How influencers’ social media posts have an influence on audience engagement among young consumers

Abstract

Purpose

Online influencers are increasingly used by brands around the globe to establish brand communication. This study aims to investigate the characteristics of social media content in terms of presentation style and brand communication among online influencers in China. The authors identified how characteristics of social media posts influence young consumers’ engagement with the posts.

Design/methodology/approach

The authors analyzed 1,779 posts from the Sina Weibo accounts of ten top-ranked online influencers by combining traditional content analysis with Web data crawling of audience engagement with social media posts.

Findings

Online influencers in China more frequently used photos than videos to communicate with their social media audience. Altogether 8% and 6% of posts carried information about promotion and event, respectively. Posts with promotional incentives as well as event information were more likely to engage audiences. Altogether 22% of the sampled social media posts mentioned brands. Posts with brand information, however, were less likely to engage audiences. Furthermore, having long text is more effective than photos/images in generating likes from social media audiences.

Originality/value

Combining content analysis of social media posts and engagement analytics obtained via Web data crawling, this study is, to the best of the authors’ knowledge, one of the first empirical studies to analyze influencer marketing and young consumers’ reactions to social media in China.

Citation

Fan, F., Chan, K., Wang, Y., Li, Y. and Prieler, M. (2023), "How influencers’ social media posts have an influence on audience engagement among young consumers", Young Consumers, Vol. 24 No. 4, pp. 427-444. https://doi.org/10.1108/YC-08-2022-1588

Introduction

Social media is omnipresent worldwide, and user numbers continue to grow. In the USA, 81% of the population reported using YouTube and 69% reported using Facebook, in 2021, with the younger population having even higher numbers (Pew Research Center, 2021). Similarly, China had more than 930 million (64%) social media users in 2021, which is 110 million more than the previous year (Kemp, 2021). The increase in the number of social media users is associated with the importance of influencer marketing. Influencers regularly create and disseminate content on social media relevant to their supporters and have thus reached many followers (Lou et al., 2019; Ye et al., 2021). The influencer market in China is the largest social media market in the world (Cervi, 2020). As the main users of social media, youngsters in China frequently follow online influencers and their social media input. Take the video platform Bilibili for example. As of April 2021, over 80% of active Bilibili users who followed top online influencers in China were youth aged 24 or below (Statista, 2022a).

Born in 1995 or after, Gen Z comprises digital natives who are heavily involved with technology (Sun et al., 2021) and a target group for influencer marketing (Lin et al., 2019). Because of their widespread use of platforms such as YouTube and Instagram, Gen Z consumers are influenced by key opinion leaders on social media in their purchase decisions (Król and Zdonek, 2020; Parker and Igielnik, 2020). Although Gen Z makes up 17% of China’s total population, they have the potential to dominate the consumer market (Statista, 2022b). The influence of opinion leaders, peers and celebrities on the purchase intention of luxury fashion is significant among Chinese Gen Z (Sun et al., 2021). A survey among young consumers in China found that respondents used Douyin (a popular social media platform) mainly for entertainment purpose. However, those who used the media for social motives demonstrated high parasocial relationships with the influencers and were more likely to purchase the products influencers endorsed (Yang and Ha, 2021).

Previous research has indicated that influencers are considered more similar to the audience, trustworthy and more effective than traditional celebrities (Schouten et al., 2020). A total of 76% of Chinese people follow at least one social media influencer (Thomala, 2021), and 44% of Gen Z and 26% of the general population reported making purchases based on influencer recommendations (Williams, 2020). COVID-19 may have strengthened the trend toward social media usage and influencer marketing (Taylor, 2020). Thus, it is not surprising that companies increasingly value influencers (Schouten et al., 2020). Among brand representatives and marketing agencies, 90% believe that influencer marketing is effective, 75% intend to dedicate a budget to influencers in 2021 (59% having a standalone budget) and 62% plan to increase the budget spent on influencers (Influencer Marketing Hub, 2021).

The number of academic articles on influencer marketing has been increasing, particularly since 2018 (Hudders et al., 2021; Vrontis et al., 2021). However, most of these studies adopt survey methodology or experimental design to measure perceived attributes of the online influencers and their impact on the self-reported parasocial relationship or purchase intention, while research on media content posted by online influencers is scant (Sundermann and Raabe, 2019; Ye et al., 2021). There is a lack of studies that measure the characteristics of social media content and relate them to their impact on objective measurements of audience engagement. Studying media content related to influencers would be a promising path because audience engagement can be measured objectively using likes, shares and comments. Crawling of actual engagement data provides the opportunity to better understand influencer marketing; for example, how different content factors and content strategies impact reactions, and engagement of target groups (Hudders et al., 2021; Sundermann and Raabe, 2019; Taylor, 2020; Ye et al., 2021).

The current study was contextualized in the source credibility model (Goldsmith et al., 2000). Credibility refers to “the extent to which the source is perceived as possessing expertise relevant to the communication topic and can be trusted to give an objective opinion on the subject” (Goldsmith et al., 2000, p. 43). Credibility was considered to be a major factor in making online influencers credible as the posts shared by them were supposed to be authentic and non-commercial (Uzunoglu and Kip, 2014).

