Revisiting opinion leadership in the digital realm: Social media influencers as proximal mass opinion leaders

Abstract

Social media influencers (SMIs) are ordinary people who rise to fame via social media. These individuals have repeatedly been labeled as opinion leaders, but often without in-depth theoretical reflection. We fill this gap by introducing a novel typology that allows for greater scrutiny in the identification of different types of opinion leaders in the modern media environment. Using this typology, we adequately capture—for the first time—opinion leadership as practiced by actors like SMIs, describing them as Proximal Mass Opinion Leaders (short: ProMOLs). We highlight that—despite their reach—ProMOLs exert a seemingly interpersonal and horizontal influence by engaging in personalized communication with opinion followers in self-built networks. Furthermore, we theoretically model the various flows of communication between ProMOLs, their communities, and other actors in the public sphere, thereby illustrating the multiple layers of ProMOLs’ impact. Finally, we discuss the model’s implications and suggest directions for future research.

Social media influencers (SMIs), popular personalities who gain fame via their successful self-presentation on social media (Piehler et al., 2022), have become an omnipresent part of (young) people’s digital media use. On platforms like Instagram or TikTok, SMIs build their personal brand around topics such as beauty (Brooks et al., 2021) or politics (Allgaier, 2020). Furthermore, SMIs casually work domain-specific recommendations to their audiences into discussions about everyday topics. In this sense, their advice-giving resembles opinion leadership, defined as influence “at the . . . level of ordinary, intimate, informal, everyday contact” (Katz and Lazarsfeld, 1955: 138).

Accordingly, the existing literature has repeatedly referred to SMIs as opinion leaders (e.g. Casaló et al., 2020; Farivar et al., 2021). Yet, the original opinion leader term may not be fully suited for SMIs, since it has traditionally been used to term influencers in interpersonal settings (e.g. Katz and Lazarsfeld, 1955), differing from the public influence that SMIs can engender. The concept of parasocial opinion leadership may initially seem more fitting (Stehr et al., 2015), as it describes how mass communicators can adopt an opinion leader role by tapping into the illusion of interpersonal contact to sway followers’ attitudes and behaviors. However, the concept was originally developed for personalities in mass media like traditional celebrities—who differ from SMIs in their origin of fame (Piehler et al., 2022)—and does not take the specific dynamics of social media into account, which allow for greater interactivity between media users and prominent media figures. SMIs in particular are known to make use of the affordances of platforms like TikTok or Instagram by repeatedly engaging in reciprocal communication with followers (Abidin, 2015), blurring the lines between social and parasocial contact (Lou, 2022).

To adapt the opinion leader concept to a media environment where forms of mass and personal communication co-occur (O’Sullivan and Carr, 2018), and in which diverse actors curate information and news for others (Thorson and Wells, 2016), a systematic categorization of different opinion leader types is needed. Such a typology would particularly help to account for novel phenomena like SMIs, who combine qualities of traditional opinion leaders and celebrities. We fill this crucial gap by first examining SMIs’ opinion leadership qualities to explore whether they generally match opinion leader traits. Second, we propose a novel categorization of different opinion leader types along two dimensions: personalized communication and status tier. We thereby, third, introduce a novel type of opinion leader, who can be placed at the center of the two dimensions and denotes self-branded media personalities such as SMIs: the proximal mass opinion leader (short: ProMOL). We highlight that ProMOLs are similar to traditional opinion leaders in that they rely on personalized communication and self-built networks to exert their influence, thereby potentially possessing persuasive advantages over other mass opinion leaders such as celebrities (Gräve and Bartsch, 2022), who we assign to the “distant mass opinion leader” type (short: DisMOL). Fourth, we theoretically model the multiple flows of communication in which ProMOLs are embedded (see Davis, 2009; Ognyanova, 2017), and which are characterized by (partial) reciprocity, collaboration, as well as accountability practices. We thereby highlight the multiple facets of ProMOLs’ influence which extend far beyond the constraints of the two-step flow of communication. Finally, we provide an agenda for future research, building on the insights of this work.

The opinion leadership concept

Opinion leadership originates from the two-step flow theory (Katz and Lazarsfeld, 1955), which postulates that mass media content is primarily relayed through opinion leaders rather than directly reaching and influencing recipients. These active and topically interested media users diffuse media messages to less invested recipients, the opinion followers, influencing their views through advice-giving (Lazarsfeld et al., 1948). Opinion leadership is carried out in the context of reciprocal exchange in social groups (e.g. in friend, work or local community groups; Keller and Berry, 2003) and exercised on an informal level (Katz and Lazarsfeld, 1955). Katz (1957) defined opinion leadership by three criteria: (a) representing values (“who one is”), (b) having a comparably higher number of ties than others (“whom one knows”), and (c) possessing high interest or knowledge on a topic (“what one knows”).

Stehr et al. (2015) integrated opinion leadership with parasocial phenomena. This literature assumes that media users can view media personalities (e.g. celebrities on TV) as intimate conversational partners or even friends, despite not actually engaging in social exchange or relationships with them (Horton and Wohl, 1956). While parasocial interactions refer to the user’s perception of being part of a “reciprocal social interaction” (Dibble et al., 2016: 23), a parasocial relationship describes “a longer-term association that . . . extends beyond the media exposure situation” (Dibble et al., 2016: 25). Stehr et al. (2015) found that parasocial relationships enable media personalities, like TV anchors, to influence audiences via specific opinion leadership functions such as providing orientation or sparking interest.

Yet, the advent of social media has again challenged the concept of opinion leadership. While Katz himself argued that opinion leadership should be retired due to identification issues (Katz and Fialkoff, 2017), the concept has regained relevance in the contemporary media environment, where information is passed on to people via social contacts within their online networks (Thorson and Wells, 2016). In the context of social media, curation by other social media users—meaning the selection, production, and filtering of information (Thorson and Wells, 2016)—significantly contributes to people’s news and information exposure (Bergström and Jervelycke Belfrage, 2018; Newman et al., 2024), implying that the role of opinion leaders still remains important today. Simultaneously, we lack clarity about which individuals classify as opinion leaders. Rather than retiring the concept, we thus argue that it is necessary to refine it to accommodate it to the modern media landscape. Moreover, opinion leadership needs to be revisited since not only peers and celebrities but also people dubbed as social media influencers (SMIs) can act as intermediaries during processes of curation and have thus been described as opinion leaders in research (Casaló et al., 2020)—even though they differ from these other personalities in central respects.

Opinion leadership qualities of social media influencers

Definitions of SMIs are diverse: some scholars describe them as third-party endorsers for organizations and advertisers (Enke and Borchers, 2019; Freberg et al., 2011), while others emphasize their large following (De Veirman et al., 2017) or influence potential (Gross and Wangenheim, 2018). A common denominator, especially in recent work, is that SMIs are ordinary people who become well known through social media (e.g. Abidin, 2015; Khamis et al., 2017; Piehler et al., 2022). Their success accordingly hinges on self-branding, that is, the presentation of a “unique selling point” to other social media users (Khamis et al., 2017: 191). In contrast, traditional celebrities (e.g. actors or athletes) gain visibility through achievements in industries that receive sustained media attention (Piehler et al., 2022). Based on their reach, SMIs are commonly grouped into nano- (⩾1000 followers), micro- (⩾10,000 followers), meso- (⩾100,000 followers), and macro-SMIs (⩾1,000,000 followers; Berne-Manero and Marzo-Navarro, 2020).

SMIs are frequently labeled as opinion leaders (e.g. