Facilitating Endorsement Efficacy: The Interplay of Parasocial Interaction, Product Placement, and Influencer Type

Abstract

Social media platforms fostering a closer and more intimate bond between celebrities and their fan bases has opened up diverse avenues for product placement. In light of this, this study endeavors to explore the profound influence of parasocial interaction (PSI) and product placement on the effectiveness of celebrity endorsement within the social media landscape. The results derived from an online experiment unveil the positive impact of parasocial interaction on consumers’ attitudes and purchase intention. Notably, the potency of PSI is enhanced when product placement is prominent. The results also uncover the crucial role of brand recall as a mediator in the relationship between parasocial interaction and endorsement outcomes. These findings shed light on the underlying mechanisms governing the influence of parasocial interaction and social media celebrity types in the realm of social media endorsement while also providing valuable insights into the moderating role of product placement. From a practical standpoint, the results underscore the critical importance of carefully selecting celebrity endorsers and strategically positioning products. Armed with this knowledge, marketers and advertisers can better explore the complex landscape of social media endorsement with greater efficacy and precision.

1. Introduction

Social media has emerged as a widely embraced platform for the acquisition of information and social interaction, capturing the attention of individuals across the globe [1]. The pervasiveness of social media has not only transformed the manner in which individuals interact with one another but has also led to a significant shift in advertising strategies, with social media influencer endorsements emerging as a highly effective marketing tool [2,3]. It is notable that social media influencers are not limited to traditional celebrities; these platforms provide numerous opportunities for individuals to express themselves and captivate audiences [4,5]. By sharing glimpses into their daily lives or creating entertaining content, certain users have amassed significant followings, thereby establishing themselves as influential figures commonly referred to as “micro-influencers” [6].

While both micro-influencers and traditional celebrities are active on social media, there are notable differences between the two. Traditional celebrities are typically associated with specific professional domains, such as entertainment or sports [5]. In contrast, micro-influencers are typically sourced from social media and are perceived as more relatable [7]. They exert influence across a range of domains, including health, fashion, and technology [8]. Micro-influencers can engage audiences through their distinctive personalities [9]. This nuanced approach renders them an attractive option for advertisers seeking authentic engagement in the digital age [10].

The advent of social media platforms has also transformed the dynamics between celebrities and their fans [11]. The advent of real-time interaction and communication facilitated by social media has led to the formation of relatively stable yet imaginative interpersonal connections between audiences and influencers, a phenomenon known as “parasocial interaction” (PSI) [12]. Furthermore, the domain of advertising on social media has diversified to encompass a multitude of formats [13]. In contrast to traditional media, which adheres to fixed time slots for advertising, micro-influencers and celebrities on social media not only share glimpses into their daily lives but also engage in the dissemination of advertising content. As a consequence, followers frequently encounter difficulties in differentiating between genuine personal narratives and content that has been sponsored by advertisers [14].

The factors that influence the persuasive impact of advertising can be classified into three main categories: the source, the message, and the audience [15]. The Elaboration Likelihood Model (ELM) further indicates that individuals adopt different information processing routes in disparate involvement contexts, which further affects the effects of celebrity endorsements. It is noteworthy that both the source and message factors of celebrity endorsement have undergone significant transformations within the context of social media [12,13]. On the one hand, there is the rise of a multitude of micro-influencers who bring diverse perspectives and content. Their relationship with followers is much closer, which also affects the followers’ information processing route [16]. Conversely, the placement of advertising messages has become more varied, flexible, and indistinct [17]. A growing body of studies has explored the influencer advertising mechanism [4,10], as well as the varying impacts of product placement strategy on social media endorsements in recent years [18,19]. Nevertheless, further research is required to elucidate the cumulative impact of influencer types and product placement on advertising effectiveness.

In light of the aforementioned considerations, this study endeavors to investigate the combined impact of the influencer type and product placement strategy on the efficacy of an endorsement. To this end, it draws upon the theoretical frameworks of PSI and ELM. Furthermore, this study aims to compare the differential impact of PSI on endorsement across various types of influencers. The findings will contribute to the existing literature on this topic by elucidating the psychological processes underlying consumer responses to social media endorsements. This will enhance our comprehension of the combined effects of the source (traditional celebrity/micro-influencer), message (explicit/implicit), and audience (high vs. low PSI with the endorser) in social media advertising. Furthermore, this study will provide empirical evidence to inform the selection of spokespersons and product placement strategies for advertisers, offering actionable insights for those navigating the complex social media landscape.

