Thesis

MSc. in Business Administration - Digital Marketing track

______________________________________________________________________________

Drivers of User engagement on TikTok

Student: Moshan Huang

Student number: 13276484

Date: 28 Jan 2022

Supervisor: Dr. Dennis Herhausen

1
Statement of originality

This document is written by Moshan Huang who declares to take full responsibility

for the contents of this document.

I declare that the text and the work presented in this document is original and that

no sources other than those mentioned in the text and its references have been used

in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of

completion of the work, not for the contents.

2
Table of contents

Statement of originality 2

List of Tables 5

Acknowledgements 6

Abstract 7

Introduction 8

Literature review 14
Social CRM 14

Engagement 15

Video characteristics 17
1. Hashtag 21
2. Weekday 22
3. Duration 22
4. Emoji 23
5. Text length 24

TikTok platform 24

Conceptual framework 26

Data and method 28

Research design 28

Stimuli 28

Procedure and measures 29

Sampling method 31

Data preparation 32

Results 33
Descriptive of the sample 33

3
 Correlations 33

Hypothesis testing 34

Discussion (of results) 38

Limitations and directions for future research 39

Managerial implications 39

Literature contributions 40

Conclusions 43

Reference 45

Appendices 56
Appendix 1: List of Calvin Klein video characteristics 56

Appendix 2: List of Levi’s video characteristics 57

Appendix 3: List of Louis Vuitton video characteristics 58

Appendix 4: List of Nike video characteristics 59

Appendix 5: List of SHEIN video characteristics 60

4
List of Tables

Table 1. Video characteristics

Table 2. Calculation of engagement rate

Table 3. Descriptive statistics of videos uploaded by different brands

Table 4. Overall means, standard deviations, correlations

Table 5. Hierarchical Multiple Regression Model of the Engagement rate

Table 6. Hierarchical Multiple Regression Model of the Popularity

5
Acknowledgements

Throughout the writing of this thesis, I have received a great deal of support and assistance. First

and foremost, I would like to convey my gratitude to Dr. Dennis Herhausen, my thesis

supervisor. He gave me lots of help and valuable advice and professional feedback. Thank you

very much.

In addition, I would like to express my gratitude to the second reader in advance for

spending the time and effort reading my thesis. I hope you will enjoy it. Lastly, I wish to express

my gratitude to my family and friends who have encouraged me to pursue a Master’s degree and

for their moral support during this closing phase of the program.

6
Abstract

Companies are always spending time and effort looking for ways to increase brand awareness

and brand recognition online. As a result, many companies are uploading many posts to their

official social media accounts, using text and video to promote their brand and products. This

helps attract and retain customers and even builds strong relationships between the company and

its customers. In fact, social media is now becoming a major competitive platform, with almost

every company or brand having its own account on different social media platforms. Extant

research provides different video characteristics to increase brand awareness and brand

recognition on different social media platforms. Based on these studies, this thesis was interested

in which characteristics have an impact on engagement rates on TikTok. In this research, five

different fashion brands were considered: Calvin Klein, Levi’s, Louis Vuitton, Nike, and SHEIN.

Through previous research, it was found that the number of hashtags, upload date (weekday or

weekend), the duration of the video, the number of emojis, and the text length of the description,

as video characteristics, may influence video engagement on social media. However, the results

of this study showed that only the duration of the video and upload date have a significant effect

on the engagement rate. Interestingly, the most significant factor in the engagement rate is the

account, meaning that brand awareness and popularity have a strong impact on the engagement

rate on social media. As not all video characteristics were able to have a significant impact on

engagement, future research should focus on more measurable video characteristics and video

content, and how different social media interact to more easily increase brand awareness and

brand recognition.

7
Introduction

Brands are always thinking about ways to improve brand awareness and brand recognition. In

order to make business thrive, they use different strategies and carry out various campaigns. So,

which strategy is the most advanced? How can campaigns be designed to evolve with the social

environment?

In a highly competitive and fast-changing environment today, many industries are under

pressure to seek ways to grow. They use strategies to interact with customers in an emotional,

cognitive and behavioral way, and Customer Relationship Management (CRM) is one of them.

Simply stated, CRM is about finding, getting, and retaining customers. Companies that

successfully implement CRM will reap the rewards in terms of customer loyalty and long-term

profitability (Chen and Popovich, 2003). As social media entered people's lives, the term "social

CRM" has emerged to describe this new way of developing and maintaining customer

relationships, which means integrating customer-facing activities, including processes, systems,

and technologies, with emergent social media applications to engage customers in collaborative

conversations and enhance customer relationships (Greenberg, 2010; Trainor, 2012). As a result,

social customers get a sense of being connected to the company and, are likely to be rewarded

for the efforts they make for the company. If things go well, they become advocates. Their

loyalty is attitudinal, not just behavioral. (Greenberg, 2010). Brands are gradually realizing the

value of social CRM and are making considerable investments in the technology. For example, a

growing number of fashion brands, including Louis Vuitton, Shein, Clavin Klein, and Levi's,

have joined the popular social media platform TikTok in recent years (Parisi, 2020; McCormick,

2021; 24 Inspiring Brands on TikTok to Fuel Your Creative Strategy, 2020). However, not all

brands succeeded in gaining large audiences. Some of them failed to be embraced by daily users

8
on social media platform (Parisi, 2020). So, why did these brands fail? How can they build

relationships with their potential customers on social media?

Fashion brands use social media to promote their products. Social media is a large ecosystem

with a complex network of interconnections and a variety of social networks and levels of

involvement. It is known as “a group of Internet-based applications that build on the ideological

and technological foundations of Web 2.0, and it allows the creation and exchange of user-

generated content (Kaplan and Haenlein, 2010). Previous research showed that customer

perceptions of the strength of the brand community on social media are an important factor for

customer satisfaction and customer intent to generate positive word-of-mouth (WOM) about the

brand (Zhang and Luo, 2016). Also, social media strategies are vital for brands to stay connected

to consumers (Yılmaz and Enginkaya, 2015). According to research, e-WOM mainly drives

online performance and has a stronger impact on offline performance (Pauwels et al., 2016). In

addition, social media utilization has had a significant effect on the success of other industries.

The amount of content shared, for example, was found to have a positive association with the

success of crowdfunding (Clauss et al., 2019). As a result, social media is a critical component of

a brand's digital social capital, and how managers use this online communication channel to

engage customers on new media will increasingly influence future business success (Shahzad et

al., 2019).

When thinking about fashion inspiration and finding the latest trends, Instagram is probably the

first option that comes to mind. However, in recent years, many fashion brands have joined

9
TikTok. Why did they enter TikTok? Is there any difference between TikTok and Instagram or

other social media platforms? In fact, compared to Facebook, Twitter, and Instagram, there is a

dearth of research on TikTok. Many brands exploring TikTok are coming to realize that TikTok

is not just another ad placement space like Facebook or Instagram. A valueless video will not be

appreciated there. Brands should not just put their ads on TikTok; it is a waste of time and does

not help with sales. Brands must customize their content to match the platform perfectly.

TikTok is a social media platform for making a variety of short-form videos, including

dance, comedy, and educational genres, ranging from 15 seconds to one minute in length

(Schwedel, 2018). Nowadays, short-form videos are becoming increasingly popular. TikTok is

the most popular app in 2020, with 850 million downloads worldwide. Until now, TikTok has

been installed more than 2.6 billion times on devices around the world (Geyser, 2021).

According to Bucknell Bossen and Kottasz (2020), older adolescent members who reported

actively producing TikTok videos were significantly more likely to agree that they did so to

satisfy needs for praise and social recognition, as well as self-identity creation and fame-seeking

desires, as well as relationship maintenance and building. In Wang's (2020) TikTok study, the

humor and camera view of the videos have substantial effects on the viewer's sense of

immersion, social presence, and entertainment. As a result, how brands can leverage TikTok's

characteristics to attract new customers and preserve relationships with existing ones is an issue

worth investigating.

Given the fact that posting videos on TikTok poses a challenge for many industries, some

researchers have started to investigate what managers in certain industries can do to overcome

the poor situation on TikTok. However, there is no clear solution as to what kind of video

characteristics the fashion industry should add on their TikTok videos. In this case, brand

10
managers need some guidance to understand the differences between different social media

platforms and to quickly adapt to the platform they work on. Otherwise, their first step in

attracting potential customers and increasing engagement on TikTok will fail.

Engagement is a measure of the relationship between the customer and the brand. Hollebeek

(2011) defines customer–brand engagement as “the level of an individual customer’s

motivational, brand-related, and context-dependent state of mind characterized by specific levels

of cognitive, emotional, and behavioral activity in direct brand interactions”. On the social media

platform, high social media engagement boosts brand relationships, affecting brand trust,

commitment, and loyalty (Dessart, 2017). This measure can assist managers in determining

which characteristics will attract customers and become brand supporters.

