Conference PaperPDF Available

Towards a Better Understanding of Online Influence: Differences in Twitter CommunicationBetween Companies and Influencers

Authors:

Abstract

In the last decade, Social Media platforms such as Twitter have gained importance in the various marketing strategies of companies. This work aims to examine the presence of influential content on a textual level, by investigating characteristics of tweets in the context of social impact theory, and its dimension immediacy. To this end, we analysed influential Twitter communication data during Black Friday 2018 with methods from social media analytics such as sentiment analysis and degree centrality. Results show significant differences in communication style between companies and influencers. Companies published longer textual content and created more tweets with a positive sentiment and more first-person pronouns than influencers. These findings shall serve as a basis for a future experimental study to examine the impact of text presence on consumer cognition and the willingness to purchase.
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
1
Towards a Better Understanding of Online Influence:
Differences in Twitter Communication Between
Companies and Influencers
Completed research paper
Diana C. Hernandez-Bocanegra
RTG User-Centred Social Media
University of Duisburg-Essen
Duisburg, Germany
Email: diana.hernandez-bocanegra@uni-due.de
Angela Borchert
RTG User-Centred Social Media
University of Duisburg-Essen
Duisburg, Germany
Email: angela.borchert@uni-due.de
Felix Brünker
RTG User-Centred Social Media
University of Duisburg-Essen
Duisburg, Germany
Email: felix.bruenker@uni-due.de
Gautam Kishore Shahi
RTG User-Centred Social Media
University of Duisburg-Essen
Duisburg, Germany
Email: gautam.shahi@uni-due.de
Björn Ross
School of Informatics
University of Edinburgh
Edinburgh, United Kingdom
Email: b.ross@ed.ac.uk
Abstract
In the last decade, Social Media platforms such as Twitter have gained importance in the various
marketing strategies of companies. This work aims to examine the presence of influential content on a
textual level, by investigating characteristics of tweets in the context of social impact theory, and its
dimension immediacy. To this end, we analysed influential Twitter communication data during Black
Friday 2018 with methods from social media analytics such as sentiment analysis and degree centrality.
Results show significant differences in communication style between companies and influencers.
Companies published longer textual content and created more tweets with a positive sentiment and
more first-person pronouns than influencers. These findings shall serve as a basis for a future
experimental study to examine the impact of text presence on consumer cognition and the willingness
to purchase.
Keywords Online Influence, Social Impact Theory, Twitter, Social Media.
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
2
1 Introduction
Influencer marketing is a marketing strategy often used in recent years on social media. The total
amount spent in 2019 is estimated to be $8 billion (Schomer 2019). In addition to traditional
advertising, influencer marketing is used by companies to increase their organic reach. Influencer
marketing is the process in which companies pay individuals who can reach a large audience to promote
their products in an authentic way (Carpenter Childers et al. 2018). Such individuals are known as
influencers (Woods 2016). They either work independently or are employed by influencer agencies.
Influencer agencies connect influencers and companies not only for regular product advertising but also
for commercial events such as Black Friday
1
.
Many studies have examined the impact of influential users on consumer purchases, for example
through electronic word of mouth (Lin and Wang 2018; Li and Wu 2018), and how to identify influential
accounts on social media as possible opinion leaders. However, this is different from the concept of
professional influencers, which has evolved into a distinct form of advertisement alongside sponsored
posts. Professional influencers are paid by companies to promote a certain product or brand. It is still
unclear how these influencers differ from traditional marketing in the way they communicate on social
media. Since they are especially hired because of their perceived authenticity, more research into their
communication style is necessary. How is their communication characterized and how does it differ
from the communication of companies?
To examine the differences in the communication of influencers and brands, we consider social impact
theory. According to Latané (1981), influence depends on the dimensions of strength, which describes
characteristics of an influence source, immediacy, meaning the closeness of the source, and the number
of sources. Thus, considering the dimensions of social impact theory might allow us to reveal meaningful
insights about online influencers, influencer marketing, brands and traditional social media marketing
practices by brands. Such findings are relevant to better understand their influence on consumer
perception or the willingness to purchase an advertised object. This is especially crucial to know for
social media managers or influencers, who are interested in optimising their social media strategies.
In order to examine the communication of influencers and companies in a structured way, we tackle one
dimension of the social impact theory at a time. In this work, the focus is on the dimension of immediacy.
Immediacy is not only defined as the temporal and geographical proximity of a source (such as an
influencer, a company). It can also be conveyed textually in a social media post (Miller and Brunner
2008). Immediacy in the sense of textual presence can reflect physical or psychological presence.
Proximity can be expressed, for example, by a high amount of words or large text blocks in social media
content (Miller and Brunner 2008). Immediacy as a psychological construct can also be related to the
use of paralanguage (e.g. emojis) or sentimental language (Poole, 2000; Rourke et al. 1999).
Investigating the dimension of immediacy will broaden the understanding of the textual presence of
social media users during a commercial event, by revealing distinct characteristics of content generated
by influencers and companies. For that reason, we consider Twitter content by the most influential
accounts in other words, by those accounts whose tweets were shared the most online and which thus
reached many people. We analyse underlying text features of such content in order to draw conclusions
about differences in author immediacy between influencers and companies. This addresses the following
research question:
RQ: How does textual content generated by influential companies and influencers differ on Twitter
during a commercial event?