We have focused on two under-researched aspects in this study. First, we will investigate the proportion of influencer posts with brand communication and its relationship with audience engagement. A study on fitness influencers on Instagram (Pilgrim and Bohnet-Joschko, 2019) showed that 71% of posts include at least one brand, out of which 90% are mentioned by name and provide direct links. While influencers in China are often considered to promote a high number of products, only a few studies provide actual numbers. There is a delicate balance because posting too much commercial content can harm the authenticity of the influencer (Hudders et al., 2021). Second, we will analyze which content characteristics are predictors of audience engagement.

We aim to answer the following research questions:

RQ1.

How many percent of the online influencers’ posts are related to advertising or carrying brand information? What are the characteristics of influencer advertising-related posts?

RQ2.

What factors influence young consumers’ engagement with online influencers on social media?

RQ3.

Does brand communication in social media posts discourage young consumers’ engagement?

The originality of the current study is the introduction of an innovative methodology of combining traditional content analysis with data analysis of messages on online social media platforms obtained via Web data crawling.

Literature review

Influencer marketing in the global context

Influencer marketing, as one of the most important marketing strategies to lead consumers’ purchasing decisions, is established on the success of e-word of mouth and relationship marketing (Brown and Hayes, 2008). After 2018, online influencer marketing-related studies grew increasingly, including several literature reviews (Chen and Chua, 2020; De Veirman et al., 2019; Hudders et al., 2021; Sundermann and Raabe, 2019; Vrontis et al., 2021; Ye et al., 2021). Nevertheless, Taylor (2020) argues that additional research is urgently needed, since influencer marketing, particularly influencer advertising, is a rare case where industry needs and academic interests intersect, yet several areas are still under-researched.

Specifically, the popularity of online influencers has attracted the attention of marketing communication practitioners (Ibáñez-Sánchez et al., 2021). Increasingly, brands invest marketing dollars in online influencers and invite them to endorse their products. Brands expect online influencer endorsement to generate brand awareness, liking and purchase intention (Jiménez-Castillo and Sánchez-Fernández, 2019). Faced with this trend of influencer advertising, brands are eager to know how to select qualified online influencers (Ye et al., 2021). First, the popularity of online influencers is a criterion for identifying online influencers to be used in advertising strategy. An important indicator of the popularity of influencers is the number of followers each online influencer owns (De Veirman et al., 2017). Based on the fan base, Berne-Manero and Marzo-Navarro’s (2020) study divided online influencers into four types. They are mega-influencer with more than 1,000,000 followers, macro-influencer with 100,000–1,000,000 followers, micro-influencer with 1,000 and 100,000 followers and nano-influencer with less than 1,000 followers. Second, online influencers’ characteristics should be considered (Vollenbroek et al., 2014). For example, an active mind (Keller and Berry, 2003), authority (Cialdini, 2006) and credibility (Vollenbroek et al., 2014) are important indicators of the potential influence of online influencers. Third, the interaction between online influencers and their social media audiences is also a major factor in determining the selection of product/brand endorsers. According to Hass (1981), interaction contributes to a person’s communication effectiveness. In social media, engagement indicators such as the number of shares and comments are used to measure the interaction between online influencers and their audiences (Vollenbroek et al., 2014). Fourth, selecting online influencers as product/brand endorsers requires consideration of the para-social relationship between online influencers and target consumers (Fan, 2021). In other words, the perceived closeness of target consumers to product/brand endorsers influences the effectiveness of influencer advertising. If consumers are attached emotionally to product/brand endorsers, they are more likely to accept endorsement messages (Gong and Li, 2017).

Although current studies have identified several factors securing the selection quality of online influencers as brand/product endorsers in the marketing strategy, few of them considered how social media content related to influencers contributes to brand/product endorsement. A survey of social media followers found that subjective evaluation of content characteristics such as information quality plays a significant role in influencing consumers’ engagement (Cheung et al., 2022). This study identifies the need for examining content characteristics to fill the research gap in social media influencer marketing. Besides, the limited studies mainly use a dual-route model [e.g. argument quality and source credibility in Ong et al.’s (2022) study] to address this topic, and rarely narrow the focus to solely explore audience engagement in influencers’ social media contents. Moreover, recent studies such as Ong et al.’s (2022) study about Instagram influencer marketing and Xiao et al.’s (2018) study about YouTube influencer marketing mainly use survey methods to address this topic. Methodological innovation in influencer marketing studies is scant. To bridge the gap, our study proposed to combine traditional content analysis with data analysis of messages on online social media platforms obtained via Web data crawling to discuss audience engagement in influencers’ social media contents.