To understand what kinds of measurable video characteristics on TikTok can increase

engagement, several video characteristics will be used in the analysis, including the number of

hashtags, upload date, the duration of the video, the number of emojis, and the text length of the

description. Five different fashion brands, including Calvin Klein, Levi's, Nike, SHEIN, and

Louis Vuitton, will be selected. Their videos posted on TikTok differ significantly in many

dimensions. For example, SHEIN is the latest fast-fashion brand, while Nike is a famous

sportswear and apparel company founded more than half a century ago. Nike is a world-

renowned company that seems to be more popular than SHEIN. However, the result is different

on their TikTok sites. The videos posted by SHEIN often have 5,000 to 6,000 likes, and some

even have millions of likes. Whereas the number of likes for videos posted by Nike usually stays

11
at around 1000 or 2000. This is the power of social media platforms to make a new brand famous

in a short time.

Despite the various studies concerning social media marketing and different platforms, research

on how video characteristics affect engagement rate on TikTok is still limited. For managers,

although they have made considerable efforts on TikTok, they still failed to promote their brands

there (Parisi, 2020). So, what are the reasons for their failure? In addition, how can they make

their brands stand out from TikTok? It is apparent that more insight is needed. Therefore, our

study aims to explore what kinds of videos characteristics drive people to take action after

watching the whole video produced by the official brand account on TikTok. Specifically, the

number of hashtags they used, the date the video was uploaded, the duration of the video, the

number of emojis they used in the description, and the text length of the description will be

examined, and the relationship between these video characteristics and the engagement rate will

be explored.

The rest of this research is outlined below. First, existing literature will be reviewed, why

brands should work on social media, what motivates people to like, comment and share on other

social media platforms, and what kind of formats are likely to be popular on TikTok. These will

be followed by hypotheses on the characteristics of the videos. The methodology and results of

the research will then be described. Based on these results, a discussion of the findings will be

presented. Next, the limitations of this study and directions for future research will be outlined.

Finally, this study will be concluded by indicating the implications of the results of this study.

The managerial contribution of this research is to provide managers with ways to improve their

12
TikTok videos to get a higher level of engagement rate and win on the highly competitive social

media platform. In terms of the academic part, this study will provide some insights related to

short-form videos and provide an understanding of what kinds of short-form video characteristics

should be added to inspire likes, comments, and shares.

13
Literature review

Social CRM

CRM is a process by which a company maximizes the use of customer information to increase

loyalty and retain customers’ business over their lifetimes. The primary goals of CRM are to (a)

build long-term and profitable relationships with chosen customers, (b) get closer to those

customers at every point of contact, and (c) maximize the company’s share of the customer’s

wallet (Choy et al., 2002). In short, CRM focuses on utilizing and developing interactions with

customers in order to increase customer satisfaction, retention, and eventually profitability.

This century has seen an explosion in the number of people using online social

networking, which has led to a new type of interaction between customers and brands. A new

term, "social CRM" emerged. Social CRM is a philosophy and a business strategy, supported by

a system and technology, designed to engage the customer in a collaborative interaction that

provides mutually beneficial value in a trusted and transparent business environment (Chan et

al., 2018). In brief, social CRM is to add a social media strategy to the CRM initiatives. It

focuses on identifying, engaging, and enabling the "influential" nodes in all iterations of the

interactions (among the company, business partners, and customers). Customers are treated as

partners in the development and improvement of products, services, and the company-customer

relationship from the start. Businesses focus on creating environments and experiences that

engage customers (Greenberg, 2010). The company now needs to connect with the growing

number of customers using the Internet to discover products, services, and companies that can

help them solve issues (Leary, 2008).

Therefore, using a social CRM strategy can help companies attract the attention of those

who use the Internet or social media. In addition, the shift from CRM to social CRM signaled

14
that the customer was no longer the object of a sale but instead defined the subject line

(Greenberg, 2010). The customer is now a social customer. This social customer is the person

who drives the need for social CRM. They require transparency and authenticity from the

companies they choose to work with. They get a sense of being somehow connected to the

company and, most likely, rewards for their efforts on behalf of the company. If things go well,

they become advocates. Their loyalty is attitudinal, not just behavioral. (Greenberg, 2010).

To build this relationship with social customers, there is nothing better than relevant and

compelling content to capture the attention of viewers. Having the right content and details will

help connect the company to those needing the product or service. Whether it is a podcast, blog

post, TikTok video, or a post on other social media platforms, creating attractive content is a key

pillar of the social CRM strategy (Leary, 2008).

Engagement

“Engagement" has been viewed as a promising concept expected to provide enhanced predictive

and explanatory power of focal consumer behavior outcomes, including brand loyalty (Avnet

and Higgins, 2006; Pham and Avnet, 2009; Schau et al., 2009; Hollebeek et al., 2014). In fact, it

has gotten a lot of attention from academics in a variety of fields, including social psychology

and marketing. Early this century, Algesheimer et al. (2005) defined engagement as "the positive

influences of identifying with the brand community through the consumer's intrinsic motivation

to interact/co-operate with community members”. They focus on three multidimensional

dimensions: utilitarian, hedonic, and social domains. Six years later, Hollebeek (2011) defined it

a bit differently: “the level of an individual customer's motivational, brand-related, and context-

dependent state of mind characterized by specific levels of cognitive, emotional, and behavioral

15
activity in brand interactions”. Also, Hollebeek focuses on three different dimensions: cognitive,

emotional, and behavioral. However, as more and more people use social media to connect with

their consumers, people tend to pay more attention to “social media engagement” than

“consumer engagement” or “community engagement”.

Social media are vast ecosystems with intricate networks of relationships and a variety of

social nets and levels of involvement. It enables "more frequent, faster, and richer interactions

among large groups of people" (Sashi, 2012). When it comes to consumer engagement, one of

the most essential features of social media is that it enables for customer-brand interactions.

Social media engagement is a context-specific occurrence of consumer engagement since

engagement varies across online media (Brodie et al., 2013).

Paek et al. (2013) measured social media engagement with 13 items on a five-point Likert

type scale adapted from Calder et al. (2009). The weakness of this method is that the data used to

measure social media engagement is derived from surveys (Bonsón et al., 2016). Instead, the

most effective way to understand the impact of the shares on the public is to interpret social

media indicators, such as page views, likes, shares, and comments. In the literature, many

methods and formulas have been developed to measure social media interaction in this way.

Bonsón and Ratkai (2013) developed a set of metrics to evaluate engagement on social media

platforms. These metrics use three different tools available on Facebook: like, comment, and

share. However, this engagement rate calculation method has become obsolete due to an

algorithm change. For example, the number of Facebook fans is no longer a valid measure in

terms of engagement rate calculation (Bitiktas and Tuna, 2020).

16
 Video characteristics

There are various video characteristics: audio and visual features; third-person or first-person

angle; humor use; background music; video length, title length, content type; relaxing

entertainment motives, social interaction motives, information-giving motives; schedule; objects,

emotions, story, message tone, and emojis. Video characteristics can be depicted in different

areas, including visual, psychological, hearing, and so on. Some authors have contributed to

enhancing the literature. The most popular video characteristics in the area of social media

engagement are presented in Table 1.

Some characteristics are hard to measure, and some are not realistic for the brands. For

example, it is hard to measure humor. Also, most of the videos uploaded by the brand are shot

from a third-person angle, so it would be useless to analyze the third-person or first-person angle.

All these video characteristics need to be removed.

Finally, the number of hashtags, uploading on weekdays or weekends, the duration of the

video, the number of emojis, and the text length of the description will all be examined in this

research.

17
Table 1. Video characteristics

Author Title Video characteristics and

finding
Wang and Ji Video Affective Content Analysis: A Survey Emotion is an important
(2015) of State-of-the-Art Methods component in the
classification and retrieval of
videos. Video affective
content analysis can be
divided into two approaches:
direct and implicit. Direct
video affective content
analysis includes the audio
and visual features. Implicit
video emotional content
analysis comes from
physiological signals and
visible behaviour.
Y. Humor and camera view on mobile short- The third-person angle
Wang(2020) form video apps influence user experience provides more sense of
and technology-adoption intent, an example immersion than a high level of
of TikTok humor use and the first-person
angle provides more sense of
immersion than the third-
person angle in short-form
mobile videos on social
media.
Liikkanen Music on YouTube: User engagement with Traditional music videos
and traditional, user-appropriated and derivative receive more views, but
Salovaara videos derivative music videos invite
(2015) more active viewer
participation through
commenting and voting.
While music videos gather
more views, engagement
differences between other
content genres are minuscule.
Chen et al. Factors Driving Citizen Engagement With Video length (negatively),
(2021) Government TikTok Accounts During the title length (positively), and
COVID-19 Pandemic: Model Development content type (partially)
and Analysis significantly influenced the
level of citizen engagement
during the COVID-19
pandemic.