As an analysis approach, we have focused on tweets that were published during Black Friday 2018. We
assume that especially commercial events such as Black Friday are an opportunity for companies to
show an online presence and make use of influencer marketing to boost their social impact.
2 Background
2.1 Social Impact Theory
The social impact theory defines influence as “any of the great variety of changes in physiological states
and subjective feelings, motives and emotions, cognitions and beliefs, values and behaviour, that occur
1
https://www.tribegroup.co/blog/black-friday-looks-bright-with-influencer-marketing?
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
3
in an individual, human or animal, as a result of the real, implied, or imagined presence or actions of
other individuals” (Latané 1981, p. 343). Moreover, the social impact theory distinguishes influence in
the three dimensions; strength, immediacy, and number. While strength focuses on the characteristics
of a source of influence, immediacy describes time and space proximity to the target of influence. The
number dimension denotes the quantity of influential sources (Latané 1981).
The presence and actions of companies and influencers on social media are assumed to lead to changes
in attitude or behaviour of their recipients, that is, social media users (Woods 2016). Regarding the
dimension of strength, assertiveness and exaggeration are identified as characteristics predictive of
influential users in the online context (Miller and Brunner 2008). On the other hand, Miller and Brunner
(2008) understood immediacy as the presence of published content. The presence of textual
contributions can be characterised by their total number and the number of words contained in each
contribution. The former was found to be a predictor for influence in anonymous, synchronous and
collaborative online communication. We assume that such immediacy features, as well as other
dimensions of the social impact theory, have an effect in a commercial context, too. Therefore, we focus
on immediacy in this work in progress (RQ) by taking a closer look at features connected to text
presence, for example, length of the contribution, and the use of frequent words or elements such as
emojis.
2.2 Influence on Social Media
Since social media have become a stage for everyday life, individuals who exert a strong influence on
others have attracted the attention of companies for marketing purposes as well as the attention of
researchers. One term, which long predates social media and goes back to the idea of opinion leaders
from theories on public opinion formation such as Lazarsfeld’s two-step flow model, is that of
“influentials” (Watts and Dodds 2007). Similarly, research has termed individualsinfluencerswho
have a high connectivity to others or who are in a central position in the (social) network that allows
them to catalyse a cascade of influence (Bakshy 2011). Kempe et al. (2003) still envisaged these
influential” users as promoting a product for free by recommending it to their friends, after being given
a free sample and liking it.
However, the contemporary concept of an influencer, as in influencer marketing, is slightly different.
The influencers advertising products on Twitter or Instagram are not simply telling their peers that they
enjoyed a particular product, but they are paid to do so, yet they still strive to appear relatable and
authentic (Newlands and Lutz 2017). Their terms are negotiated with the advertising agency and
specified in a written contract. The amount of money influencers make can be high as much as
$20,000 from a single contract in some instances (Carpenter Childers et al. 2018). As a result,
“influencer” is no longer only a label applied by academics to accounts with a high follower count, but it
is considered a distinct profession. There are social media users who aspire to become professional
influencers, and books (Graham, 2019; Welch, 2019) and events (Stoldt et al. 2019) that promise to show
them how to do so. We focus on the influential communication of these individuals who have a high
connectivity to other users and who are paid by companies to promote a product, who can also be
described as “professional influencers”.
Hashtags, mentions, URLs, emojis and the number of words are among the elements that are often used
to examine online communication, and that have an effect on the influence of content (Cossu et al. 2015).
Miller and Brunner (2008) found that a higher number of words within a text contribution corresponds
to social influence in anonymous, collaborative online communication. Findings show that URLs whose
webpages evoke positive feelings or which are assessed as interesting are more likely to spread, although
a prediction of influence concerning URLs has been found to be unreliable (Bakshy et al. 2011). Hashtags
symbolise social influence expressed by the so-called neighbourhood effect. It describes that if
individuals within a network show the adoption of a trend by using content-related hashtags, others
tend to be influenced in their behaviour and also join the community (Backstrom et al. 2006).
Mentioning other users (mentions) has been considered to have an impact on sentimental influence in
social media (Wu and Ren 2011). Moreover, findings regarding emojis have shown that they promote
perceived playfulness in text messages, which positively affects electronic word-of-mouth and is highly
related to online influence (Hsieh and Tseng 2017).
Another approach for examining textual contributions on social network sites is analysing
communication styles. Linked to personality traits, communication styles may present the author’s
stable individual predispositions during communication (Page et al. 2013). Research results on the
relationship between personality and influence have been somewhat conflicting (Winter et al. 2020).
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
4
However, the use of specific words such as pronouns (e.g. I, me, we, they) has been related to personality,
social skills, leadership ability and the quality of relationships between people (Pennebaker 2011).
Quercia et al. (2011) considered categories that reflect language use for examining influential tweets.