Online influencers and their prevalence in China

The rise of online influencers and influencer advertising rockets globally, with China no exception. Online influencers in China actively use diverse online communication platforms to engage consumers. Thomala’s study (2021) found that most Chinese people, around 80%, follow at least one online influencer. In China, some scholars perceive online influencers as a new type of celebrity (Chen et al., 2021a, 2021b; Fan, 2021; Hung, 2020). They develop their online personal images and charisma by updating social media feeds about their daily life, expertise and individual thoughts (Hung, 2020). They differ from traditional celebrities created by professionals in the entertainment industry (Hung, 2020). Online influencers are mostly self-made opinion leaders with strong popularity among social media users (Gräve, 2017; Schouten et al., 2020). In the literature of influencer studies, scholars currently use the following terms to name online influencers: social media influencers (Coates et al., 2019; Shan et al., 2020; Sundermann and Raabe, 2019), key opinion leaders (Zhu and Wang, 2020), bloggers (Balabanis and Chatzopoulou, 2019; Gannon and Prothero, 2018), vloggers (Ladhari et al., 2020; Lee and Watkins, 2016), internet celebrities (Chen et al., 2021a, 2021b; Roberts, 2010) micro-celebrities (Khamis et al., 2016; Martínez and Olsson, 2019) and wang hong, which means popular internet figure in the Chinese context (Cheng, 2018; Xu and Zhao, 2019).

In China, scholars have explored online influencer-related topics from the perspective of comparative study with celebrities (Fan, 2021), parasocial interaction between online influencers and audiences (Chen et al., 2021a, 2021b; Gong and Li, 2017; Shan et al., 2020) and the characteristics of online influencers (Chen et al., 2021a, 2021b; Fan, 2021; Gong and Li, 2017; Lou et al., 2019; Shan et al., 2020). However, Vrontis et al.’s (2021) systematic review of online influencer marketing-related studies found that although studies about online influencer marketing have been conducted in China, such studies on the USA are currently outnumbering research on China by a factor of 8:1. Therefore, further research deserves to be conducted on the largest consumer market in the world (Vrontis et al., 2021). This also justifies the importance of our research topic.

Influencer advertising and its persuasiveness

Studies have been conducted to measure the communication effects of influencer advertising around the globe. A market survey with 14,000 millennial respondents found that influencer advertising drives sales performance. Specifically, 30% of the survey participants claimed that they prefer purchasing products recommended by online influencers (Barker, 2017). Overall, current studies support the effectiveness of influencer advertising on consumer persuasion. One study on how YouTube influencers and their vlogs affect consumers’ perceptions of luxury brands showed that if online influencers review and recommend luxury brand products in vlogs, perception of and purchase intention for the advertised brands are significantly enhanced among consumers who watched these vlogs, compared with those who did not (Lee and Watkins, 2016). Another qualitative interview study found that online influencers could effectively motivate consumers to purchase the advertised products and spread word-of-mouth about them (Chapple and Cownie, 2017). However, not all online influencers are effective at influencer advertising. One experimental study of Instagram influencers with a large fan base found that endorsing a variety of product categories lowers the uniqueness of the advertised brands and dampens consumer attitudes toward them (De Veirman et al., 2017).

Four influencer marketing theories were proposed in the literature to explain influencer marketing, including the source credibility model, the source attractiveness model, the celebrity–brand fit model and the meaning transfer model (Chan and Fan, 2020). These four models identified key factors that affect impact audience response, including source credibility, source attractiveness and perceived congruency between the celebrity and the brands. However, all these identified factors are subjective qualities perceived by the audience. There is a research gap to examine how the contents presented by the celebrity/online influencers influence audience response. A content analysis of influencer advertising for 10 luxury brands found that product advertisement is the most commonly used type of influencer advertising, accounting for 33% of influencer advertisements (Zhu and Wang, 2020). Product advertisement with promotional incentives generates the highest number of shares among target consumers, whereas vlogs featuring promotional incentives generate the highest number of likes and comments (Zhu and Wang, 2020).

Second, the characteristics of online influencers play an essential role in the effectiveness of their endorsements. A survey among travel vlog viewers found that physical attractiveness, social attractiveness and perceived credibility of online influencers had a positive relation with audience identification with the influencer (Le and Hancer, 2021). An experimental study found that consumers perceive a high level of similarity to and trust in online influencers, and this perceived similarity and trust later add value to the effectiveness of online influencer endorsement (Schouten et al., 2020). One online survey conducted with 500 YouTube users in Taiwan found that the expertise of online influencers indirectly affects consumers’ impulse purchase behaviors for advertised products (Chen et al., 2021a, 2021b). In addition, congruence between influencers and consumers also influences advertising persuasiveness. A study about adults going online in China found that the fit between the images of online influencers and the self-images of adult consumers contributes to the effectiveness of influencer advertising (Shan et al., 2020). Another survey summarized that consumers’ attitudes toward advertising messages and purchase intention for advertised products are improved if the image of the online influencer is matched with the self-image of the consumer (Shan et al., 2020). Besides this, congruence between online influencers and advertised products significantly improves the attitude of consumers toward the products (Kim and Kim, 2021).