2. Literature Review and Hypothesis Development

2.1. Parasocial Interaction on Social Media

Celebrity endorsers are defined as individuals with high public visibility who endorse products and are themselves revered as consumers [11]. The form of advertising which uses celebrity endorsers to promote a product/brand is defined as celebrity endorsement [20]. Traditional celebrity endorsers include entertainment and sports stars and experts in politics, health, and other fields. The advent of social media has led to the emergence of a new form of celebrity known as the “micro-influencer” [21], whose identity is based on the recognition and affection of their followers [2,22].

“Parasocial interaction” (PSI) describes a form of one-sided interaction that exists primarily in the imagination of the audience and is part of a broader concept known as the “parasocial relationship” (PSR) [23,24]. Compared to traditional media, social media platforms provide a direct dialogue space between celebrities and their followers, and followers’ feedback is more timely, allowing both parties to achieve a sense of equal and direct communication [25,26]. An increasing number of brands are utilizing social media influencers, both celebrities and micro-influencers, to promote their products. This trend was particularly prominent during and after the COVID-19 pandemic [27]. Social media influencers who increase the frequency of interaction with fans are more likely to establish a high level of PSI with followers [28,29]. PSI can not only enhance fans’ affection for the spokesperson [30,31] but also transfer the level of favorable brand/product attitudes and even positive evaluations of brand/product reputation [14]. Notably, while PSI can boost the effectiveness of celebrity endorsements, highly commercial social media platforms can have the opposite effect. Furthermore, once followers detect the commercial intent behind an influencer’s posts, they may question the influencer’s credibility. This skepticism can damage the influencer’s PSI and undermine the effectiveness of their endorsements [32].

The majority of previous studies support that PSI has a positive effect on consumers’ positive attitudes and purchase intentions, which can also boost consumers’ willingness to engage with the brand [2,16,33]. Through interactions with followers on social media, advertising campaigns could promote the brand’s values and benefits, leading to increased brand influence [34]. This relatively stable PSI would increase the followers’ attention to the celebrity and subsequently enhance the audience’s brand recall of the celebrity’s social media endorsement. Based on the previous literature, the following hypothesis is proposed:

H1. 

PSI has a positive effect on (a) endorsed brand recall, (b) attitude toward the advertising (Aad), (c) attitude toward the brand (Ab), and (d) purchase intention (PI).

2.2. The Mediation Role of Brand Recall

Brand recall is a pivotal metric for assessing advertising effectiveness. It refers to the consumers’ ability to accurately remember brands that were presented within media content without any external cues or assistance [35]. Brand recall depends on the consumers’ familiarity with the brand [36]. Additionally, the endorser’s familiarity and trustworthiness can affect the followers’ memory of their endorsed brands [37].

Boerman [38] found that the disclosure of social media influencers on Instagram positively affects fans’ brand recall. Followers are more familiar with the influencers with whom they engage in a high level of PSI, which can lead to preference and reduce the fear of uncertain risks [39]. Therefore, social media influencers might be considered to have a higher level of source credibility [40]. High source credibility can bring a favorable attitude towards the brand and also increase the willingness to read the implanted advertisement content [41].

Previous studies provided ample evidence for the positive influence of PSI on brand recall [42,43], as well as the subsequent favorable effect of brand recall on consumers’ attitudes and purchase intention [44,45]; hence, it is reasonable to infer a potential chain of influence from PSI to brand attitudes and behaviors mediated by brand recall. Hence, we propose the following hypothesis:

H2. 

brand recall mediates the relationship between PSI-Aad, PSI-Ab, and PSI-PI.

2.3. Product Placement

Product placement is defined as the exposure of a product or brand in a media program that is paid for by a corporation [46]. Product placement is widely used in television dramas, movies, variety shows, and online games. The prominence of product placement is the level of product placement that has the characteristics of product placement, with the aim of attracting the audience’s attention [47]. Product placement affects the audience’s memory, attitude, and behavior intention [48]. The rise of the internet and social media has enriched the context and form of product placement. Due to the high frequency with which consumers use social media, brand/product placement social media can have more frequent interactions with consumers [49].