High use of first-person pronouns in communication correlates with the personality trait of neuroticism
(Stirman and Pennebaker 2001) and is related to self-focus (Quercia et al. 2011). In comparison, second
and third-person pronouns symbolise social engagement (Rude et al. 2004). Moreover, Quercia et al.
(2011) conducted a sentiment analysis to test the relationship between expressed sentiment on the level
of communication styles and influence on Twitter. They found that influential users tend to express a
sense of community and negative emotions by often using second- and third-person pronouns as well
as words with a negative connotation.
3 Research Design
3.1 Case Description
The “Black Friday” is a commercial event on the day after Thanksgiving. Over the years, Black Friday
has evolved into one of the most popular shopping events in Western culture. Social media can be
deployed to distribute marketing communication to further promote this event, which can be enhanced
by influencer marketing. Since these marketing strategies affect purchase behaviour and the willingness
to purchase, we consider data regarding the Black Friday as relevant to gain new insights in the
commercial context.
3.2 Data Collection
This case study examines influential content during a commercial event. To this end, we have performed
a sequence of steps. First, the study builds upon Twitter communication related to the commercial
phenomenon of “Black Friday” (see also Brünker et al. 2020). The data was gathered with a self-
developed Java crawler using the Twitter4J library. The collected tweets contained at least one of the
following hashtags: “BlackFriday”, “CyberMonday”, “BlackFridayDeals”, “BlackFriday18”,
“BlackFridaySale”, “BFCM”, “BFCM18”, “BlackFriday2018”. These terms were chosen based on
Twitter’s trending hashtags and comparison to past Black Friday related Twitter communication. Due
to the surrounding event “Cyber Monday Week”, the hashtags “CyberMonday2018” and
“CyberMonday18” were also considered as relevant indicators for commercial communication during
the period. Furthermore, only tweets declared by Twitter as English were considered. In total, we
gathered Twitter communication on the Black Friday, from 23 November 2018 00:00 to 23 November
2018 23:59. The selection of keywords was based on the usage frequency during the event, as well as
hashtags appearing in the Trends section on Twitter. The tracking yielded a total of 392,606 users
creating 542,551 tweets.
In order to identify influential users and their content, we used social network analysis to compute the
degree centrality of each user, particularly the in-degree. We defined influential users to be those users
who were retweeted the most within the dataset (Oh et al. 2015). In order to cover the majority of shared
tweets, we considered the 200 most retweeted users during the examined time period, with a resulting
number of 808 tweets for further analysis. These highly retweeted users are responsible for the majority
of online communication in the examined case one user may be retweeted thousands of times and is
therefore represented several times in the dataset (Stieglitz et al. 2017). The reason we considered the
top 200 retweeted users is the distribution of retweets in the dataset. A few users were responsible for
the majority of retweets; thus, to catch the majority of influential users, we focused on the most
retweeted ones and excluded the cases belonging to the long tail from the analysis.
To facilitate the analysis of data, we first manually classified users into companies, influencers and other
specific roles. Similar suitable categorisations such as influential individual, media or promotional
account have been used before to precisely analyse influencers (Bokunewicz & Shulman 2017).
Therefore, we followed an inductive category formation (Mayring 2014), and manually checked each of
the Twitter profiles. This check involved a number of aspects of the profile, for example the content of
profile description, URLs linked as personal websites or connected social media platforms (e.g.
Instagram, YouTube, or blogs) as well as a sample of the last tweets shared. We started with one category
and benchmarked each account against the criteria of the category. Following that, we either classified
the account into the existing category or created a new one. For example, we assigned users to the
category of influencers who have a high connectivity to other users and are likely to be paid by companies
to promote a product. Figure 1 shows two examples. This step involved two independent researchers
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
5
who came to a substantial agreement of Cohen’s kappa = 0.724, signalling a sufficient level of inter-
coder reliability (Landis and Koch, 1977).
Figure 1. Examples of Twitter Influencers
Tweets were preprocessed by removing punctuation, stop words and the hashtags related to Black
Friday. In order to identify the differences in communication style between the roles of interest
(influencers and companies), we performed a content analysis consisting of the following tasks:
determining metrics related to common tweet elements (hashtags, URLs, emojis and mentions),
detecting the most frequent words in tweets, and examining communication style (Quercia et al. 2011),
especially first-, second- and third-person pronouns. Metrics based on the number of elements such as
hashtags, URLs, emojis, and mentions were extracted using regular expressions, character patterns that
represent those elements. We use the Python library wordcloud (Mueller 2019) to identify relevant
words based on their weighted frequency in the dataset. We also analysed the URLs in the tweets to
understand what kind of websites are referenced by companies and influencers with the library
pydomains (Sood 2018), which identifies the type of website from a set of predefined categories, such as
shopping (e.g. amazon.co.uk), bank, phishing, malware, press, adult, and others.
To further identify communication styles, we extracted the number of pronouns in the first person (e.g.
I, we, me, my, myself), the second person (e.g. you, yourself), and the third person (e.g. he, they, his,
her). Since pronouns are sometimes omitted in informal language (for example, “just got home” instead
of “I just got home”), two coders identified instances of this phenomenon (called ellipsis) in the data set
and discussed cases of initial disagreement until they agreed. The reliability of this approach was
calculated as κ = 0.66 (first person pronouns), κ = 1.00 (second person), κ = 0.67 (third person) after
two coders used the same procedure on a smaller subset (100 tweets). This allows us to calculate the
number of sentences that are in the first, second or third person even when the pronoun is absent, and
it gives us a measure of how informal the linguistic register used in the tweets is.