Third, advertising messages themselves also affect the persuasiveness of influencers. A survey found that social media content attributes of prestige and expertise and interaction strategies of interactivity and self-disclosure had a positive impact on parasocial relationship. Parasocial relationship had a positive impact on purchase intention (Aw et al., 2022). One experimental study showed that if young online influencers on Instagram disclose that their social media feeds are sponsored or contain advertising messages from brands, the level of advertising recognition and brand recall is higher than without clear disclosure (Boerman, 2020). Another experiment found that ambiguous disclosure of sponsorship in branded communication posts resulted in a lower level of perceived influencer transparency than an explicit disclosure. In the same study, influencer transparency was positively correlated with purchase intention (Woodroof et al., 2020). Ji et al.’s (2018) content analysis of Greenpeace China Weibo posts and audience engagement showed that the framing of social media posts significantly affects the number of likes, comments and shares they garner. Take Greenpeace China’s environmental messages for example. Posts framed with the notion of environmental responsibility generate a large number of likes and shares, while posts framed in a conflicting manner generate a higher number of comments (Ji et al., 2018). Another content analysis of ten health influencers’ posts about health information found that when influencers’ Sina Weibo posts feature efficacy-related information, they generate fewer likes from online audiences (Zou et al., 2021). Overall, current studies about message framing by online influencers mainly focus on one specific topic.

Studies in China have also addressed how the relationship between online influencers and audiences affects the persuasiveness of influencer advertising. Fan’s (2021) qualitative study found that compared with traditional celebrities, audiences in China perceive a more distant relationship with online influencers. Although the audience relationship with online influencers is not as close as that with traditional celebrities, the effectiveness of online influencers for product endorsement cannot be overlooked. An online survey conducted among 511 Chinese consumers found that the closeness of their relationship with online influencers was positively associated with their attitude toward advertising messages (Chen et al., 2021a, 2021b). Another online survey study with 500 Taiwanese consumers proved that the closeness of their attachment to online influencers was positively affected by the level of self-discourse influencers exhibit in front of their followers. If followers are attached to online influencers, their purchase intention and behaviors toward advertised products are enhanced (Chen et al., 2021a, 2021b).

To conclude, current studies have widely analyzed the effectiveness of influencer advertising from the perspectives of the influencer advertising type, influencers’ characteristics, message attributes and influencer–audience relationship. However, there are insufficient empirical studies that investigate the influence of content attributes and communication styles on audience engagement. As audience engagement is expected to have a positive impact on parasocial relations and purchase intention, the current study examines how objective measurements of influencers’ social media contents, including their formats and presentation styles, impact audience engagement.

Method

This study adopted a hybrid method of data mining and content analysis to examine social media posts of the top ten online influencers and consumer engagement. The combination of these two approaches enhances efficiency, systematization as well as objectivity (Lewis et al., 2013). This study aims to analyze the content of influencers’ social media posts from Sina Weibo. Weibo is China’s leading micro-blogging platform, with more than 500 million monthly active users (Luqiu and Yang, 2020; Sina Financial, 2021). As a microblogging service similar to Twitter and Facebook (Yang et al., 2012), its usage is more associated with Instagram, as online influencers use it for social and economic gain (Yan et al., 2018). Also, Sina Weibo is often used for content analysis research (Ji et al., 2018).

Sampling method

The current study adopts purposive sampling. The sample frame is the ranking list of the most popular online influencers in China co-published by eNet Research Center and the magazine China Internet Week. The top ten online influencer accounts were selected based on the number of fans on Sina Weibo as of April 2021. The number of fans ranged from 16.2 million to 41.6 million. A Web crawling application based on Python was used to collect the posts from these ten accounts on Sina Weibo. With the restriction from the Sina Weibo publisher, the application only allowed us to crawl around 1,000 Weibo posts per each online influencer’s account since its inception. The data was crawled on May 2, 2022. The ten online influencers had different activity levels. Jia Ma was the most active one and he posted 1,003 posts in the period between January 31, 2022 and May 2, 2022. On the other hand, Papi Jiang (Rank no. 2) and Ziqi Li (Rank no. 4) were the least active ones. Altogether 621 and 273 posts were crawled, respectively. A total of 86% of all the sampled posts were posted within one year before the crawling date. XPath Syntax was applied to extract Weibo texts, picture links, video links, length of posts and audience engagement with the posts expressed in terms of likes, comments and shares.

In considering the coding resources, a systematic sampling of about 200 posts per online influencer account was determined. In other words, each fifth post was sampled for coding. Each Weibo post formed the unit of analysis. The final sample consisted of 1,779 posts.

Coding procedure

This study applies manual content analysis. Ideally, the coding frame should be developed from the source credibility model. However, the variables in this model such as perceived trustworthiness and expertise cannot be coded using content analysis. A coding frame was therefore developed based on the research objectives and guided by the literature. To answer Research Question 1, we code the presence or absence of three types of promotional messages, either promoting a social cause or an activity related to a social cause, a sales promotion or a physical event/activity. To answer Research Question 2, we code the number of Chinese or foreign brands mentioned in the posts. For characteristics of the post, we code the presence or absence of pictures and videos in the posts. Information on the length of the post and audience engagement was obtained from data crawling. Appendix summarizes the variables coded and their operational definitions. Except for several Chinese and foreign brands, all variables were coded either 0 for absent or 1 for the present. To train the coders and get them familiar with the coding frame, two selected coders worked on a pilot of 20 posts by an online influencer not selected in the final coding. After that, one of the two coders coded all the posts, and 10% of the sample was coded by the other coder independently. The inter-coder reliability scores for all variables ranged from 0.83 to 0.91. Acceptable inter-coder reliability of Krippendorff’s alpha of 0.8 was reached for all the variables (Krippendorff, 2004).