18
Khan Social media engagement: What motivates The strongest predictor for
(2017) user participation and consumption on liking and disliking videos
YouTube? was the relaxing entertainment
motive; commenting and
uploading were strongly
predicted by the social
interaction motive; sharing
was strongly predicted by the
information giving motive.
Hoiles et al. Engagement and Popularity Dynamics of The video view count has a
(2017) YouTube Videos and Sensitivity to Meta- casual effect on the subscriber
Data count of the channel. Also, it
was found that going “off-
schedule” can actually
increase the popularity of a
channel.
Trzcinski Predicting popularity of online videos using Visual and temporal features
and Rokita Support Vector Regression (e.g., view counts, social
(2017) features) can be helpful in
predicting future video
popularity. Additionally,
social features represent a
much stronger signal in terms
of video popularity prediction
than visual ones.
Schwenzow Understanding videos at scale: How to extract Several important video
et al. (2021) insights for business research features, including objects,
emotions, text content, story,
and message tone, have had
an impact on the illustrative
application.

Munaro et To engage or not engage? The features of Medium-length and long

al. (2021) video content on YouTube affecting digital videos posted during non-
consumer engagement business hours and weekdays,
and those using subjective
language styles, less-active
events, and temporal
indications are more likely to
receive views, likes, and
comments.
Li et al. Video mining: Measuring visual information Visual variation and video
(2019) using automatic methods content can have a positive
effect on the success of
crowdfunding campaigns.

19
Tellis et al. What Drives Virality (Sharing) of Online Information-focused content
(2019) Digital Content? The Critical Role of negatively influences sharing.
Information, Emotion, and Brand Prominence Content that uses drama and
drama-supporting elements
(e.g., a plot, characters, or
surprise) positively impacts
engagement and the arousal of
emotions. High levels of
brand prominence (longer
duration, earlier placement) in
ads negatively impact sharing.
Geuens et Emotional advertising: Revisiting the role of Emotional ads using product-
al. (2011) product category congruent emotions lead to
more positive ads and brand
attitudes than non-emotional
ads, irrespective of whether
the product is utilitarian or
hedonic, or a low- or high-
involvement product.
Hautz et al. Let Users Generate Your Video Ads? The The influence of different
(2014) Impact of Video Source and Quality on video generators (user vs.
Consumers' Perceptions and Intended agency) on spectators'
Behaviors perception and intended
behavior depends on the
technical quality of the video.
In the case of low technical
quality, users as generators
have a significantly stronger
positive effect on source
trustworthiness and expertise
– and thereby on consumers'
intended behaviors – than
AGVs (Agency-generated
videos).
McShane et Emoji, Playfulness, and Brand Engagement Emoji presence increases
al. (2021) on Twitter engagement with tweets, with
more emoji leading to more
likes and retweets. The effect
of emojis on brand
engagement varies depending
on the nature of the interplay
between emojis and text, and
the subsequent effect of this
interplay on perceived
playfulness.

20
 Shin et al., The power of e-WOM using the hashtag: Hashtag utilization has
(2018) focusing on SNS advertising of SPA brands significantly positive effects
on e-WOM. It allows users to
discuss certain topics, thus
widening the scope of
potential customers.

1. Hashtag

The use of hashtags, represented by the pound sign (#), indicates that a message is relevant to a

particular topic. The hashtag was first used to categorize messages. It can make searching for

information easier. For example, if a user wanted to find information about healthcare, a search

for the term "healthcare" would yield results, but a search for #healthcare would ensure that all

results were relevant to the topic (Lovejoy et al., 2012). Hashtags allow users to cluster around

specific topics—essentially creating discursive clusters around a shared interest (Bode et al.,

2015). The more hashtags the video maker adds to their videos, the more likely users are to find

those videos on TikTok. This greatly influences social networking sites: they can easily be used

as a communication medium between brands and their customers. In Shin et al.'s (2018) study,

hashtag utilization has significantly positive effects on informativeness, enjoyment, interactivity,

attitudes toward advertising, and e-WOM.

However, many people now find it disturbing to see a post with too many hashtags. Using

too many hashtags often looks unnatural to followers. In addition, it can even distract viewers

from the main message of the post (Cadena, 2020). Therefore, it can be assumed that the number

of hashtags negatively impacts the engagement rate.

H1: There is a negative relationship between the number of hashtags and the engagement rate.

21
2. Weekday

A TikToker may always consider “when is the best time to post on TikTok?” Although,

according to Soares et al. (2021), characteristics related to posting day and time are not related to

engagement on Facebook, they may vary across platforms. Munaro et al. (2021) have already

studied this characteristic on YouTube and found that posting a video on weekends positively

affects the number of comments but posting on weekends does not positively affect the number

of views and likes. In addition, videos posted on certain weekdays (Monday, Tuesday, and

Thursday) were significantly more popular than those posted during the weekend. In research

from Landsberg (2021), Tuesday 9 am, Thursday 12 am, and Friday 5 am have particularly high

levels of engagement on TikTok. Therefore, we will measure upload date by 2 categories in this

research: weekend and weekday. According to the study mentioned above, it can be assumed that

posting on weekdays will have a higher engagement rate on TikTok.

H2: There is a positive relationship between uploading on weekdays and the engagement rate.

3. Duration

TikTok is a short-form video platform with a duration of 15 seconds to 3 minutes. Munaro et al.

(2021) found that medium and long YouTube videos posted during non-business hours and

weekdays are more likely to receive views, likes, and comments. This suggests that video length

affects video popularity. But TikTok, unlike YouTube, does not support long videos. So, what is

the appropriate length of a TikTok video for viewers to have the patience to watch, like,

comment and share it?

22
 During the COVID-19 pandemic, videos were frequently published on government social

media platforms. Chen et al. (2021) found that video length significantly influenced the level of

citizen engagement on TikTok. At that time, they found that shorter videos elicit greater citizen

engagement on TikTok. Therefore, it can be proposed that for the fashion industry, video length

has a negative effect on engagement rate.

H3: There is a negative relationship between the duration of the video and the engagement rate.

4. Emoji

Emojis, which are pictographs that represent facial expressions, people, places, or things, are

becoming a mainstream form of communication (McShane et al., 2021). It originated from

smileys, which first evolved into emoticons, followed by emoji and stickers in recent years.

Smiley first appeared in the 1960s and is regarded as the first expression symbol. Smiley is a

yellow face with two dots for eyes and a wide grin that is printed on buttons, brooches, and t-

shirts. By the early 1980s, this symbol had become widespread, emerging as a permanent feature

of western popular culture (Stark and Crawford, 2015). People use emojis as a way of adding

personality to messages, enhancing media richness, reciprocating emotion, and conveying both

emotion and intention. The use of emojis in marketing activities can enhance the appeal of these

activities and bring them closer to the younger generation. It can also have an impact on

consumers, including optimizing the consumer experience, improving purchase intention, and

changing perceptions of brands (Bai et al., 2019). More importantly, the presence of an emoji

will increase brand engagement on Twitter (McShane et al., 2021). Therefore, it can be expected

that the number of emojis positively affects the engagement rate.

23
H4: There is a positive relationship between the number of emojis and the engagement rate.

5. Text length

In research from Chen et al. (2021), they found that title length is a video characteristic that is

related to the number of shares, likes, and comments. Title length refers to the total number of

words describing the video’s content in the title of an uploaded video. When customers want

information about a brand or are looking for a discount, they will be happy to read large sections

of text. For example, short video lengths, longer descriptions, more reliable video sources, and

lower medical information content were more popular with audiences during the COVID-19

pandemic (Liu et al., 2021). However, young people nowadays are increasingly impatient and

more reluctant to read long paragraphs of text (Levine et al., 2007). Therefore, it is expected that

text length of description is negatively related to the engagement rate.

H5: There is a negative relationship between the text length of the description and the

engagement rate.

TikTok platform

The above predictions will be examined on TikTok. TikTok is one of the most dominant social

media platforms with 1 billion active monthly users and is used by consumers and brands alike

(Statista, 2021). According to TikTok Statistics – Everything You Need to Know (2021), the

average time spent on TikTok per day is 52 minutes, and younger people ages four to fifteen

spend an average of 80 minutes per day. What’s more, this platform has an incredibly high

engagement rate.

24
 Over the past two years, many fashion brands have joined TikTok. Because TikTok is a

newly popular social media platform, many fashion brands are still looking to find ways to

engage their audience. Therefore, it is interesting to see how these fashion brands use their

videos to influence viewers’ behaviors on TikTok. Consumers may be influenced by the videos’

different characteristics in their actions to consume and contribute to the posts. The five

characteristics mentioned above are important factors for TikTok users in deciding whether to

watch the whole video or not, and thus whether to like, comment, or share it.

25
Conceptual framework

The proposed conceptual framework considers the use of the number of hashtags, weekday

uploading, the duration of the video, the number of emojis, and text length of the description as

independent variables, with engagement rate as the dependent variable. In particular, this

research will investigate and compare whether these TikTok video characteristics positively or

negatively influence the engagement rate of the video. Moreover, it will be explored what the

combined effects are.