In order to compare the content differences between influencers and companies, we performed
statistical tests on distributions of length, the number of words, URLs, hashtags, mentions, emojis, and
the number of first, second- and third-person pronouns per tweet.
Lastly, we performed a sentiment analysis (cf. Quercia et al. 2011) to classify the sentiment expressed in
a text as positive, neutral or negative. The objective is to identify the attitudes and emotions that
characterise the content of each tweet. For that reason, we used the tool VADER (Hutto and Gilbert
2014).
4 Results
As a first step, we categorised the 200 most retweeted Twitter accounts (authors) into different roles, in
accordance with the above-mentioned procedure. After the removal of tweets generated by accounts that
were deleted or suspended and, thus, could not be classified at the time of analysis, the total number of
tweets that we used for the analysis is 763, by 187 authors. Table 1 shows the number of tweets and
authors for each role.
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
6
Role
Examples
# of
tweets
# of
authors
Companies
Cafes and restaurants, airlines, glasses vendors, shops
264
66
Influencers
Professional bloggers, Twitterers, YouTubers,
Instagrammers
54
20
Others
Artists, government and politicians, media, individuals,
social activists, communities, sports teams
445
101
Total
763
187
Table 1. Categorisation and number of tweets and authors per role
We aimed to test differences between the two roles regarding length of tweets, and number of elements
included within tweets (emojis, hashtags, URLs, and mentions), as well as the use of first, second and
third person pronouns. In this respect, and given that data did not fulfill the assumption of normality,
we used a nonparametric test for comparison, i.e. Mann-Whitney U test. In addition, chi-square tests
were used to compare percentages of tweets with positive, negative and neutral opinions, within each
role, as well as the percentages of use of different personal pronouns and the percentages of URL types
referenced in tweets within each role. Results indicate that companies tend to write longer tweets, with
more hashtags, and emojis than influencers. In fact, tests indicate that:
The length of tweets (in characters) was significantly greater for companies (Mdn=154) than
for influencers (Mdn=117), U=5018, p<.001, d=0.50, as well as the number of words per
tweet, which was significantly greater for companies (Mdn=24) than for influencers
(Mdn=17), U=4857, p<.001, d=0.52. These d values indicate a medium effect size (Cohen
1988). See also Figure 2.1.
The number of other elements per tweet was also significantly larger for companies than for
influencers. This is the case for the number of URLs per tweet (companies: Mdn=1, 75th
percentile=2; influencers: Mdn=1, 75p=1), U=5558, p=.005, d=0.01, hashtags
(companies: Mdn=1, influencers: Mdn=0), U=5676, p=.012, d=0.28 and emojis
(companies: Mdn=0, 75th percentile=1; influencers: Mdn=0, 75p=0), U=5771, p=.006,
d=0.37. Despite the significant difference in the case of URLs, the very small d value indicates
a small effect size, therefore the difference can be considered trivial. The d values for emojis and
hashtags indicate a small to medium-sized effect. See also Figure 2.2.
The percentages of tweets with positive, neutral or negative opinions differ significantly
between companies and influencers, X2 (2, N = 318) = 9.54, p <.01. See also Figure 2.4.
The analysis of frequent words in the dataset reveals that both companies and influencers frequently
communicate about special “deals”, “giveaways”, and “competitions” where individuals might “win”
certain prizes. However, influencers also try to convey messages involving emotions and feelings, as with
the frequently used expression “Friday feeling”, or messages with a critical intent, such as those related
to politics, as evidenced by frequent mentions of “Trump”.
The number of mentions per tweet was not significantly greater for companies than for influencers, nor
the number of second and third-person pronouns (Mdn=0, both companies and influencers). However,
the results indicate that companies use more first-person pronouns than influencers. The number of
first-person pronouns per tweet was significantly greater for companies than for influencers
(companies: Mdn=0, 75th percentile=1; influencers: Mdn=0, 75p=0), U=5386, p=.002, d=0.23
(indicating a small to medium-sized effect). Additionally, in the case of companies, the use of first-
person pronouns surpasses the use of second and third person pronouns, as depicted in Figure 2.3. This
difference was also confirmed using a chi-square test (X2 (2, N = 318) = 17.69, p < .001), which indicates
that the distribution of use of pronouns by companies is significantly different from a distribution that
assumes that all pronoun types are equally likely to be used. In the case of influencers, this difference is
not significant. Moreover, the number of ellipsis (omissions of, for example, pronouns) was higher for
companies than for influencers (companies: Mdn=0, 75th percentile=1; influencers: Mdn=0,
75p=0), U= 5581, p=.004, d=0.46 (a small to medium-sized effect). We found a positive correlation
between the number of ellipsis used and the length of the tweet, r=0.15, p<0.05. However, this
correlation can be considered as very weak (Evans 1996).