Data analysis

Descriptive statistics of characteristics of social media posts, their advertising-related content and brand communication are compiled for the entire sample as well as for the individual online influencers. Chi-square statistics are computed to see if there are differences in the characteristics of posts and advertising content among the ten influencers. Multiple linear regression was conducted with each of the three audience engagement measures (i.e. like, share and comment) as dependent variables. The predictors were the characteristics of the posts, the presence or absence of promotional activities as well as presence or absence of brand communication.

Results

Descriptive statistics about the selected online influencers and their social media posts

The top ten online influencers on Sina Weibo were selected for our study. Their demographic profiles and the typical content of their posts are summarized in Table 1. Five of them are female online influencers. Jia Ma, the most popular online influencer, is a male online influencer with more than 41 million followers on Sina Weibo. Mr Deca’s mailbox, the least popular online influencer in our study, is a male online influencer with over 16 million followers. There is age and education information for six online influencers. For those with such information, most of them are aged 29–35 with at least a bachelor’s degree. The posts they shared varied in content. For example, online influencer No. 1 mainly shared his thoughts and feelings. The posts initiated by online influencer No. 3 comprised mostly of promotional information about cosmetic products and snacks. Online influencer No. 4 often shared content about traditional Chinese food and culture. Most of the online influencers adopted a humorous approach.

Table 1 also displays translations of the most liked posts for each online influencer. The Weibo post with the highest number of likes for No. 2, a female online influencer, was about a family issue. She is perceived as a feminist in China. She triggered a controversial discussion among the online audience with her decision to name her baby after the father’s family. She released a vlog on Sina Weibo to address this issue, which recorded the highest number of likes. The most-liked post by influencer No. 3 was a promotional message inviting the audience to watch his live stream and enter a lucky draw. Influencer No. 4’s most liked post promoted her video series about how to dye blue calico, a traditional dyeing skill in China. As for the remaining four most liked posts, two of them, from influencers Nos. 1 and 5, were about personal feelings, while influencer No. 8’s post was an endorsement of a Chinese TV drama. Influencer No. 10’s post encouraged Weibo users to share with him a few sentences about a book they read recently and liked.

Characteristics of influencer advertising-related posts

To answer Research Question 1, we tabulate the percentage of advertising-related posts for each online influencer in Table 2. Overall, the percentages of advertising-related posts were low. The percentages of promotional incentives, event-related posts and advocacy were 8.1%, 5.7% and 1.3%, respectively. Chi-square tests indicated that the distribution of these advertising-related posts was uneven among the ten online influencers. The influencer labeled No. 3 returned the highest frequency of advertising-related content. Altogether 4%, 42% and 21% of his posts contained advocacy, promotional incentives and event information, respectively. Close to one-quarter of all the posts (23.3%) have branded content. Chinese brands (17.5%) were more frequently mentioned than foreign brands (7.9%). These posts altogether mentioned foreign brands 161 times and Chinese brands 385 times. The ratio of mentioning Chinese brands to foreign brands was 2.4 to 1, indicating the dominance of Chinese brands over foreign brands in the social media platform. Advertising content was dominated by two online influencers (i.e. Austin Li and Teen Quanyou Liu). These two influencers together accounted for 39% and 42% mentioning of Chinese brands and foreign brands, respectively.

Most of these mentioned brands belonged to the product category of service (e.g. transport services, e-commerce and online platforms) and entertainment (e.g. media, movies and TV shows). This was the case for both foreign brands and Chinese brands. The frequently mentioned foreign brands included Twitter, Instagram and Walmart. The frequently mentioned Chinese brands were e-commerce platforms such as JD.com, TMall.com and Taobao. Online influencers often encourage the audience to look for a certain product or to get a sales promotion incentive by visiting these popular e-commerce platforms.

Characteristics of the posts and audience engagement statistics

Table 3 summarizes the visual presentation, mean length of posts and audience engagement statistics. The online influencers on average used around 66 words per post to share Weibo updates. A significant difference occurred among the ten online influencers in terms of length of post [F(9, 1769) = 14.1, p < 0.001]. Overall, the selected online influencers used visual aids to enrich their social media content. Online influencers used pictures and photos more frequently than videos. Altogether 47% of social media posts had photos and pictures, whereas only 18.5% of social media posts contained videos.

Not all online influencers and their social media posts could effectively engage online audiences. The average number of likes per influencer post was 26,990. A significant difference was found among them [F(9, 1769) = 55.7, p < 0.001]. Two female online influencers labeled as No. 4 and No. 2 enjoyed the highest number of likes. The least effective in generating likes was influencer No. 8. The number of comments was also used to measure audience engagement on social media. Although the average number of comments per influencer post was 8,666, a significant difference existed [F(9, 1769) = 15.2, p < 0.001]. More people shared their views and comments on the posts of the male online influencer labeled No. 3, while the female online influencer labeled No. 8 generated the lowest number of comments per post. In addition, the number of shares also reflected audience engagement. The average number of shares per online influencer post was 5,593. Again, a significant difference was found among them [F(9, 1769) = 8.9, p < 0.001]. As with the number of likes, influencer No. 4 embraced the highest number of shares, whereas influencer No. 8 had the lowest. Audience engagement statistics had a high correlation. The Pearson correlation coefficients between like and comment, like and share as well as comment and share were 0.67, 0.68 and 0.91, respectively (all significant at the 0.001 level).

Factors predicting audience engagement on social media

To answer Research Question 2, multiple regression was conducted to identify characteristics of social media posts that have a significant influence on audience engagement. Three multiple regression equations were compiled, with the same set of predictors, and a different engagement statistic for each equation. Table 4 summarizes the results. For the prediction of the number of likes, advocacy content, videos and foreign brand endorsement were not significant indicators, whereas the remaining five factors significantly influenced the number of likes [R² = 0.05, F(8,1770) = 10.82, p < 0.001]. They were the length of post (b = 0.08, t = 3.09, p < 0.01), with/without photos and pictures (b = −0.06, t = −2.41, p < 0.05), with/without promotional incentives (b = 0.12, t = 4.66, p < 0.001), with/without event information (b = 0.10, t = 3.89, p < 0.01) and with/without Chinese brand endorsement (b = −0.12, t = −4.52, p < 0.001).

The prediction of the number of comments and number of shares was similar, except that presence of pictures or presence of video had no significant impact on comments or shares.

To answer Research Question 2, the results showed that advertising contents were more important than presentation styles in influencing young consumers’ engagement with social media posts.

Audience engagement in online brand communication

Table 4 also highlighted the multiple regression results of whether brand communication discourages audience engagement. The interesting observation is that there are inconsistencies between the influence of advertising-related content and the mention of brands on young consumers’ engagement with the posts. On the one hand, young consumers were more likely to interact with posts with promotion incentives (b = 0.12, t = 4.66, p < 0.001) and event information (b = 0.10, t = 3.89, p < 0.01). On the other hand, young consumers were less likely to interact with posts mentioning brands (Chinese brands: b = −0.12, t = −4.52, p < 0.001). Therefore, Research Question 3 was answered. Brand communication, highlighting the promotional activities and brand events (e.g. fashion shows), significantly enhanced the audience engagement on social media.

Discussion and conclusion

To analyze the phenomenon of influencer marketing in China, 1,779 posts were crawled and coded from the top ten online influencers’ Sina Weibo accounts. Among the analyzed posts, 23.3% of the influencer advertising posts are about brand information. Advertising-related posts were not prevalent. The percentages of posts with promotion and event were both below 10%. Advocacy social media posts were rare, with only 1%. Photos/visuals are more frequently used by online influencers than videos on Sina Weibo.

The three dominating factors in predicting young consumers’ post engagement were posts with promotional content, posts about events and posts not mentioning Chinese brands. In addition, longer posts and posts without a picture can further increase the number of likes among social media audiences.

Combining traditional content analysis with data analysis of messages on online social media platforms obtained via Web data crawling, our study is, to the best of the authors’ knowledge, one of the first empirical studies to analyze influencer advertising and young consumers’ reactions to social media. It provides a prototype for the future content analysis of social media platforms. Theoretically, this study fills the literature gap in evaluating young consumers’ engagement behaviors based on the characteristics of the media content and brand communication. One more theoretical contribution of our study is that variables such as the length of the post, the use of visuals, promotional tactics, event marketing and brand origin are introduced to the study about consumers’ engagement with online influencers. These variables have varied impacts on consumers’ engagements with online influencers. The introduction of these variables to consumer engagement-related studies lays the evidence-based foundations for future studies to enrich their theoretical frameworks by involving such variables.

In our study, several interesting findings deserve attention from scholars and marketing communication practitioners. From the source credibility model, explicit brand communication will diminish the credibility of the online influencer. First, our study found that the selected online influencers do not often share and endorse brands on their Weibo accounts. On average, 23.3% of content is influencer advertising related to brand information. This can be explained by the intention of online influencers to manage their authenticity. Fan’s interview study (2021) found that compared with traditional celebrities, online influencers are less appreciated among Chinese respondents. To handle this issue, online influencers may intentionally improve and manage their social media content quality and develop an authentic image among social media users. In building credibility, organic content with self-disclosure and point of view might be more helpful than content with brand and product advertising (Hudders et al., 2021). From the perspective of brands, if social media users less value online influencers, they are less likely to pay to endorse the influencer’s brands and products. However, our finding contradicts Zhu and Wang’s (2020) study about influencer advertising of Chinese luxury brands. Their study highlighted that product advertisement was the most frequently featured advertising theme on bloggers’ social media accounts. Such different findings might be explained by the diverse metric used to collect Weibo data. As for our study, we targeted online influencers and collected all available Weibo posts. However, Zhu and Wang’s (2020) study crawled Weibo data about ten luxury brands in China. The varied study focus may lead to different results. Future studies are needed to crawl and analyze social media data by considering the characteristics of both online influencers and brands.

The contradicting influence of advertising-related content and brand communication on audience engagement is worth noting. This demonstrates that young consumers are highly selective in information processing. They are interested in information about promotional offers but are not interested in brand information. The marketing communication implication is that online influencers should skillfully include information about sales promotion without explicitly mentioning specific brands.