Prominent product placement is defined as explicit product placement, while subtle placement is defined as implicit product placement [46]. The prominence of product placement is found to be positively related to brand image and recall [48,50]. Song and Xualso found that explicit product placement leads to higher brand recall compared to implicit product placement [51]. In social media contexts, the product placement also influences audiences’ attitudes and purchase intentions [52,53]. Chan found that brand attitudes were better for explicit than implicit product placement [18]. He attributed this to the increased familiarity from simple exposure effects, which in turn boosted brand favorability. When products or brands are placed in environments overwhelmed with information, audiences tend to have a better brand attitude towards the brand when the placement is more prominent.

Prior research has found that explicit product placement is more significant in enhancing the audience’s memory and recognition of brands/products. Jin and Muqaddam [19] found that consumers perceive higher corporate credibility and exhibit more positive brand attitudes when social media influencers are present in product placements on Instagram, compared to only the brand present in the advertising. PSI mediates between the placement types and corporate credibility. In this study, followers are more likely to notice and remember the explicit placement when they engage in high PSI with the social media influencer. However, implicit product placements may escape notice, as followers are predominantly drawn to the influencer’s personal message. Thus, the following hypotheses are proposed:

H3. 

the prominence of the product placement has a positive effect on endorsed (a) brand recall, (b) Aad, (c) Ab, and (d) PI.

H4. 

the positive effect of PSI on endorsed brand recall, Aad, Ab, and PI will be stronger when product placement is more explicit (vs. implicit).

The hypothesized model is summarized in Figure 1.

2.4. Traditional Celebrity vs. Micro-Influencer

While both traditional celebrities and micro-influencers are active on social media platforms, there exist distinct differences between them. Traditional celebrities are rooted in mass media channels [5]. Micro-influencers, on the other hand, are a product of the social media age and are perceived by users as being “one of us” [7]. The integration of social media and smartphones has become a significant aspect of daily life nowadays; hence, micro-influencers are increasingly influential. Social media platforms themselves serve as a stage for self-expression, and one of the most important characteristics of successful micro-influencers is their personal charm [54].

Compared to traditional celebrities, micro-influencers have a more widespread presence in real life [8] and can garner attention and influence among certain users through their unique strengths and characteristics [9,30]. For followers, the quality of information and perceived similarity of micro-influencers are the keys to their appeal [55]. Users on social media tend to follow bloggers who are similar to themselves or their own social circle [56], which makes them more likely to trust the micro-influencer [5,29].

These micro-influencers often combine paid advertising content with their personal life, making it more subtle compared to straightforward advertising promotion by traditional celebrities [4]. Based on the recognition of and trust in a social media influencer, followers tend to receive personalized marketing content, especially for those with a smaller number of fans [26]. Gerlich (2023) found that influencers with smaller fan bases, categorized as micro- or nano-influencers, could be more persuasive. This is because they often have stronger relationships with their followers compared to mega- and macro-influencers [57]. Through their practical knowledge and unique charisma, they can become reliable sources of information and bring positive results to advertising [2,7,22].

Recent studies have paid more attention to the different influences of traditional celebrities and micro-influencers and also to micro-influencers with a different number of followers. Sheng, Lee, and Lan found that followers have a greater degree of identification with and more similarity and trust in micro-influencers than traditional celebrities [58]. PSI plays a much more significant role in advertising effectiveness for micro-influencers than traditional celebrities [59]. Notably, a follower’s perceived weak PSI with a micro-influencer may have adverse effects on endorsement effectiveness; however, the same is not found to be true for traditional celebrities’ influence [60]. Based on the profound findings of the previous literature, we propose the following research question:

RQ1. 

what are the different influences of PSI on traditional celebrity and micro-influencer endorsement effectiveness?

Previous research has found that explicit product placement advertisements have significantly higher brand recall and advertising effectiveness than implicit advertisements, but explicit product placement advertisements are also more likely to elicit negative reactions from the audience [18]. Followers are familiar with the content style of social media influencers, making it easier for advertising information to capture their attention [26]. However, the promotion of product placement on a social media influencer’s account carries the risk of disrupting their PSI, as followers may resist excessive promotional content within the influencer’s social media channel [58]. Therefore, the influence of product placement may have an interaction effect with celebrity identity, as indicated in the following research question:

RQ2. 

what is the interaction effect of social media influencer types and product placement on endorsement effectiveness?