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
7
Figure 2. From left to right, top to bottom: (1) Length of tweets, (2) Elements in tweets, (3)
Communicative style (Percentage of each personal pronoun used in tweets), (4) Sentiment analysis
(Percentage of tweets with each type of polarity), (5) Types of URLs referenced in tweets.
The sentiment analysis results show that there is a difference of the proportion of tweets that contain
positive, neutral or negative words, between companies and influencers (X2 (2, N = 318) = 9.54, p <.01),
as depicted in Figure 2.4. In particular, we found that in the case of companies, the proportion of tweets’
polarity is significantly different from a distribution that assumes that the content is tweeted in the same
polarity proportion, i.e. 33% of tweets would have each polarity (X2 (2, N = 318) = 73.17, p <.001). In the
case of influencers, such difference is not significant.
Lastly, the URL analysis showed that there is a difference in the proportion of URL types referenced in
tweets between companies and influencers (X2 (4, N = 409) = 33.31, p <.001), as depicted in Figure 2.5.
In particular, for influencers, only 3 out of 54 (5.6%) of tweets contained direct links to shopping
websites, compared with 51 out of 264 (19.3%) for companies. This difference is statistically significant
(X2 (1, N= 318) = 6.02, p = .014)). In contrast, of the tweets by influencers, 31.5% contained a link to a
press website, compared with 7.2% of tweets by companies (X2 (1, N = 318) = 26.33, p< .001). In
addition, all 21 links to audio or video hosting websites were posted by company accounts.
5 Discussion
The present study exploratively approached the utilisation of Twitter by companies and influencers as
well as their communication styles, in the context of a large-scale commercial event. Although the main
objectives of influencers and companies are identical, that is, to advertise special deals and competitions,
their strategies differ markedly. Examining the findings under social impact theory and its dimension of
immediacy, we can deduce that there are indeed differences between companies and influencers
regarding characteristics of the text, such as its length or sentiment (we observed that influencers tend
to communicate in a more negative way than companies do). The results show that companies use
significantly more characters in their tweets, and use emojis and hashtags much more frequently than
influencers. In consequence, one can argue that companies are more interested in arousing positive
emotions and thus promoting themselves and their products. Therefore, their strategies might be
perceived as less authentic (they only highlight positive aspects of their products or services, with no
apparent capacity for self-criticism), compared to influencers, who might be perceived as more authentic
and credible, since they are more critical in their discussion about products and deals. In particular,
most of the negative statements expressed by influencers in the data set involves criticism to
unnecessary purchases, e.g. The trouble with #BlackFriday is you always buy something you don't need
just because it's cheap(sic.). This perceived authenticity has been identified as one of the key
characteristics of influencer marketing (Newlands and Lutz 2017). Finally, although the difference in
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
8
the number of URLs between companies and influencers is negligible, companies are significantly more
likely to include direct links to shopping websites and to audio or video hosting websites in their tweets,
whereas influencers are more likely to refer to press websites. The latter may be perceived as more
neutral and as a more indirect form of promoting a product, further contributing to perceived
authenticity.
Considering that another key feature of influencer marketing is that they suggest a personal, almost
intimate relationship with their audience (Newlands and Lutz 2017), it is surprising that influencers
were less likely to use first-person pronouns in their posts than companies. One possible explanation is
that businesses deliberately use many such pronouns in an effort to seem more personable.
Our contributions are novel. Similar research on characteristics of influential tweets did not study them
with a focus on the new phenomenon of influencer marketing. For example, focusing on online influence
based on a textual level, Miller and Brunner (2008) found that the length of messages positively
correlates with social influence. However, their research was conducted on anonymous, collaborative
social networks. Context and nature of communication are thus different from those in public social
media communication. Therefore, the relationship between online influence and textual characteristics
needs to be checked for this domain, as well.
Moreover, Quercia et al. (2011) considered communication styles within influential tweets in a similar
manner as we did. However, their approach can rather be seen as connecting the dimensions of
immediacy and strength by relating communication styles to specific user types. Instead, we are not
addressing such a connection, nor identifying the actual personality of influencers and companies based
on their used communication style, but aiming to gain a better understanding of the differences between
traditional social media marketing and influencer marketing.
In order to address the effect that communication styles of companies and influencers have on users’
perception, we need to look beyond the dimension of immediacy and instead relate to the dimension of
strength from social impact theory. It is also conceivable that the impact of used communication styles
will multiply the more often influencers or companies publish content to advertise a product. This
assumption considers the relation that an interplay of immediacy and number may have to online
influence. In summary, a more comprehensive analysis of online influence can be accomplished by
considering the effects that the social impact dimensions have on each other.
6 Conclusion and Further Steps
As a conclusion, the results suggest various communication strategies as a possible explanation for the
identified differences between the two roles. Consequently, in order to analyse the potential impact of
such differences on individuals, we suggest examining the perception of content that fits the
communication patterns of companies and influential people first.
We plan to expand this research in progress, based on the findings presented in this study. We provide
first insights into differences in the communication of influencers and companies. These findings will
be used as a foundation for further investigating differences in influential content on Twitter during a
commercial event. As future work, we plan to conduct a user experiment, design the test conditions and
extract tweets to be compared based on the identified differences between companies and influencers.
In this way, we plan to examine how influential content of influencers and companies on Twitter affects
the willingness to purchase a product, in order to gain a better understanding about the impact of the
dimension of immediacy from social impact theory, in the context of commercial social media
communication. Further research in this sense might show the effectiveness of distinct communication
strategies during a large-scale commercial event on Twitter.
Based on our preliminary findings, we also plan to examine the dimension of strength from social impact
theory by investigating the characteristics of companies and influencers and how they are perceived by
their recipients. Moreover, the dimension of number shall be addressed by considering how the amount
of content published by influencers and companies affects the dimensions of strength and immediacy,
as well as the willingness to buy an advertised product. In addition, we also plan to extend our analysis
to non-textual content. In doing so, we aim to gain a broader view of influence on Twitter during a
commercial event under the scope of the social impact theory. Furthermore, it should be noted that our
preliminary analysis is limited to the top 200 users using the English language. Future examination may
consider original content by more users, as well as cultural differences in the way influential tweets are
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
9
in other languages. However, the purpose of this study was to analyse the content of the most influential
accounts.
References
Backstrom, L., Huttenlocher, D., Kleinberg, J., and Lan X. 2006. “Group Formation in Large Social
Networks: Membership, Growth, and Evolution,” in Proceedings of the 12th ACM SIGKDD
International Conference on Knowledge Discovery and Data Mining, Philadelphia, PA, pp. 44-54.
Bakshy, E., Hofman, J. M., Mason, W. A., and Watts, D.J. 2011. “Everyone’s an Influencer: Quantifying
Influence on Twitter,” in Fourth ACM International Conference on Web Search and Data Mining,
Hongkong, China, pp. 65-74.
Bokunewicz, J.F., and Shulman, J. 2017. Influencer Identification in Twitter Networks of Destination
Marketing Organizations,” Journal of Hospitality and Tourism Technology (8:2), pp. 205-219.
Brünker, F., Marx, J., Ross, B., Mirbabaie, M., and Stieglitz, S. “‘The Tireless Selling Machine’
Commercial Deployment of Social Bots during Black Friday Season on Twitter.” In Entwicklungen,
Chancen und Herausforderungen der Digitalisierung, Gronau, N., Heine, M., Krasnova, H., and
Pousttchi, K. (eds.): Band 1: Proceedings der 15. Internationalen Tagung Wirtschaftsinformatik
2020, Potsdam, Germany, March 2020.
Carpenter Childers, C., Lemon, L. C., and Hoy, M. G. 2018. “#Sponsored #Ad: Agency Perspective on
Influencer Marketing Campaigns.” Journal of Current Issues & Research in Advertising (40:3), pp.
258-274.
Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences (2nd ed.), Hillsdale, NJ:
Lawrence Erlbaum Associates.
Cossu, J. V., Dungué, N., and Labatut, V. 2015. “Detecting Real-World Influence Through Twitter.” In
Second European Network Intelligence Conference, Karlskrona, Sweden, pp. 83-90.
Evans, J.V. 1996. Straightforward Statistics for the Behavioral Sciences, Pacific Grove, Calif:
Brooks/Cole Publishing.
Hsieh, S. H., and Tseng T.H. 2017. “Playfulness in Mobile Instant Messaging: Examining the Influence
of Emoticons and Text Messaging on Social Interaction.” Computers in Human Behavior (69), pp.
405-414.
Gilbert, C. H. E., & Hutto, E. (2014, June). Vader: A parsimonious rule-based model for sentiment
analysis of social media text.” In Eighth International Conference on Weblogs and Social Media
(ICWSM-14).
Graham, M. 2019. Social Media Marketing 2019: 30 Powerful Strategies to Become an Influencer for
Billions of People on Facebook, Instagram, YouTube, LinkedIn and Others. Independently published.
Latané, B. 1981. “The Psychology of Social Impact,” American Psychologist (36:4), 343.
Li, X., and Wu, L. 2018. “Herding and Social Media Word-of-Mouth: Evidence from Groupon,” MIS
Quarterly (42:4), pp. 1331-1351.
Lin, Z., and Wang, Q. 2018. “E-Commerce Product Networks, Word-of-Mouth Convergence, and
Product Sales,” Journal of the Association for Information Systems (19:1), pp. 23-39.
Mayring, P. 2014. Qualitative Content Analysis: Theoretical Foundation, Basic Procedures and
Software Solution, Klagenfurt, Germany: SSOAR.
Miller, M. D., and Brunner, C. C. 2008. “Social Impact in Technologically-Mediated Communication:
An Examination of Online Influence,” Computers in Human Behavior (24:6), pp. 2972-2991.
Mueller, A. 2018. “WordCloud 1.5.0.” doi: 10.5281/zenodo.594952.
Newlands, G., and Lutz, C. 2017. “So Close, yet son far Away: The Paradox of Status and Distinction
Among Instagram Influencers,” AoIR Selected Papers of Internet Research.
Sood, G. 2018. “PyDomains: Classifying the Content of Domains”.
https://github.com/themains/pydomains
Australasian Conference on Information Systems Hernandez-Bocanegra et al.
2020, Wellington Towards a Better Understanding of Online Influence
10
Oh, O., Eom, C., and Rao, H. R. 2015. “Research Note Role of Social Media in Social Change: An
Analysis of Collective Sense Making During the 2011 Egypt Revolution,” Information Systems
Research (26:1), pp. 210-223.
Page, X., Knijnenburg, B. P., and Kobsa, A. 2013. „FYI: Communication Style Preferences Underlie
Differences in Location-Sharing Adoption and Usage,” in Proceedings of the 2013 ACM International
Joint Conference on Pervasive and Ubiquitous Computing, Zurich, Switzerland, pp. 253-162.
Pennebaker, J. W. 2011. “The Secret Life of Pronouns,New Scientist, (211:2828), pp. 42-45.
Poole, D. M. 2000. “Student Participation in a Discussion-Oriented Online Course: A Case Study,”
Journal of research on computing in education (33:2), pp. 162-177.
Quercia, D., Ellis, J., Capra, L., and Crowcroft, J. 2011. “In the Mood for Being Influential on Twitter,”
in IEEE Third International Conference on Privacy, Security and Trust and Social Computing,
Boston, MA, pp. 207-314.
Rourke, L., Anderson, T., Garrison, D. R., and Archer, W. 1999. “Assessing Social Presence in
Asynchronous Text-Based Computer Conferencing,” The Journal of Distance Education/Revue de
l’Éducation à Distance (14:2), pp. 50-71.
Rude, S., Gortner, E. M., and Pennebaker, J. W. 2004. “Language Use of Depressed and Depression
Vulnerable College Students,” Cognitions & Emotion (18:8), pp. 1121-1133.
Schomer, A. 2019. “Influencer Marketing: State of the social media influencer market in 2020,”
Business Insider (Available at https://www.businessinsider.com/influencer-marketing-report).
Stieglitz, S., Mirbabaie, M., Schwenner, L., Marx, J., Lehr, J., and Brünker, F. 2017. “Sensemaking and
Communication Roles in Social Media Crisis Communication.” in Proceedings of the 13th International
Conference on Wirtschaftsinformatik (WI), St. Gallen, Switzerland.
Stirman, S. W., and Pennebaker, J. W. 2001. “Word Use in the Poetry of Suicidal and Nonsuicidal
Poets,” Psychosomatic Medicine (63:4), pp. 517-522.
Stoldt, R., Wellman, M., Ekdale, B., & Tully, M. (2019). Professionalizing and Profiting: The Rise of
Intermediaries in the Social Media Influencer Industry,” Social Media + Society (5:1),
2056305119832587.
Watts, D.J., and Dodds, P.S. 2007. “Influentials, Networks, and Public Opinion Formation,” Journal of
Consumer Research (34:4), pp. 441-458.
Welch, D. 2019. Social Media Marketing 2019: Secret Strategies to Become an Influencer of Millions
on Instagram, YouTube, Twitter, and Facebook While Advertising Your Personal Brand in 2018 and
Beyond. Independently published.
Winter, S., Neubaum, G., Stieglitz, S., and Ross, B. 2020. “#Opinionleaders: a comparison of self-
reported and observable influence of Twitter users,” Information, Communication & Society.
Woods, S. 2016. #Sponsored: The Emergence of Influencer Marketing. University of Tennessee.
Wu, Y., and Ren, F. 2011. “Learning Sentiment Influence in Twitter,” in International Conference on
Future Computer Science and Application, Hongkong, China, pp. 119-122.
Acknowledgements
This work was supported by the Deutsche Forschungsgemeinschaft (DFG) under grant No. GRK 2167,
Research Training GroupUser-Centred Social Media”.
Copyright
Copyright © 2020 authors. This is an open-access article licensed under a Creative Commons
Attribution-NonCommercial 3.0 New Zealand, which permits non-commercial use, distribution, and
reproduction in any medium, provided the original author and ACIS are credited.
ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
In recent years, mainstream E-commerce platforms have changed the way consumer goods are merchandized online. While such platforms increasingly discover social elements to drive sales, social media start implementing E-commerce features themselves. This emerging intersection also provides space for bot activity in a Social Commerce (SC) context, which is sparsely understood. In this short paper, we investigate the deployment of social bots during a large-scale commercial event. To this end, we applied bot detection techniques to the Twitter communication during the 2018 Black Friday season. Using three distinct metrics, we identified 42 bot-like accounts. A manual classification of 11,889 tweets found those bots to be primarily deployed in pre-transactional phases of SC to promote third party products and to initiate external transactions. Our results further suggest adapting detection metrics to event-specific user behavior. Further research aims at the dynamics, impact, and network positions of SC bots.
Article
Full-text available
This study examines the relationship between travel influencers (e.g., bloggers and social media personalities) and destination marketers within the changing travel and tourism industry. Through in-depth interviews, observations, and document analysis, we explore the tensions between travel influencers and destination marketers that shape the way travel is promoted, labor is compensated, and professional structures are negotiated. We examine a new breed of travel and tourism worker—intermediaries who seek to professionalize and formalize the relationship between influencers and destination marketers while simultaneously solidifying their own role within the industry. Intermediaries promote and facilitate relationships based on structured flexibility—formalized agreements designed to satisfy a brand’s campaign goals yet open enough for influencers to pursue their unique needs. By examining the relationships between digital content creators, destination marketers, and third-party intermediaries, this article provides insight into how digital media industries negotiate the tension between participation and control.