Figure 1. Conceptual framework

H1 Hashtag

H2 Weekday

H3 Duration Engagement rate

H4 Emoji

Text
H5 length

H1: There is a negative relationship between the number of hashtags and the engagement rate.

H2: There is a positive relationship between uploading on weekdays and the engagement rate.

H3: There is a negative relationship between the duration of the video and the engagement rate.

26
H4: There is a positive relationship between the number of emojis and the engagement rate.

H5: There is a negative relationship between the text length of the description and the

engagement rate.

27
Data and method

Research design

In order to scientifically test the hypotheses, the data was scraped from the platform TikTok. In

the research, the independent variables were the number of hashtags, weekday, the duration of

the video, the number of emojis, and the text length of the description, and the engagement rate

was measured as the dependent variable. To test the video characteristics only, four control

variables were added. In the first step, we only test the engagement rate with the control of the

different accounts. Then, in step 2, the five different independent variables will be entered.

Stimuli

Brand choosing

For the fashion brands, Calvin Klein, Levi’s, Louis Vuitton, Nike, and SHEIN were selected.

These five brands represent different fashion trends. Calvin Klein represents the modern fashion

apparel industry, Levi’s stands for the classic jeans industry, Louis Vuitton represents the luxury

industry, Nike is famous for its sportswear products, and SHEIN represents the fast fashion

industry. Choosing more than one brand will avoid contingency and ensure the applicability of

other brands. In this research, 100 videos of five different fashion brands were collected from

TikTok to see if there was a relationship between the different video characteristics and the

engagement rate of the video.

28
 Procedure and measures

In this research, all the data of top brands will be scraped on TikTok. All the videos will at least

have one month for TikTok users to like, comment, and share the videos. The influence of brand

popularity will be minimized by using control variables.

Independent variables: Hashtag, date, duration, emoji, and text length

The independent variable duration of the video was directly scraped from the TikTok platform.

On TikTok, the number of hashtags, the number of emojis, and the text length of the description

can be easily counted. The weekday variable is first scraped from TikTok and then transformed

into "weekday" or "weekend".

Dependent variables: Engagement rate

The independent variables can be collected from TikTok platform without calculation. However,

the dependent variable, the engagement rate needs some calculation. To calculate engagement

rate, the number of views, likes, comments, and shares will be first collected from TikTok. Then,

as there is a gap between the existing research on the calculation of engagement rate and this

study, a small change to the calculation of engagement rate based on the Bonsón and Ratkai

(2013) study is needed. In our case, the popularity and reputation of a certain brand may

influence the number of fans, and then the level of engagement rate. In addition, this calculation

is better suited to calculating the engagement rate for all posts on a given account rather than a

single post. However, in our case, the engagement rate of single video is the dependent variable

29
and it is important to compare the engagement rates of these videos. Therefore, it is necessary to

address these issues and simplify the engagement rate formula. The mean number of likes per

post (P2) will be replaced by the number of likes of that video, the mean number of comments

per post (C2) will be replaced by the number of comments on that video, and the mean number

of shares per post (V2) will be replaced by the number of shares of that video. In addition, the

number of fans will be replaced by the number of times that video has been viewed, which can

help reduce the impact of brand reputation on engagement rate. The engagement rate is the

modified engagement rate (E), calculated by dividing the number of likes, comments and shares

by the number of views. As the number of likes tends to be much higher than the number of

comments and shares for videos on the TikTok platform, popularity (P) will be analyzed in the

robustness test. It is calculated by dividing the number of likes by the number of views.

Table 2. Calculation of engagement rate

Category Sign Formula Measure

Popularity P1 Number of posts liked/total Percentage of posts that have been
(P) posts liked
P2 Total likes/total posts Mean number of likes per post
P3 (P2/number of fans) × 1000 Mean number of likes per post per
1000 fans
Commitment C1 Number of posts Percentage of posts that have been
(C) commented/total posts commented on
C2 Total comments/total posts Mean number of comments per post
C3 (C2/number of fans) × 1000 Mean number of comments per post
per 1000 fans
Virality (V) V1 Number of posts shared/total Percentage of posts that have been
posts shared
V2 Total shares/total posts Mean number of shares per post
V3 (V2/number of fans) × 1000 Mean number of shares per post per
1000 fans
Engagement E P3 + C3 + V3 Stakeholder engagement index
Source: Bonsón and Ratkai (2013)

30
Control variables: CK, Levi’s, LV and Nike

After labeling the brand to which each video belonged, the dummy variables of CK, Levi’s, LV,

and Nike were created. For the dummy variable of CK, videos from CK were coded as 1, and

other videos were coded as 0. For the dummy variable of Levi’s, videos from Levi's were coded

as 1, and other videos were coded as 0. For the dummy variable of LV, videos from LV were

coded as 1, and other videos were coded as 0. For the dummy variable of Nike, videos from Nike

were coded as 1, and other videos were coded as 0. These control variables can help minimize

the impact of brand popularity and reputation.

Sampling method

The data was collected on TikTok from September 1 to December 1, 2021. All the selected

videos are uploaded in 2021. Most of them are produced in June and July. One of the

requirements when choosing a video is to minimize the impact of external factors of the video.

For example, the video posted by Louis Vuitton on July 6, 2021, has been deleted. This video

has a surprisingly high engagement rate of over 14% and has even gotten 29.5 million views and

4 million likes, which is obviously higher than others. This is due to cooperation with the famous

South Korean boy band, BTS. As a result, many fans of this band have come to view, like,

comment, and share this video. This also led to much higher figures for this video than for other

videos posted by Louis Vuitton. In this case, the engagement rate of this video was affected by

the fame of this Korean band, which would reduce the association between the engagement rate

and the video characteristics.

31
 Data preparation

All data were prepared and analyzed in IBM SPSS Statistics 28. First, the data gathered via

TikTok were checked for missing and extreme values. A frequency test was conducted on all

variables. The frequency table showed that no missing values were found. For some extreme

values and outliers, they were replaced by other to give a sample of 20 per account. In the

research, most of the data is numerical, so these scores do not need to be coded in reverse.

In the normality part, descriptive statistics, skewness, kurtosis, and normality tests were

performed. All items of duration and text length have a skewness between 0 and 0.5, and the

kurtosis is between -0.5 and -1. For the independent variable hashtag, both skewness and kurtosis

are between 0 and 1. For weekday, skewness is between -2.5 and -2, and kurtosis is between 2

and 2.5, which indicates that the most frequent scores are grouped toward the right end of the

distribution, and it has many scores in the tails. For emoji, both skewness and kurtosis are

between 1 and 1.5, which indicates that the most frequent scores are grouped toward the left end

of the distribution, and the distribution is pointy. For the dependent variable, the engagement

rate, both the skewness and kurtosis are between 1 and 2. This means that the most frequent

scores are grouped toward the left end of the distribution, and the distribution is pointy. Finally,

for the control variables CK, Levi’s, LV, and Nike, their skewness is between 1.5 and 2, and

their kurtosis is between 0 and 0.5. This means that the distribution is positively skewed.

Although in this study the skewness and/or kurtosis of weekday, emoji, engagement rate, CK,

Levi’s, LV, and Nike are not between -1 and 1, we assume that the impact will be rather small.

Because the sample is large enough, the risk is reduced. Neither skewness nor kurtosis would

make a substantive difference in the analysis.

32
Results

Descriptive of the sample

Table 3 shows the average views, likes, comments, shares, and followers of videos uploaded by

different brands. Through this table, it is obvious that these brands have totally different levels of

views, likes, comments, and shares.

Table 3. Descriptive statistics of videos uploaded by different brands

Category Calvin Levi's Louis Vuitton Nike SHEIN

Klein
Mean views 5,309 1,576,898 4,819,785 38,045 301,405
Mean likes 192 48,221 113,950 2,393 34,426
Mean comments 7 171 858 45 548
Mean shares 4 102 359 38 133
Followers 63K 432.7K 1.4M 2M 3.3M

Correlations

For the correlation analysis, SPSS provides a table of correlation coefficients, which is the

correlation matrix, for all the combinations of variables. From the correlation carried on, duration

is the strongest predictor among all other independent variables, with a Pearson correlation

coefficient of r = 0.324 and the significance value less than 0.001. The second strongest

predictor, emoji, has a Pearson correlation coefficient of r = 0.294, p < 0.01. For the remaining

variables, weekday, hashtag, and text length, they were not statistically significant with 0.178, -

0.159, and -0.045, respectively. Furthermore, this correlation table showed that not only did

duration and emoji correlate with engagement rate, but duration also correlated with emoji and

text length.