3. Methods

3.1. Research Design and Sample

This study employs a 2 (celebrity identity: traditional celebrities/micro-influencers) by 2 (product placement: explicit placement/implicit placement) between-subjects factorial design experiment to test the hypotheses. To further enhance the effectiveness of the results, traditional celebrities and micro-influencers were selected through a comprehensive ranking system based on Weibo-related metrics. Prior to commencing the primary experiment, the author conducted a preliminary trial and selected the final celebrity endorsers and the degree of advertising prominence based on the results of the questionnaires. In the primary experiment, a virtual Weibo interface was integrated into the questionnaire software, and participants completed the survey after viewing the materials.

3.2. Pretest

In order to ascertain the most suitable celebrity endorsers for the forthcoming experiment, an initial online survey was conducted as a pretest. In particular, only female celebrities were selected in order to preclude any potential gender-related biases from influencing the experiment. Furthermore, celebrities who had previously endorsed toothpaste were deliberately excluded from this study. The popularity index on Chinese social media platforms (Xiaohongshu and Weibo) was used as the basis for selecting four celebrities for testing purposes. The selected celebrities were Liu Yifei, Qi Wei, Li Ziqi, and Papi Jiang. Liu Yifei and Qi Wei were selected as representatives of traditional celebrities, whereas Li Ziqi and Papi Jiang were chosen as representatives of micro-influencers.

An online survey was conducted with a total of 154 respondents. Participants were asked to provide ratings on perceived credibility, product–celebrity congruence, and their level of parasocial interaction with the four celebrities. The questionnaire comprised three dimensions, perceived credibility [59], product–celebrity congruence [60], and parasocial interaction [61,62], and they were all measured by a Likert scale ranging from 1 to 7. The results indicated that traditional celebrity Liu Yifei obtained the highest scores with a mean of 5.0 (SD = 1.149). Following closely was the micro-influencer Li Ziqi, who received an average score of 4.73 (SD = 1.151). Consequently, Liu Yifei and Li Ziqi were selected as the final experimental celebrities for this study.

3.3. Stimulus

The stimuli included the determination and design of the product category, as well as the creation of the actual stimulus. Toothpaste was selected as the experimental product category, and a virtual product was developed for three specific reasons. Primarily, the selected experimental product was required to be a necessity, thereby ensuring that respondents would have a genuine need for it in their daily lives. This would help to reduce the potential impact of extraneous factors on the experiment. Secondly, considerations were made to address the needs of both men and women, as the demand for this type of product is relatively similar across genders. This approach helped to mitigate any uncontrollable factors that might have arisen from gender differences during the course of the experiment. Lastly, to prevent participants from forming fixed, preconceived impressions of existing brands, which could introduce uncontrollable experimental factors, a self-designed virtual brand was utilized for this study.

The key aspect of creating the stimulus was to highlight significant differences between explicit placement and implicit placement. These differences were discerned based on factors such as product/brand size [63], exposure time, placement location, brand positioning, interaction with characters, and close-ups [44]. Following the classification criteria, four sets of experimental materials were created using Photoshop: Liu Yifei and explicit placement, Liu Yifei and implicit placement, Li Ziqi and explicit placement, and Li Ziqi and implicit placement. Participants were randomly assigned to these four groups through the Sojump platform, resulting in a total of 154 valid questionnaires. The final stimulus materials are shown in Appendix A.

The results are as follows: For traditional celebrities, participants in the explicit placement condition perceived the placement to be much more prominent (M = 5.533, SD = 0.929) than those who were in the implicit placement condition (M = 3.911, SD = 1.704), t(29) = 4.578, p < 0.001; for the micro-influencer, the explicit product placement condition (M = 5.622, SD = 1.053) was also perceived to be more prominent than the implicit placement condition (M = 4.422, SD = 1.813), t(29) = 3.315, p < 0.01. A significant difference exists between explicit and implicit product placements. Therefore, the four groups of posters can be used in the main experiment.