Second, although some scholars claim that the use of photos in social media campaigns allows engagement with online audiences (Luttrell, 2018), our study found that the use of photos in social media posts was not a significant predictor of the number of comments and shares. Interestingly, the photo has a negative correlation with the number of likes. Based on this result, we conclude that text plays a more important role than photos in engaging social media audiences in China. From the marketing communication perspective, content/text is treated as an essential element in influencer marketing, rather than the photos. One possible explanation is that the photos or videos posted by the influencers are unrelated to the content of their posts, leading to a sense of alienation from the audience. Influencer advertising is one type of native advertising, a paid form of content marketing to share commercial messages in the editorial format (Kim and Kim, 2021). To improve the effectiveness of content marketing, it is necessary to emphasize the high quality of the content. Another interesting finding is a positive correlation between post length and the number of likes, but not comments and shares. In other words, the longer the post and the fewer the photos that go with the post, the more likes will be generated. This rather intriguing result might be explained by the fact that longer texts attract more readers because they contain a great variety of content and are easier to search for. Our result was consistent with previous studies that long-text posts were more effective than short-text posts in China (Le and Aydin, 2022; Rohde and Mau, 2021). It may be because of the impression that long text is more authentic. Rohde and Mau (2021) suggested that long text enabled more elaborative storytelling and as a result more likely to blend marketing persuasion messages with non-commercial information.

Third, online influencers seldom share advocacy-related social media posts. This might be related to their perceived low credibility and vulnerable reputation in the audience’s mind. Some interviewees in Fan’s (2021) interview study mentioned that online influencers are not quite righteous. If online influencers lack credibility and repute, it will be of concern to organizations or brands wishing to involve them in endorsements relating to social causes.

Limitations and future studies

The study provides new insights and facilitates future research agendas on influencer advertising in China. However, this research has its limitations. First, the current findings heavily address the descriptive statistics about texts and pictures of influencer advertising and consumer engagement on social media, without analyzing embedded emojis, video content and so forth. Besides, the top social media environment in China is changing rapidly. Young consumers’ engagement behaviors on new platforms such as TikTok, Xiaohongshu and Zhihu may differ from their engagement with Sina Weibo platform. Thus, future research could investigate how online influencers interact in these new platforms. Third, content analysis has the limitation of understanding consumers’ subjective response. Future studies could use surveys, interviews or ethnographic methods to provide a more comprehensive understanding of influencer marketing and their responses to social media.

Table 1

Profiles of top ten online influencers and typical and most liked post contents

No.	Online influencer	No. of fans (‘000)	Sex	Age	Education	Content of posts is mostly about…	Post with the highest number of likes (translated post content)	Likes for this post
1	Jia Ma	41,622	Male	35	Bachelor	Personal feelings, humor	Not easy #why I don't share my pain # Weibo accompany you for the Chinese New Year	96,895
2	Papi Jiang	33,119	Female	35	Master	Funny videos, pets, feelings as a woman	A formal clarification about the online issue a few months ago	1,114,617
3	Austin Li	29,108	Male	29	Bachelor	Promotion of cosmetics and snacks	Li Jiaqi’s Double Eleven is crazy! Big sale for well-known brands! At 7 pm October 20th, be sure to watch Jiaqi’s live streaming. Big surprise is waiting for you. Follow + Add the hashtag of # advance notice for Li Jiaqi’s Double Eleven# + Forward, like and comment this Weibo post. Then, you will earn a chance to (1) become one of the 111 lucky persons to receive one big-brand handbag for free per person! (2) become one of the 1111 lucky persons to receive a RMB666 voucher per person! (3) become one of the 1111 lucky persons to receive a gift package from Never Mind Cafe!	1,936,257
4	Ziqi Li	27,569	Female	29	Junior high school	Food	As the old saying goes “the student surpasses the master”. From farming to harvesting + Description of how to dye and make clothes + It's hard to imagine how many people still remember how to produce the handmade Blue Calico! #Li Ziqi dyeing Blue Calico#	1,243,623
5	13 Dai	20,268	Female	No information	No information	Funny, talk show	#When I am with a good friend#, I look like this…	32,520
6	Uncle Tong Dao	18,554	Male	34	Bachelor	Constellation/astrology	#Hello Celebrity# This time, we have Sagittarius guest--@Gong Jun Simon + Description of how fans react to the muscular side of Simon + Rumors about Sagittarius people as “playboys” are not true! I confess that I met someone as sweet as Wen Kexing in my life. What is the most popular comment at KTV? Forward this “Gong Jun Simon”, you can also make money∼	106,846
7	UK Baojie	18,362	Female	35	Doctor	Daily life and culture of other countries	#Zheng Kai announced that his wedding ceremony will be held this year # Zheng Kai and Miaomiao will have a wedding ceremony this year! In the birthday party, Zheng Kai announced that his wedding ceremony will be arranged this year, and he also shared his daughter's nickname with the public. Miao Miao sent a happy birthday post to Zheng Kai on Weibo	108,264
8	My ex is a jerk	17,088	Female	No information	No information	Humor, feelings and pets	TV drama “I am your eye”. Really touching friendship!	5,690
9	Teen Quanyou Liu	16,856	Male	No information	No information	Humor, feelings	I accidentally clicked on Shen Teng’s ex-profile picture. Oh my God + [the content creator's surprised feeling about the photos] Help me!	74,237
10	Mr Deca’s mailbox	16,235,695	Male	No information	No information	Encouragement and positive thinking	Which book have you been reading recently? Please share a paragraph you like very much in this book	33,726