33
Table 4. Overall means, standard deviations, correlations

M SD 1 2 3 4 5 6 7 8 9

1 Hashtag 1.560 1.067

2 Weekday 0.860 0.349 -0.086

3 Duration 21.740 9.436 -0.069 0.072

4 Emoji 0.520 0.717 0.144 0.132 0.202*

5 Text Length 10.520 6.096 -0.003 -0.008 0.417*** 0.143

6 Engagement 0.056 0.033 -0.159 0.178 0.324*** 0.294** -0.045

7 CK 0.200 0.402 -0.264** -0.159 -0.250* -0.364*** 0.196 -0.240*

8 Levis 0.200 0.402 0.396*** 0.058 0.187 0.301 0.291** -0.196 -0.250*

9 LV 0.200 0.402 0.090 -0.014 -0.455*** -0.364 -0.669*** -0.267** -0.250* -0.250*

10 Nike 0.200 0.402 -0.240* 0.202* 0.408*** 0.091 0.340*** 0.135 -0.250* -0.250* -0.250*

Note. Statistical significance: *p <.05; **p <.01; ***p <.001

Hypothesis testing

Next, a hierarchical regression analysis is used to examine the relationship between the

independent variables (hashtag, weekday, duration, emoji, and text length) and the dependent

variables (engagement rate). Hierarchical multiple regression was performed to research the

ability of hashtags, weekday, duration, emoji, and text length to predict engagement rate after

controlling the dummy variables of different accounts. Table 5 provides an overview of the

regression analysis. As a first step of the hierarchical multiple regression, four predictors have

been entered: the dummy variables for Calvin Klein, Levi’s, Louis Vuitton, and Nike. This was

done for these four dummy variables so that a shared variability of these variables with the

predictive variables could be controlled. The observed effect of those independent variables on

engagement rate is therefore independent of the effect of this control variable.

In the first step of the multiple regression, the relationship between these two models was

not very significant, F (4, 95) = 16.295, p < 0.001, and explained 40.7% of the variance in the

engagement rate. After the entry of hashtag, weekday, duration, emoji, and text length at step 2,

34
the total variance explained by the model as a whole was 48.2%. The introduction of the five

independent variables explained an additional 7.5% variance in engagement rate, after

controlling for the account (R² Change = 0.075, F (5, 90) = 2.608, p = 0.030). In the final model,

six out of nine predictor variables were statistically significant, with LV recording a higher beta

value (β = -0.607, p < 0.001) than Levi’s (β = -0.559, p < 0.001), CK (β = -0.524, p <0.001),

Nike (β = -0.384, p < 0.001), duration (β = 0.212, p = 0.037) and date (β = 0.164, p = 0.042). In

other words, if the duration of the video increases for one, the engagement rate will increase for

0.212; if the upload date changes from weekends to weekdays, the engagement rate will increase

by 0.164. On average, LV obtains a 0.607 point lower score in engagement rate than other

accounts; Levi’s obtains a 0.559 point lower score in engagement rate than other accounts; Nike

obtains a 0.384 point lower score in engagement rate than other accounts.

In this table 5, hashtag, as an independent variable, is not significant and has a low and

negative beta value of -0.094. Therefore, hypothesis 1 is not supported. Weekday, however, is

significant at the 0.05 level. This provides support for hypothesis 2. Duration is also significant

at the 0.05 level, but it has a positive beta value, which is the opposite of what was proposed in

the hypothesis 3. Therefore, hypothesis 3 was rejected. The fourth and fifth ones are both not

significant. Therefore, hypothesis 4 and hypothesis 5 cannot be confirmed.

35
Table 5. Hierarchical Multiple Regression Model of the Engagement rate

R R² R² change B SE β t
Step 1 0.638 0.407***
CK -0.054 0.008 -0.647*** -6.474
Levi’s -0.051 0.008 -0.611*** -6.116
LV -0.056 0.008 -0.668*** -6.689
Nike -0.029 0.008 -0.347*** -3.473
Step 2 0.694 0.482* 0.075*
CK -0.043 0.010 -0.524*** -4.220
Levi’s -0.046 0.009 -0.559*** -5.177
LV -0.050 0.011 -0.607*** -4.623
Nike -0.032 0.009 -0.384*** -3.404
Hashtag -0.003 0.003 -0.094 -1.064
Weekday 0.016 0.008 0.164* 2.063
Duration 0.001 0.000 0.212* 2.123
Emoji 0.003 0.005 0.057 0.574
Text length -0.001 0.001 -0.150 -1.266
Note. Statistical significance: *p <.05; **p <.01; ***p <.001

Popularity can be used to test the robustness of the engagement rate. Popularity, as

mentioned in the procedure and measure, is calculated in a similar way to the engagement rate.

The results showed that in the regressions, popularity has a similar impact to the engagement

rate, with only slight variations in the data (Table 6). Duration has the strongest impact on both

popularity and engagement. Of the five independent variables, only two of them, duration and

weekday, were consistently significant in their relationship with engagement rate and with

popularity. However, after entering the five independent variables, the popularity model explains

an additional 8% of the variance, which is 0.5% higher than the variance of the engagement rate.

In addition, the beta values are also different. The duration variable had a slightly higher impact

on popularity than on engagement rate, while the weekday variable had a higher impact on

engagement rate than on popularity. Overall, the differences between the data are very small.

This means that the results are reliable.

36
Table 6. Hierarchical Multiple Regression Model of the Popularity

R R² R² change B SE β t
Step 1 0.627 0.393***
CK -0.053 0.008 -0.647*** -6.403
Levi’s -0.048 0.008 -0.594*** -5.874
LV -0.053 0.008 -0.649*** -6.419
Nike -0.027 0.008 -0.335** -3.313
Step 2 0.688 0.473* 0.080*
CK -0.043 0.010 -0.526*** -4.204
Levi’s -0.044 0.009 -0.536*** -4.921
LV -0.048 0.011 -0.588*** -4.446
Nike -0.030 0.009 -0.374** -3.289
Hashtag -0.003 0.003 -0.107 -1.198
Weekday 0.015 0.008 0.163* 2.028
Duration 0.001 0.000 0.218* 2.167
Emoji 0.002 0.005 0.051 0.517
Text_len -0.001 0.001 -0.156 -1.304
Note. Statistical significance: *p <.05; **p <.01; ***p <.001

37
Discussion (of results)

The purpose of this study was to test the impact of the ability of the number of hashtags,

weekday, duration of the video, the number of emojis, and text length of the description on the

level of engagement rate. Prior research has shown that these five elements have an impact on

different social media posts, but it has not been demonstrated whether they have an impact on the

engagement rate on TikTok videos. In this case, a correlation test was first conducted and only

the duration of the video and the number of emojis are correlated with engagement rate. Then,

hierarchical multiple regressions were conducted to examine this impact after controlling for

different accounts. The testing provided evidence that engagement rate was influenced both

positively by weekday and negatively by duration. The remaining predictors, the number of

emojis, the number of hashtags, and the text length of the description, cannot be fully accepted.

In addition, these can only explain the 7.5% variance in engagement rate, and only two out of

five variables (weekday and duration) are strongly associated with the dependent variable

engagement rate. These unsatisfactory results may because of the small sample size of only 100

observations, so that no support for conventional significance levels could be found.

Furthermore, although some other researchers found that the number of hashtags, the number of

emojis and text length in description have an impact on other platforms, there is insufficient

evidence to conclude that there is an impact on the engagement rate on TikTok. Of these three

variables, the number of hashtags and text length had a non-significant negative effect on

engagement rate, while the number of emojis had a non-significant positive effect on

engagement rate. To test if the results are robust, engagement rate was compared with the

popularity. The results showed that the effect of engagement rate and popularity were nearly the

38
same, which indicate that the result is reliable. This research also suggests that the account has a

strong influence on engagement rate, although it was just used as a control variable.

Limitations and directions for future research

This research design has the advantages of reliability and reproducibility. However, it does have

some limitations. For example, the five brands may not be able to represent all the fashion

industries. Different brands may have different situations. Furthermore, 20 observations for each

brand may not be enough. Future research should study a larger sample and find more

measurable video characteristics. In addition, while the number of hashtags, the number of

emojis, and the length of text in the descriptions did not have a significant impact, future

research could further discuss their negative/positive impact. Finally, future research could also

focus more on what kind of video content and how they interact with other social media

platforms can more easily increase brand awareness and brand recognition.

Managerial implications

From a management perspective, this research has several important implications, particularly in

terms of social media marketing. Marketers should still work on the characteristics of the videos

and assess which features could be improved. The results of this study showed that videos

uploaded on a weekday and with a longer duration are more likely to receive a higher

engagement rate on TikTok. However, the number of hashtags, the number of emojis, and the

text length in the description have no impact on the engagement rate on TikTok. Furthermore,

39
companies should recognize the importance of brand image and brand popularity and the impact

they can have on engagement rate.

Literature contributions

Hashtags

The outcomes of this study have shown that there is no relationship between the number of

hashtags and engagement rate (H1). This finding contrasts with previous research, which

indicated that more hashtags distract viewers from the main message of the post (Cadena, 2020).

However, the fact is that the number of hashtags does not affect the engagement rate. Some users

who do not use hashtags to search for information may not care about the use of hashtags. The

hashtags usually appear at the end of the description. Sometimes there are so many words in the

description that the only way to see the hashtags is to click on "See more", so many users may

ignore the hashtags.