3.4. Procedure

The experiment was conducted using a questionnaire embedded in a html5 page. Firstly, the social media experience was created using the “Yiqixiu” html5 application to simulate the Weibo browsing experience, utilizing a combination of images and text. Secondly, Photoshop was used to create sixteen posters, which were divided into four groups, with each group containing four posters. Finally, the four groups of posters were integrated into a virtual Weibo page using the html5 application, and a corresponding caption was provided for each poster. The explicit advertisements included the tagline “Herbs for a fresher feel and more effective whitening,” along with the hashtag “#QingBenToothpaste#”. The implicit advertisements were context-dependent and did not directly mention the words “QingBen Toothpaste”.

The experimental procedure was conducted in three steps. First, the participants were randomly assigned to one experimental condition after entering the invitation link. The participants were required to watch a celebrity video as a prelude to the formal experiment. Second, the participants were prompted to scroll down and view four posters with corresponding text, featuring a spokesperson. Additionally, replay was prohibited, and participants were instructed to close the page. Finally, respondents returned to the questionnaire page to complete the survey.

To ensure the quality and validity of our data, we implemented an exclusion criterion based on response time. Participants who completed the entire questionnaire in less than 180 s (3 min) were excluded from the final analysis. This decision was based on our pretesting phase, which indicated that thoughtful completion of the questionnaire typically required approximately 5 min. The 180 s threshold was established to filter out responses that were likely completed without careful consideration of the questions. Given the complexity and depth of the questions related to parasocial interaction and brand recall, we determined that responses submitted in less than 3 min were unlikely to reflect thoughtful engagement with the survey content.

Finally, 300 validated questionnaires were collected, with 75 samples per group; around 41.3% (n = 124) of the participants were male, and 58.7% (n = 176) were female. The age of the participants ranged from 18 to 45 years. The demographics of participants are summarized in

Table 1. Demographics of sample.

Gender	Frequency	Percentage
Male	124	41.33
Female	176	58.67
Age
18–25	112	37.33
26–35	176	58.67
36–45	12	4.00
Education
High school or below	5	1.67
Junior college	31	10.33
Bachelor	212	70.67
Master or above	52	17.33

.

3.5. Measurement

Follower–celebrity PSI: We adopted the established PSI scale to measure this variable [64,65,66]. The original PSI scale was developed in traditional mass media research. Therefore, the word “television” in items 4 and 5 was replaced with “social media” to meet the social media context (8 items, α = 0.90, M = 5.15, SD = 0.10).

Product placement: The scale was developed based on the study by D’Astous and Chartier [47]. We modified the sentence expressions of three selected items to align with the experimental scenario: “I observed the toothpaste brand in the picture, which made a profound impact on me, and I am able to recall relevant images”, “I did take notice of the toothpaste brand endorsed by the spokesperson in the picture”, and “I believe that the information about this toothpaste brand is easily noticeable”. A seven-point Likert scale was used (α = 0.97, M = 4.66, SD = 1.36).

Brand recall: Early measurements of advertising recall included two types of tests: recall tests and recognition tests. Recall tests involve independently recalling the content of an advertisement, while recognition tests are typically referred to as assisted recall [67]. A virtual brand that was completely unfamiliar to the participants was used, so a recognition test was employed. Three items were used to test the recall of “the brand name”, “packaging”, and “product features” in advertising placement. A Likert five-point scale was used for the measurement, with 1 indicating “no memory” and 5 indicating “very impressive” (α = 0.78, M = 4.20, SD = 1.28).

Advertising effectiveness: Advertising effectiveness consists of three dimensions: advertising attitude, attitude toward the brand, and purchase intention [68,69,70]. The attitude toward the ad dimension was measured using an established attitude measurement scale (a 7-point semantic differential scale) consisting of six items: attractive/unattractive, interesting/uninteresting, pleasant/unpleasant, memorable/forgettable, likable/dislikable, and fun/boring (α = 0.92, M = 4.99, SD = 1.18). The attitude toward the brand dimension was measured using a 7-point semantic differential scale consisting of six items: like/dislike, valuable/worthless, appreciable/inappreciable, useful/useless, beneficial/not at all beneficial, and touching/unmoving (α = 0.93, M = 4.97, SD = 1.14). Purchase intention was assessed through a 7-point Likert scale question [71,72], asking whether participants are willing to buy, recommend to others, prioritize buying, and whether or not the celebrity endorsement has stimulated their purchase intention (α = 0.94, M = 4.68, SD = 1.48).

