyers directly John Wiley ampSons Hoboken NJ

Sierra Saacutenchez J (2012) Factors influencing a studentrsquos decision to pursue acommunications degree in Spain Intangible Capital 8 43ndash60 doi103926ic277

Silverman D (2012) IAB internet advertising revenue report 2011 Full Year resultsInteractive Advertising Bureau and PricewaterhouseCoopers lthttpwwwiabnetmediafileIAB_Internet_Advertising_Revenue_Report_FY_2011pdfgt Accessed 052113

Simmons G (2008) Marketing to postmodern consumers introducing the internetchameleon European Journal of Marketing 42 299ndash310 doi10110803090560810852940

Singh V K Jain R amp Kankanhalli M (2011) Mechanism design for incentivizingsocial media contributions In S C H Hoi J Luo S Boll D Xu R Jin amp I King(Eds) Social media modeling and computing (pp 121ndash143) Springer LondonLondon lthttprdspringercomchapter101007978-0-85729-436-4_6gt Accessed101012

Smith A N Fischer E amp Yongjian C (2012) How does brand-related user-generatedcontent differ across YouTube Facebook and Twitter Journal of InteractiveMarketing 26 102ndash113 doi101016jintmar201201002

Smith P R amp Zook Z (2011) Marketing communications integrating offline and onlinewith social media Kogan Page London [ua]

Sterne J (2010) Social media metrics how to measure and optimize your marketinginvestment John Wiley Hoboken NJ

Suh B Hong L Pirolli P amp Chi E H (2010) Want to be retweeted Large scaleanalytics on factors impacting retweet in Twitter network In Social Computing(SocialCom) 2010 IEEE Second International Conference on Social Computing (pp177ndash184) Presented at the 2010 IEEE Second International Conference on SocialComputing (SocialCom) Minneapolis MN IEEE doi101109SocialCom201033

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

10 F Sabate et alEuropean Management Journal (2014) ndash

Sun E Rosenn I Marlow C amp Lento T (2009) Gesundheit Modeling contagionthrough Facebook news feed In Proceedings of the third international AAAIconference on weblogs and social media Presented at the International AAAIConference on Weblogs and Social Media San Jose CA AAAI Press lthttpsnapstanfordeduclasscs224w-readingssun09contagionpdfgt Accessed 101012

Trusov M Bucklin R E amp Pauwels K (2009) Effects of word-of-mouth versustraditional marketing findings from an internet social networking site Journalof Marketing 73 90ndash102

Tuten T L (2008) Advertising 20 social media marketing in a Web 20 world PraegerPublishers Westport CT

Urban G L (2003) Customer advocacy is it for you (Working Paper No 175)lthttpdigitalmiteduresearchpapers175_Urban_Trustpdfgt Accessed 101012

Verhoef P C amp Lemon K N (2013) Successful customer value management keylessons and emerging trends European Management Journal 31(1) 1ndash15doi101016jemj201208001

Xiang Z amp Gretzel U (2010) Role of social media in online travel informationsearch Tourism Management 31 179ndash188 doi101016jtourman200902016

Ye Q Law R amp Gu B (2009) The impact of online user reviews on hotel room salesInternational Journal of Hospitality Management 28 180ndash182 doi101016jijhm200806011

Zhang L Peng T-Q Zhang Y-P Wang X-H amp Zhu J J H (2014) Content or contextwhich matters more in information processing on microblogging sites Computersin Human Behavior 31 242ndash249 doi101016jchb201310031

ARTICLE IN PRESS

Please cite this article in press as Ferran Sabate Jasmina Berbegal-Mirabent Antonio Cantildeabate Philipp R Lebherz Factors influencing popularity of branded content inFacebook fan pages European Management Journal (2014) doi 101016jemj201405001

11F Sabate et alEuropean Management Journal (2014) ndash