Weekday

One of the main goals of this study was to find out if there was a relationship between uploading

on weekdays and the engagement rate. Prior research has shown that there is no such relationship

on Facebook. However, it does have an impact on YouTube. The results of this study showed

that there was no correlation between weekday and engagement rate. The result changed when

other influencing factors were considered. The relationship between weekday and the

engagement rate became significant in the regression model. As there are always many factors

influence social media engagement rate, it is unrealistic to consider the effect of a particular

40
factor on engagement rate in isolation, which means the regression models are more accurate.

Therefore, there is a positive relationship between weekday and the engagement rate (H2).

Duration

Duration is always an important factor in engagement rate, and the effect of duration on

engagement rate varies across platforms. For example, on YouTube, medium-length videos are

more supportive (Munaro et al., 2021); on TikTok, the situation is different due to time

constraints, and longer videos uploaded by government TikTok accounts will receive a higher

number of likes and comments (Chen et al., 2021). However, this cannot be applied to the

fashion industry. The results of this study suggest a positive relationship between duration and

engagement rate, which is contrary to the proposed hypothesis (H3).

Emoji

Emoji is also an important factor that video makers will consider. Prior research has shown that

the presence of emojis will increase brand engagement on Twitter. In the correlation test, the

result showed that the number of emojis and the engagement rate did correlate. However, the

result changed when other influencing factors were considered. The relationship between the

number of emojis and the engagement rate disappeared. As there are many factors that influence

social media engagement rate, it is unrealistic to consider the effect of a particular factor on

engagement rate in isolation, which means that the regression models are more accurate.

Therefore, there is no significant relationship between the number of emojis and the engagement

rate (H4).

41
Text length

Lastly, as the independent variable, text length is expected to have a negative relationship with

the engagement rate. Chen et al. (2021) have already explored this using government TikTok

accounts to improve health information dissemination. However, this cannot be applied in the

fashion industry. There is no relationship between text length in the description and the

engagement rate (H5). That is, whether the description is long or short, it does not affect the

engagement rate.

42
Conclusions

Brands are always thinking ways to promote their products. Using social CRM strategy can help

them to make business thrive. But how can this be achieved? This research aims to improve the

TikTok videos by focusing on the video characteristics. It examined the effect of video

characteristics on the engagement rate on the TikTok platform. Prior literature showed that

different characteristics may have an impact on the engagement rate. However, almost all these

characteristics were examined on other social media platforms (e.g., Twitter), with some of them

coming from text posts rather than video posts. Therefore, this study aims to fill this research gap

by running multiple regressions on the video characteristics to see how they affect engagement

rate on TikTok. It was expected that all these variables may have a certain level of impact on the

engagement rate. However, it was found that this impact is rather small. The video

characteristics can only explain 7.5% of the engagement rate. In addition, only two out of five

independent variables, weekday and duration, have a significant impact on the engagement rate.

This indicates that videos uploaded on weekdays and with a longer duration will have a higher

engagement rate. The number of hashtags, the number of emojis, and the text length in

description do not influence the engagement rate on TikTok. Interestingly, it was found that the

accounts can explain around 40% of the engagement rate, which means that people will take

action after viewing the whole video depending on whether they know the brand or not. This

suggests that the engagement rate is significantly influenced by its brand reputation and

popularity. The reason for this may be found on the TikTok platform. Since a consumer’s brand

knowledge positively influences their participation in social media brand engagement activities

(Osei-Frimpong et al., 2019) and TikTok is a new platform, users may first follow or watch

videos from well-known and popular brands. This can explain why SHEIN is so popular on

43
TikTok. Because SHEIN is currently very popular and receiving a lot of attention among young

people. Managers, therefore, need to focus not only on video characteristics but also on how to

increase brand awareness and work with each other on different social media platforms.

44
Reference

24 Inspiring Brands on TikTok to Fuel Your Creative Strategy. (2020). Wallaroo Media. https://

wallaroomedia.com/blog/social-media/tiktok-brands-examples-strategy/

Algesheimer, R., Dholakia, U. M., and Herrmann, A. (2005). The Social Influence of Brand

Community: Evidence from European Car Clubs. Journal of Marketing, 69(3), 19–34.

https://doi.org/10.1509/jmkg.69.3.19.66363

Avnet, T., & Higgins, E. T. (2006). How Regulatory Fit Affects Value in Consumer Choices and

Opinions. Journal of Marketing Research, 43(1), 1–10.

https://doi.org/10.1509/jmkr.43.1.1

Bai, Q., Dan, Q., Mu, Z., and Yang, M. (2019). A Systematic Review of Emoji: Current

Research and Future Perspectives. Frontiers in Psychology, 10.

https://doi.org/10.3389/fpsyg.2019.02221

Bitiktas, F., and Tuna, O. (2020). Social media usage in container shipping companies: Analysis

of Facebook messages. Research in Transportation Business and Management, 34,

100454. https://doi.org/10.1016/j.rtbm.2020.100454

45
Bode, L., Hanna, A., Yang, J., and Shah, D. V. (2015). Candidate Networks, Citizen Clusters,

and Political Expression. The ANNALS of the American Academy of Political and

Social Science, 659(1), 149–165. https://doi.org/10.1177/0002716214563923

Bonsón, E., and Ratkai, M. (2013). A set of metrics to assess stakeholder engagement and social

legitimacy on a corporate Facebook page. Online Information Review, 37(5), 787–803.

https://doi.org/10.1108/oir-03-2012-0054

Bonsón, E., Royo, S., and Ratkai, M. (2016). Facebook Practices in Western European

Municipalities. Administration and Society, 49(3), 320–347.

https://doi.org/10.1177/0095399714544945

Brodie, R. J., Ilic, A., Juric, B., & Hollebeek, L. (2013). Consumer engagement in a virtual brand

community: An exploratory analysis. Journal of Business Research, 66(1), 105–114.

https://doi.org/10.1016/j.jbusres.2011.07.029

Bucknell Bossen, C., and Kottasz, R. (2020). Uses and gratifications sought by pre-adolescent

and adolescent TikTok consumers. Young Consumers, 21(4), 463–478.

https://doi.org/10.1108/ yc-07-2020-1186

Cadena, K. (2020, May 11). Hashtags: How Many is Too Many? Bell + Ivy.

https://www.bellivy.com/blog/how-many-hashtags

46
Calder, B. J., Malthouse, E. C., and Schaedel, U. (2009). An Experimental Study of the

Relationship between Online Engagement and Advertising Effectiveness. Journal of

Interactive Marketing, 23(4), 321–331. https://doi.org/10.1016/j.intmar.2009.07.002

Chan, I. C. C., Fong, D. K. C., Law, R., & Fong, L. H. N. (2018). State-of-the-art social

customer relationship management. Asia Pacific Journal of Tourism Research, 23(5),

423–436. https://doi.org/10.1080/10941665.2018.1466813

Chen, I. J., and Popovich, K. (2003). Understanding customer relationship management (CRM).

Business Process Management Journal, 9(5), 672–688. https://doi.org/

10.1108/14637150310496758

Chen, Q., Min, C., Zhang, W., Ma, X., and Evans, R. (2021). Factors Driving Citizen

Engagement With Government TikTok Accounts During the COVID-19 Pandemic:

Model Development and Analysis. Journal of Medical Internet Research, 23(2),

e21463. https://doi.org/10.2196/21463

Choy, K., Lee, W., and Lo, V. (2002). Development of a case based intelligent customer–

supplier relationship management system. Expert Systems with Applications, 23(3),

281–297. https://doi.org/10.1016/s0957-4174(02)00048-9

47
Clauss, T., Niemand, T., Kraus, S., Schnetzer, P., and Brem, A. (2019). Increasing crowdfunding

success through social media: The importance of reach and utilisation in reward-based

crowdfunding. International Journal of Innovation Management, 24(03), 2050026.

https://doi.org/10.1142/s1363919620500267

Dessart, L. (2017). Social media engagement: a model of antecedents and relational outcomes.

Journal of Marketing Management, 1–25.

https://doi.org/10.1080/0267257x.2017.1302975

Geuens, M., de Pelsmacker, P., & Faseur, T. (2011c). Emotional advertising: Revisiting the role

of product category. Journal of Business Research, 64(4), 418–426.

https://doi.org/10.1016/j.jbusres.2010.03.001

Geyser, W. (2021). TikTok Statistics – Revenue, Users and Engagement Stats (2021). Influencer

Marketing Hub. https://influencermarketinghub.com/tiktok-stats/#toc-3

Greenberg, P. (2010). The impact of CRM 2.0 on customer insight. Journal of Business and

Industrial Marketing, 25(6), 410–419. https://doi.org/10.1108/08858621011066008

Hautz, J., Füller, J., Hutter, K., & Thürridl, C. (2014c). Let Users Generate Your Video Ads?