4. Results

4.1. Manipulation Check

We conducted an ANOVA to see if the manipulation of product placement prominence was successful. The results showed that people in the explicit placement condition perceived the product placement as much more prominent than those in the implicit placement condition (Mexplicit = 5.24, SD = 1.15, Mimplicit = 4.08, SD = 1.31, F(1, 298) = −8.13, p < 0.001). As such, the manipulation of product placement prominence was successful.

4.2. Hypothesis Testing

First, we conducted a series of analyses using PROCESS macro [73]. The results suggested that PSI had a positive effect on brand recall (point estimate = 0.26, SE = 0.09, t = 2.74, p < 0.01), Aad (point estimate = 0.33, SE = 0.12, t = 2.62, p < 0.01), Ab (point estimate = 0.37, SE = 0.07, t = 5.55, p < 0.001), and PI (point estimate = 0.51, SE = 0.09, t = 5.93, p < 0.001). As such, H1 was supported. Next, mediation analyses suggested that brand recall mediated the effect of PSI on Aad (point estimate = 0.10, SE = 0.02, 95%-CI [0.06, 0.15]), Ab (point estimate = 0.11, SE = 0.03, 95%-CI [0.07, 0.17]), and PI (point estimate = 0.16, SE = 0.04, 95%-CI [0.10, 0.24]). Therefore, H2 was also supported. The results are summarized in

Table 2. Path coefficient of testing model.

Hypothesis Testing			Coefficient	SE	t	p	95%CI
H1a	PSI→Brand recall	Support	0.26	0.09	2.74	<0.01
H1b	PSI→Aad	Support	0.33	0.12	2.62	<0.001
H1c	PSI→Ab	Support	0.37	0.07	5.55	<0.001
H1d	PSI→PI	Support	0.51	0.09	5.93	<0.001
H2a	Brand recall mediates the relationship between PSI and Aad	Support	0.10	0.02		<0.001	0.06, 0.15
H2b	Brand recall mediates the relationship between PSI and Ab	Support	0.11	0.03		<0.001	0.07, 0.17
H2c	Brand recall mediates the relationship between PSI and PI	Support	0.16	0.04		<0.001	0.10, 0.24

.

Also, we observed significant main effects of product placement on brand recall (Mexplicit = 4.43, SD = 1.14, Mimplicit = 3.96, SD = 1.36, t(298) = 3.28, p < 0.001), Aad (Mexplicit = 5.63, SD = 0.93, Mimplicit = 4.36, SD = 1.05, t(298) = 11.17, p < 0.001), Ab (Mexplicit = 5.63, SD = 0.93, Mimplicit = 4.38, SD = 0.98, t(298) = 10.50, p < 0.001), and PI (Mexplicit = 5.47, SD = 1.07, Mimplicit = 3.89, SD = 1.40, t(298) = 11.02, p < 0.001). The results are summarized in

Table 3. Endorsement effectiveness in different placement conditions.

Advertising Effectiveness		M	SD	T(298)	p
Brand recall	Explicit placement	4.43	1.14	3.28	<0.01
	Implicit placement	3.96	1.36
Attitude toward advertising	Explicit placement	5.63	0.93	11.17	<0.001
	Implicit placement	4.36	1.05
Attitude toward brand	Explicit placement	5.63	0.93	10.50	<0.001
	Implicit placement	4.38	0.98
Purchase intention	Explicit placement	5.47	1.07	11.02	<0.001
	Implicit placement	3.89	1.40

. Therefore, H3 was supported.

Furthermore, the positive effect of PSI on Aad (point estimate = 0.37, SE = 0.07, t = 5.40, p < 0.001), Ab (point estimate = 0.39, SE = 0.12, t = 3.21, p < 0.01), and PI (point estimate = 0.32, SE = 0.16, t = 2.01, p = 0.05) depended on the product placement prominence. As predicted, when the product placement was explicit, PSI increased Aad (point estimate = 0.70, SE = 0.10, t = 6.66, p < 0.001), Ab (point estimate = 0.75, SE = 0.10, t = 7.46, p < 0.001) and PI (point estimate = 0.83, SE = 0.13, t = 6.27, p < 0.001). However, the effect of PSI on Aad (point estimate = 0.37, SE = 0.07, t = 5.40, p < 0.001), Ab (point estimate = 0.37, SE = 0.07, t = 5.55, p < 0.001), and PI (point estimate = 0.51, SE = 0.09, t = 5.93, p < 0.001) was weaker when the product placement was implicit. Furthermore, the effect of PSI on brand recall was not moderated by product placement prominence (point estimate = 0.19, SE = 0.17, t = 1.11, p = 0.27). This suggested that, for both explicit and implicit product placement, PSI influenced brand recall positively. As such, H4 was partially supported.