Source: Authors’ own work

Table 2

Characteristics of advertising-related social media posts by online influencers

No.	KOL	No. of sampled posts	Advocacy, %	Promotion, %	Event, %	Mention foreign brand, %	Mention Chinese brand, %	Mention any brand, %
1	Jia Ma	200	1.0	2.5	1.0	3.5	13.5	16.5
2	Papi Jiang	124	3.2	4.0	5.6	7.3	16.1	22.6
3	Austin Li	200	4.0	42.0	21.0	18.5	28.5	42.0
4	Ziqi Li	53	0.0	9.4	3.8	1.9	20.8	20.8
5	13 Dai	200	1.0	5.0	6.0	10.5	15.0	23.5
6	Uncle Tong Dao	201	0.0	2.5	0.0	0.5	17.4	17.9
7	UK Baojie	200	0.5	4.0	4.0	20.5	23.5	41.0
8	My ex is a jerk	200	0.0	0.5	0.5	2.0	5.0	6.5
9	Teen Quanyou Liu	200	1.5	9.5	13.0	8.5	32.0	35.5
10	Mr Deca’s mailbox	201	1.5	1.0	1.0	1.0	5.0	5.0
	Total	1,779	1.3	8.1	5.7	7.9	17.5	23.3
	Chi-sq. (significance)		22.4**	365.6***	146.3***	122.8***	98.1***	168.7***

Notes:

***p < 0.001; **p < 0.01

Source: Authors’ own work

Table 3

Summary of presentation of social media posts and audience engagement

No.	KOL	Length of post	With picture, %	With video, %	Like	Comment	Share
1	Jia Ma	67	48.5	11.0	5,612	872	1,305
2	Papi Jiang	68	29.8	26.6	116,865	14,375	13,867
3	Austin Li	94	45.0	23.0	81,919	55,122	26,371
4	Ziqi Li	92	45.3	32.1	188,617	34,842	27,032
5	13 Dai	51	68.0	38.0	1,913	324	1,099
6	Uncle Tong Dao	67	36.3	19.4	5,180	772	1,850
7	UK Baojie	86	58.5	21.0	8,419	578	1,718
8	My ex is a jerk	29	65.5	11.0	450	109	361
9	Teen Quanyou Liu	73	55.0	7.5	5,214	677	597
10	Mr Deca’s mailbox	56	10.9	9.0	8,855	478	677
	Mean	66	47.0	18.5	26,990	8,666	5,593
	F-stat/Chi-sq. (significance)	14.1***	207.8***	108.8***	55.7***	15.2***	8.9***

Note:

***p < 0.001

Source: Authors’ own work

Table 4

Summary of results of multiple linear regression for predicting audience engagement

	Standardized coefficients beta
	Like	Comment	Share
Length of post	0.08*	0.03	0.04
With picture	−0.06*	−0.00	0.01
With video	0.03	0.02	0.00
Advocacy	0.02	0.01	0.01
Promotion	0.12***	0.25***	0.18***
Event	0.10***	0.16***	0.16***
With foreign brand	−0.04	−0.04	−0.04
With Chinese brand	−0.12***	−0.12***	−0.12***
Adjusted R sq.	0.05	0.11	0.07
F-stat (significance)	10.82***	27.02***	17.2***

Notes:

*p < 0.05; ***p < 0.001

Source: Authors’ own work

Appendix. Coding variables

Advocacy: Does the post include an activity related to a social cause?

0 = Post includes no PSA

1 = Post includes PSA

Promotion: Does the post include promotion activities providing incentives that drive immediate action (e.g. lucky draw)

0 = Post includes no promotion

1 = Post includes a promotion

Events: Does the post include time or location specific activities that drive immediate action (e.g. fashion show)

0 = Post includes no event

1 = Post includes an event

Foreign brands

0 = Post includes no foreign brand

1 = Post includes foreign brand

Number of foreign brands

Write the number of foreign brands in the post. If unsure about the brand’s origin, check the brand online.

Chinese brands

0 = Post includes no Chinese brand

1 = Post includes Chinese brand

Number of Chinese brands

Write the number of Chinese brands in the post. If unsure about the brand’s origin, check the brand online.

Brands (deduced from above, no coding is needed)

0 = Post includes no brand

1 = Post includes either a Chinese brand or a foreign brand

Variables that can be extracted from data crawling and no need to code length of post

Use Excel formula to compile: = len()

Engagement (already part of the data crawling)

• Likes

• Shares

• Comments

Photos/visuals

0 = Post includes no photos/visuals

1 = Post includes photos/visuals

Videos

0 = Post includes no videos

1 = Post includes a video