The Impact of Video Source and Quality on Consumers’ Perceptions and Intended

48
 Behaviors. Journal of Interactive Marketing, 28(1), 1–15.

https://doi.org/10.1016/j.intmar.2013.06.003

Hoiles, W., Aprem, A., & Krishnamurthy, V. (2017). Engagement and Popularity Dynamics of

YouTube Videos and Sensitivity to Meta-Data. IEEE Transactions on Knowledge and

Data Engineering, 29(7), 1426–1437. https://doi.org/10.1109/tkde.2017.2682858

Hollebeek, L. D. (2011). Demystifying customer brand engagement: Exploring the loyalty nexus.

Journal of Marketing Management, 27(7–8), 785–807.

https://doi.org/10.1080/0267257x.2010.500132

Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and

opportunities of Social Media. Business Horizons, 53(1), 59–68.

https://doi.org/10.1016/j.bushor.2009.09.003

Khan, M. L. (2017). Social media engagement: What motivates user participation and

consumption on YouTube? Computers in Human Behavior, 66, 236–247.

https://doi.org/10.1016/j.chb.2016.09.024

Landsberg, N. (2021). Best Times to Post on TikTok (+6 Tips to Grow Your TikTok Following).

Influencer Marketing Hub. https://influencermarketinghub.com/best-times-to-post-on-

tiktok/

49
Leary, B. (2008), “Social CRM: customer relationship management in the age of the socially-

empowered customer”, working paper, Sage CRM Solutions, Scottsdale, AZ, available

at: www.box.net/shared/static/adqb9k3pqy.pdf (accessed November 11, 2008).

Levine, L. E., Waite, B. M., and Bowman, L. L. (2007). Electronic Media Use, Reading, and

Academic Distractibility in College Youth. CyberPsychology & Behavior, 10(4), 560–

566. https://doi.org/10.1089/cpb.2007.9990

Li, X., Shi, M., & Wang, X. S. (2019). Video mining: Measuring visual information using

automatic methods. International Journal of Research in Marketing, 36(2), 216–231.

https://doi.org/10.1016/j.ijresmar.2019.02.004

Liikkanen, L. A., and Salovaara, A. (2015). Music on YouTube: User engagement with

traditional, user-appropriated and derivative videos. Computers in Human Behavior,

50, 108–124. https://doi.org/10.1016/j.chb.2015.01.067

Liu, J., Lu, C., and Lu, S. (2021). Research on the Influencing Factors of Audience Popularity

Level of COVID-19 Videos during the COVID-19 Pandemic. Healthcare, 9(9), 1159.

https://doi.org/10.3390/healthcare9091159

50
Lovejoy, K., Waters, R. D., and Saxton, G. D. (2012). Engaging stakeholders through Twitter:

How nonprofit organizations are getting more out of 140 characters or less. Public

Relations Review, 38(2), 313–318. https://doi.org/10.1016/j.pubrev.2012.01.005

McCormick, P. (2021). Shein: The TikTok of Ecommerce. Not Boring by Packy McCormick.

https:// www.notboring.co/p/shein-the-tiktok-of-ecommerce

McShane, L., Pancer, E., Poole, M., and Deng, Q. (2021). Emoji, Playfulness, and Brand

Engagement on Twitter. Journal of Interactive Marketing, 53, 96–110.

https://doi.org/10.1016/j.intmar.2020.06.002

Munaro, A. C., Barcelos, R., Francisco Maffezzolli, E. C., Rodrigues, J. P., and Paraiso, E.

(2021). To engage or not engage? The features of video content on YouTube affecting

digital consumer engagement. Journal of Consumer Behaviour, 20(5), 1336–1352.

https://doi.org/10.1002/cb.1939

Osei-Frimpong, K., McLean, G., and Famiyeh, S. (2019). Social media brand engagement

practices. Information Technology and People, 33(4), 1235–1254.

https://doi.org/10.1108/itp-05-2018-0220

Paek, H. J., Hove, T., Jung, Y., and Cole, R. T. (2013). Engagement across three social media

platforms: An exploratory study of a cause-related PR campaign. Public Relations

Review, 39(5), 526–533. https://doi.org/10.1016/j.pubrev.2013.09.013

51
Parisi, D. (2020). Less aspirational, less produced: Luxury brands are (finally) making headway

on TikTok. Glossy. https://www.glossy.co/fashion/luxury-brands-are-finally-making-

headway-on-tiktok/

Pauwels, K., Aksehirli, Z., and Lackman, A. (2016). Like the ad or the brand? Marketing

stimulates different electronic word-of-mouth content to drive online and offline

performance. International Journal of Research in Marketing, 33(3), 639–655.

https://doi.org/10.1016/j.ijresmar.2016.01.005

Pham, M. T., & Avnet, T. (2009). Rethinking Regulatory Engagement Theory. Journal of

Consumer Psychology, 19(2), 115–123. https://doi.org/10.1016/j.jcps.2009.02.003

Sashi, C. (2012). Customer engagement, buyer‐seller relationships, and social media.

Management Decision, 50(2), 253–272. https://doi.org/10.1108/00251741211203551

Schau, H. J., Muñiz, A. M., & Arnould, E. J. (2009). How Brand Community Practices Create

Value. Journal of Marketing, 73(5), 30–51. https://doi.org/10.1509/jmkg.73.5.30

Schwedel, H. (2018). A Guide to TikTok for Anyone Who Isn’t a Teen. Slate Magazine. https://

slate.com/technology/2018/09/tiktok-app-musically-guide.html

52
Schwenzow, J., Hartmann, J., Schikowsky, A., and Heitmann, M. (2021). Understanding videos

at scale: How to extract insights for business research. Journal of Business Research,

123, 367–379. https://doi.org/10.1016/j.jbusres.2020.09.059

Shahzad, B., Awan, K. M., Abdullatif, A. M., Ikramullah Lali, M., Saqib Nawaz, M., Ayesha,

U., and Khan, M. (2019). Quantification of Productivity of the Brands on Social Media

With Respect to Their Responsiveness. IEEE Access, 7, 9531–9539.

https://doi.org/10.1109/ access.2019.2891081

Shin, J., Chae, H., & Ko, E. (2018). The power of e-WOM using the hashtag: focusing on SNS

advertising of SPA brands. International Journal of Advertising, 37(1), 71–85.

https://doi.org/10.1080/02650487.2017.1401519

Soares, J. C., Limongi, R., and Cohen, E. D. (2021). Engagement in a social media: an analysis

in higher education institutions. Online Information Review, ahead-of(ahead-of-print).

https://doi.org/10.1108/oir-06-2020-0242

Stark, L., & Crawford, K. (2015). The Conservatism of Emoji: Work, Affect, and

Communication. Social Media and Society, 1(2), 205630511560485.

https://doi.org/10.1177/2056305115604853

53
Statista. (2021, October 13). TikTok global monthly active users 2018–2021.

https://www.statista.com/statistics/1267892/tiktok-global-

mau/# %7E:text=TikTok%20global- %20monthly%20active%20users%202018%2D2

021&text=In%20September%202021%2C%20social%20video,the%20biggest%20soci

al%20networks%20worldwide.

Tellis, G. J., MacInnis, D. J., Tirunillai, S., & Zhang, Y. (2019). What Drives Virality (Sharing)

of Online Digital Content? The Critical Role of Information, Emotion, and Brand

Prominence. Journal of Marketing, 83(4), 1–20.

https://doi.org/10.1177/0022242919841034

TikTok Statistics - Everything You Need to Know [Sep 2021 Update]. (2021, September 27).

Wallaroo Media. https://wallaroomedia.com/blog/social-media/tiktok-statistics/

Trainor, K. J. (2012). Relating Social Media Technologies to Performance: A Capabilities-Based

Perspective. Journal of Personal Selling and Sales Management, 32(3), 317–331.

https:// doi.org/10.2753/pss0885-3134320303

Trzcinski, T., and Rokita, P. (2017). Predicting Popularity of Online Videos Using Support

Vector Regression. IEEE Transactions on Multimedia, 19(11), 2561–2570.

https://doi.org/10.1109/tmm.2017.2695439

54
Wang, S., and Ji, Q. (2015). Video Affective Content Analysis: A Survey of State-of-the-Art

Methods. IEEE Transactions on Affective Computing, 6(4), 410–430.

https://doi.org/10.1109/taffc.2015.2432791

Wang, Y. (2020). Humor and camera view on mobile short-form video apps influence user

experience and technology-adoption intent, an example of TikTok (DouYin).