4.3. Differences Between Celebrity and Micro-Influencer

To answer the research question, we performed additional analyses. For research question 1, the results suggested that the effect of PSI on brand recall also depended on the celebrity type (point estimate = 0.30, SE = 0.14, t = 2.14, p = 0.03). In particular, for micro-influencers, PSI positively influenced the brand recall (point estimate = 0.50, SE = 0.09, t = 5.31, p < 0.001). However, for traditional celebrities, the effect of PSI on brand recall was not significant (point estimate = 0.20, SE = 0.10, t = 1.89, p = 0.06). In contrast, the celebrity type did not moderate the effects of PSI on Aad (point estimate = 0.04, SE = 0.11, t = 0.32, p = 0.74), Ab (point estimate = 0.12, SE = 0.11, t = 1.12, p = 0.26), or PI (point estimate = 0.05, SE = 0.14, t = 0.37, p = 0.71).

For research question 2, the results showed that there is no interaction between social media influencer type and product placement prominence on brand recall (point estimate = 0.11, SE = 0.29, t = 0.37, p = 0.71), Aad (point estimate = 0.30, SE = 0.23, t = 1.32, p = 0.19), Ab (point estimate = 0.15, SE = 0.22, t = 0.65, p = 0.52), or PI (point estimate = −0.003, SE = 0.29, t = −0.01, p = 0.99).

5. Discussion and Implications

This study presents empirical evidence of the synergistic effects of PSI, influencer types, and product placement on social media endorsement. The experimental results demonstrate that parasocial interaction enhances consumers’ brand recall, attitude toward the advertising and brand, and purchase intention. The impact of PSI on attitudes and purchase intention is mediated by brand recall. Moreover, the impact of PSI on Aad, Ab, and PI is subject to the degree of product placement salience. The positive effects of PSI on Aad, Ab, and PI are more pronounced when the product placement is more overt. In addition, the impact of PSI on brand recall varies according to the type of influencer. The positive effect of PSI on brand recall is more pronounced for micro-influencers. One potential explanation is that micro-influencers are perceived to be more accessible and relatable [7]. In this context, parasocial interaction with micro-influencers may be perceived as more authentic, which in turn produces a stronger effect on the recall of the micro-influencer-endorsed brand.

5.1. Theoretical Implications

The findings of the present study fill a void in the literature and thus contribute to PSI research by further investigating the effect of PSI on endorsement effectiveness within different social media influencer types and different product placement scenarios. First, consistent with the findings of previous research, our findings further confirm the positive effect of PSI on social media endorsement effectiveness [11,16]. This study also supports the notion that brand recall mediates the relationship between PSI and consumers’ attitudes and purchase intention. While the majority of previous studies have directly examined the influence of PSI on consumers’ attitudes and purchase intentions, this study highlights the importance of brand recall through which PSI influences consumers’ responses to celebrity endorsement on social media. The mediation model further elucidates the psychological mechanisms underlying followers’ responses to influencer endorsement on social media.

Second, this research sheds light on the role of product placement in celebrity endorsement. Product placement has a positive impact on brand recall but also moderates the relationship between PSI and advertising effectiveness. Specifically, the impact of PSI is found to be more pronounced when the product placement is explicit. While both PSI and product placement are recognized as pivotal factors influencing celebrity endorsements on social media [18,40], this study provides further understanding of the influence of PSI within different product placement strategies on social media. The past literature suggests that explicit advertising placements may evoke followers’ aversion, thereby affecting advertising effectiveness and the relationship with influencers [74]. According to ELM, some research also found that fans might process messages through a central route when they have a closer relationship with an influencer, making implicit placements more effective [75]. However, our results show that in an environment saturated with advertisements, explicit placements perform better. For followers with PSI with influencers, who frequently follow celebrity social media updates, exposure to an explicit message leads to better advertising outcomes.