Article
Full-text available
Purpose Destination marketing organizations (DMOs) use Twitter to promote attractions and special events and to build brand awareness. Tweets of a DMO spread through a complex network of connected accounts. Some of these are more influential than others due to their position within the network. This paper aims to use a network analysis of 14 DMOs to identify the categories of influencers that have the greatest reach. Design/methodology/approach NodeXL was used to download and analyze network data from Twitter during July 2016 for a collection of DMOs promoting US cities. Accounts in the networks were ranked using several measures of relative influence such as the number of times the accounts mentioned/retweeted others or were mentioned in posts about the DMO. The most influential accounts in the network were identified and coded by category. Findings Media, promotional accounts and those of individuals were determined to be influential by each metric considered. Stakeholders such as hotels and restaurants occupy positions of low importance in the networks and generally do not capitalize on opportunities provided by the DMOs. Practical implications DMOs can seek out strategic partnerships with key influencers to maximize their effectiveness. Additionally, stakeholders can improve their Twitter presence by interacting with the DMOs and other influential accounts. Originality/value This paper identifies influencers that can aid in DMOs’ marketing campaigns. It also presents a methodology that can monitor the effectiveness of such campaigns, something absent in the current literature.
Conference Paper
Full-text available
Social media is becoming increasingly important during crisis situations. Affected people are now enabled to provide helpful on-site information, and emergency service organisations can use social media to inform people and communicate with them. This study addresses how different communication roles in social media affect sensemaking during crises situations. To this end, we conducted a study on Twitter during the Brussels attacks of 2016. We collected a sample of 3,223,197 tweets, which included a total of 1,535,943 participants. Our study reveals that, whereas information distribution dominates early crisis stages, attention-keeping gains in importance in subsequent stages. It is decisively depending on the characteristics and retweet behaviour of certain communication roles that information is being consulted by individuals in a situation of either lack of information or information overload.
Article
Social media have become forums of discussions on political and societal debates in which individual users may forward information or influence others. While prior studies either employed network analyses or surveys to identify opinion leaders and their characteristics, the present investigation combines these two approaches to address the relationship between observable and self-perceived influence. For this purpose, a retweet network of Twitter communication on the Brexit debate (N = 15,018) was analyzed in relation to a survey on motives and personality traits that was filled out by a subsample of active users (N = 98). Results showed that users’ eigenvector centrality (as a measure of influence in the network) was significantly related to their political interest and their number of followers, but not to self-perceived opinion leadership. According to a comparison of self-assessment and network position, those with stronger motivations to distribute relevant information tended to overestimate their influence in the network. Implications for the identification of opinion leaders are discussed.
Article
Modern online retailing practices provide consumers with new types of real-time information that can potentially increase demand. In particular, showing sales information to a customer can increase certainty about product quality, inducing consumers to herd. This effect can be particularly salient for experience goods due to their quality being inherently highly uncertain. Social media word-of-mouth (WOM) can increase product awareness as product information spreads via social media, increasing demand directly while amplifying existing quality signals such as past sales. This study examines the mechanisms behind the strategy of facilitating herding and the strategy of integrating social media platforms to understand the potential complementarities between the two strategies. We conduct empirical analysis using data from Groupon.com, which sells goods in a fast cycle format of "daily deals." We find that facilitating herding and integrating social media platforms are complements that generate sales, supporting the idea that it is beneficial to combine the two strategies on social media platforms. Furthermore, we find that herding is more salient for experience goods, consistent with our hypothesized mechanisms, while the effect of social media WOM is similar for experience goods and search goods.
Article
As digital and social media continue to grow in terms of usage and advertising expenditures, the advertising industry has been forced to develop innovative strategies. One such strategy is influencer marketing, which connects online personas with brands or services that target audiences trust and engage with regularly. This study provides insight into the experiences with and perceptions of influencer marketing among those working in advertising agencies. Nineteen U.S. advertising agency professionals were interviewed. Results show that the billion-dollar influencer marketing industry is largely uncharted territory that impacts strategic decision making and requires a shift in agency process for effective implementation.
Article
Mobile instant messaging is the communication technology revolution that is changing the way people communicate. The adoption rate is surging and the massive growth is experienced worldwide. Prior studies on instant messaging have been primarily concerned with its task-relevant or functional aspects, while less attention is paid to social interaction aspects. Drawing from media richness theory, present research highlights that social interaction factors can be facilitated by an enriched communication channel to create closer interpersonal social relationships. Specifically, the findings illustrate that the combined effect of text messaging and emoticon use increases information richness, which leads to perceived playfulness in mobile instant messaging. In addition, the perceived playfulness fostered in the instant messaging process plays a driving role in facilitating social connectedness, identity expressiveness between users and advocacy of mobile instant messaging usage.