Computers in Human Behavior, 110, 106373.

https://doi.org/10.1016/j.chb.2020.106373

Yılmaz, H., and Enginkaya, E. (2015). Brand followers: motivations and attitudes of consumers

to follow brands in social media. International Journal of Internet Marketing and

Advertising, 9(1), 3. https://doi.org/10.1504/ijima.2015.068356

Zaglia, M. E. (2013). Brand communities embedded in social networks. Journal of Business

Research, 66(2), 216–223. https://doi.org/10.1016/j.jbusres.2012.07.015

Zhang, M., and Luo, N. (2016). Understanding relationship benefits from harmonious brand

community on social media. Internet Research, 26(4), 809–826.

https://doi.org/10.1108/ intr-05-2015-0149

55
Appendices

Appendix 1: List of Calvin Klein video characteristics

Video Hashtag Date Duration Emoji Text View Like Comment Share Engagement
Length
1 1 Weekday 8 0 14 4,503 169 37 3 0.0464

2 1 Weekend 14 0 16 3,728 115 4 5 0.0333

3 1 Weekday 16 0 11 9,233 311 15 4 0.0357

4 1 Weekday 15 0 13 5,948 213 5 6 0.0377

5 1 Weekday 17 0 7 5,999 213 2 3 0.0363

6 1 Weekday 7 0 19 2,645 132 3 4 0.0526

7 1 Weekday 17 0 8 4,412 188 5 1 0.0440

8 1 Weekday 9 0 6 3,945 187 10 4 0.0510

9 1 Weekend 21 0 16 3,971 128 4 10 0.0358

10 1 Weekday 9 0 11 11,600 530 19 4 0.0477

11 1 Weekday 16 0 9 4,965 145 2 2 0.0300

12 1 Weekday 32 0 16 2,847 173 2 7 0.0639

13 1 Weekend 20 0 19 5,419 130 9 5 0.0266

14 1 Weekday 12 0 7 9,117 236 6 2 0.0268

15 1 Weekday 25 0 11 8,079 161 4 1 0.0205

16 1 Weekday 15 0 16 3,311 166 4 14 0.0556

17 1 Weekend 13 0 18 3,996 137 3 2 0.0355

18 1 Weekday 24 0 9 4,667 215 1 4 0.0471

19 1 Weekday 24 0 16 3,529 181 2 5 0.0533

20 1 Weekend 27 0 16 4,259 102 2 2 0.0249

56
Appendix 2: List of Levi’s video characteristics

Video Hashtag Date Duration Emoji Text View Like Comment Share Engagement
Length
1 2 Weekday 17 0 18 10,800 264 7 0 0.0251

2 2 Weekday 9 2 16 6,971 313 5 7 0.0466

3 2 Weekday 22 1 16 1,600,000 22,900 83 51 0.0144

4 1 Weekend 10 1 12 2,600,000 46,200 62 39 0.0178

5 2 Weekday 36 0 18 1,300,000 91,900 610 345 0.0714

6 2 Weekday 23 0 12 3,600,000 104,700 478 254 0.0293

7 1 Weekday 14 0 3 8,452 590 18 8 0.0729

8 3 Weekday 10 0 17 3,200,000 95,600 268 89 0.0300

9 3 Weekend 26 0 11 248,000 3,743 33 23 0.0153

10 4 Weekday 31 2 10 7,956 284 3 10 0.0373

11 3 Weekday 23 3 18 2,600,000 87,200 206 120 0.0337

12 2 Weekday 29 1 14 2,500,000 54,900 59 75 0.0220

13 1 Weekday 43 1 6 1,200,000 97,900 490 302 0.0822

14 3 Weekday 24 3 16 10,900 965 55 9 0.0944

15 4 Weekday 36 0 22 16,000 776 14 21 0.0507

16 3 Weekday 39 0 17 421,200 34,500 202 107 0.0826

17 3 Weekday 10 2 10 7,675 481 7 8 0.0646

18 2 Weekday 37 1 17 4,100,000 132,800 420 273 0.0326

19 2 Weekday 35 1 20 5,200,000 158,100 307 172 0.0305

20 3 Weekday 31 1 8 2,900,000 30,300 101 131 0.0105

57
Appendix 3: List of Louis Vuitton video characteristics

Video Hashtag Date Duration Emoji Text View Like Comment Share Engagement
Length
1 2 Weekday 12 0 5 3,100,000 34,300 189 140 0.0112

2 2 Weekday 6 0 2 2,500,000 35,200 382 109 0.0143

3 2 Weekend 15 0 2 5,900,000 89,500 550 245 0.0153

4 1 Weekend 14 0 1 5,000,000 79,300 418 244 0.0160

5 1 Weekday 12 0 1 247,900 7,901 150 84 0.0328

6 2 Weekday 15 0 2 1,000,000 47,200 299 353 0.0479

7 2 Weekday 15 0 2 124,100 15,100 104 111 0.1234

8 2 Weekday 15 0 2 4,600,000 80,400 955 260 0.0177

9 2 Weekday 14 0 2 4,900,000 124,700 682 238 0.0256

10 2 Weekend 12 0 2 1,100,000 13,300 139 64 0.0123

11 1 Weekday 12 0 4 796,100 60,700 1,475 1,327 0.0798

12 1 Weekday 15 0 1 3,900,000 145,700 874 366 0.0377

13 2 Weekday 12 0 2 5,400,000 67,800 571 201 0.0127

14 2 Weekday 13 0 3 270,800 14,700 257 96 0.0556

15 2 Weekday 13 0 2 332,300 11,400 68 129 0.0349

16 2 Weekday 15 0 2 1,000,000 41,500 651 137 0.0423

17 2 Weekday 15 0 2 124,500 17,000 294 84 0.1396

18 2 Weekday 13 0 3 5,600,000 46,600 934 390 0.0086

19 2 Weekday 14 0 7 47,000,000 1,300,000 7,820 2,466 0.0279

20 1 Weekday 12 0 1 3,500,000 46,700 350 142 0.0135

58
Appendix 4: List of Nike video characteristics

Video Hashtag Date Duration Emoji Text View Like Comment Share Engagement
Length
1 1 Weekday 32 0 14 30,500 2,045 45 40 0.0698

2 1 Weekday 29 1 11 26,500 1,827 33 20 0.0709

3 2 Weekday 31 2 17 31,500 1,919 38 11 0.0625

4 1 Weekday 33 1 14 29,200 1,602 37 11 0.0565

5 0 Weekday 30 0 24 24,800 1,352 16 6 0.0554

6 1 Weekday 31 2 14 26,700 1,716 30 8 0.0657

7 1 Weekday 30 0 10 34,600 2,213 34 12 0.0653

8 2 Weekday 32 0 15 35,700 2,657 35 40 0.0765

9 2 Weekday 30 0 11 33,900 1,954 80 19 0.0606

10 1 Weekday 22 0 11 45,900 3,279 30 20 0.0725

11 1 Weekday 40 0 16 40,600 2,895 85 21 0.0739

12 0 Weekday 30 0 14 29,400 1,607 40 19 0.0567

13 2 Weekday 20 0 15 36,700 2,821 42 27 0.0787

14 1 Weekday 30 1 13 34,500 2,361 32 18 0.0699

15 0 Weekday 30 1 27 25,500 1,477 24 15 0.0595

16 1 Weekday 32 1 17 39,100 1,928 96 20 0.0523

17 1 Weekday 30 1 9 31,300 1,954 28 13 0.0637

18 1 Weekday 31 1 11 125,000 6,752 109 424 0.0583

19 1 Weekday 22 1 17 28,400 1,568 26 3 0.0562

20 1 Weekday 23 1 13 51,100 3,929 36 12 0.0778

59
Appendix 5: List of SHEIN video characteristics

Video Hashtag Date Duration Text Emoji View Like Comment Share Engagement
Length
1 4 Weekday 21 12 1 85,800 6,505 239 36 0.0790

2 3 Weekday 20 9 1 75,500 4,387 243 24 0.0616

3 0 Weekday 29 5 1 128,700 17,500 243 77 0.1385

4 3 Weekday 12 6 1 89,800 9,360 269 60 0.1079

5 0 Weekend 32 10 1 91,000 8,402 466 43 0.0979

6 4 Weekend 32 13 0 965,200 76,900 1,347 473 0.0816

7 1 Weekday 30 24 0 66,100 2,719 190 18 0.0443

8 0 Weekday 34 6 1 1,200,000 178,100 980 258 0.1494

9 3 Weekend 20 4 1 66,700 5,095 58 21 0.0776

10 0 Weekday 27 5 1 325,600 41,100 1,156 447 0.1312

11 0 Weekday 9 7 2 91,300 4,685 398 12 0.0558

12 0 Weekday 26 10 1 464,300 83,000 526 266 0.1805

13 4 Weekday 27 8 1 60,700 2,900 752 29 0.0606

14 0 Weekday 50 7 1 1,700,000 196,600 2,500 741 0.1176

15 2 Weekday 11 6 2 178,700 17,200 312 31 0.0982

16 0 Weekday 18 5 1 78,200 6,325 219 25 0.0840

17 0 Weekday 19 6 1 71,200 3,737 132 38 0.0549

18 4 Weekday 10 11 1 101,100 8,363 168 7 0.0845

19 4 Weekend 25 9 1 98,600 8,458 517 46 0.0915

20 0 Weekday 24 9 1 89,600 7,190 245 4 0.0830

60