Third, the effect of PSI on brand recall is moderated by social media influencer type. The finding that PSI has a more pronounced effect on brand recall for micro-influencers may be because their perceived accessibility and reliability contribute to the authenticity of PSI [27,57]. This authenticity makes them pay more attention to the influencer’s social media content, which leads to a higher recall of the endorsed brand. Previous studies found that influencers with a small number of fans have a stronger persuasion effect [57,76]. It is noteworthy that the celebrity and “micro-influencer” used for testing in this study both have more than 70 million fans, and the effect of PSI is still prominent for the micro-influencer. Given that brand recall has a significant impact on consumers’ attitude and purchase intention, it follows that PSI is more significant for micro-influencer endorsements. Not only is this consistent with the findings of prior studies [57,58], but it also further enhances our comprehension of the underlying mechanisms that account for the increased influence of PSI in micro-influencer endorsements compared to traditional celebrities.

Last, the results indicate that there is no interaction effect between influencer type and product placement. Explicit product placement has better effects than implicit product placement in both traditional celebrity and micro-influencer endorsement. This may be attributed to the information-dense and highly commercial environment of social media [76], where implicit placements are frequently overshadowed by competing stimuli, thereby diminishing their capacity to foster brand recall and the subsequent advertising outcomes.

5.2. Industrial Implications

From a managerial perspective, recognizing the influence of PSI on the effectiveness of product placement can guide advertising strategies and campaign design. Micro-influencers and social media platforms should pay more attention to promoting PSI and PSR with followers. In addition to the quality of content, micro-influencers can strengthen PSI with fans through enhancing interaction, disclosure, or update frequency, which will further contribute to endorsement effectiveness and their own market value subsequently. On the other hand, micro-influencers also need to balance commercial content with content quality. Fans follow them for their useful or interesting information and not for advertising. Therefore, too much commercial content can damage fan trust, thereby harming their PSI, advertising effectiveness, and personal brand value.

Marketers should consider leveraging PSI by incorporating elements that foster a sense of connection and interaction with consumers. By understanding the positive effects of PSI on consumers’ attitudes and purchase intentions, practitioners can strategically utilize advertising content and product placement to enhance advertising effectiveness. In addition, when selecting brand endorsers, greater emphasis should be placed on the level of interaction between the influencer and their followers. This approach ensures that the selected endorsers have the ability to effectively engage with their audiences, thereby maximizing the impact of PSI in the endorsement process.

Additionally, the role of product placement prominence should not be overlooked. Advertisers should adopt explicit product placement within the highly commercial social media context since it is more effective compared to implicit placement, as this study reveals that PSI effects are stronger when the product placement is more prominent. Therefore, how to skillfully integrate advertising with the influencer’s content and whether to label the ads are issues that need to be considered and balanced in market practice.

5.3. Limitations and Future Research

This study has some clear limitations, which in turn suggest avenues for future research. First, while our use of a convenience sample from a specific Chinese online platform allowed us to efficiently collect data from active social media users, it potentially limits the generalizability of our findings. We chose this platform due to its popularity among our target demographic and its diverse user base, which aligns well with the typical audience for influencer marketing. However, we acknowledge that users of this platform may not be fully representative of the broader population. Also, using two female celebrities on a single social media platform may constrain the generalizability of our results. While this approach allowed for controlled comparison, it may not capture the full spectrum of influencer types and social media environments. Therefore, caution should be exercised when extrapolating these findings to broader contexts. Future research should aim to examine the relationships between variables by incorporating a wider range of celebrity types and encompassing various social media platforms, within different cultural contexts.

Although this study establishes a relationship between PSI and advertising effectiveness, it is important to note that causality cannot be definitively determined because PSI is not manipulated in this study. Therefore, alternative explanations or confounding factors cannot be ruled out. In reality, however, PSI is very difficult to manipulate in an experiment since it is a continuous variable. Future studies could use longitudinal approaches to establish more robust causal relationships.

Last but not least, this study only compares explicit and implicit product placement, whereas the product placement and message forms vary across social media. Researchers could further explore the influence of different forms of advertising on consumers’ responses, including but not limited to advertising avoidance and sharing intentions. By examining a broader range of advertising formats and their impacts on consumer responses, researchers can provide a more comprehensive understanding of the multifaceted dynamics at play in the realm of social media advertising. This would allow marketers to make informed decisions regarding their advertising strategies and optimize the results of their campaigns.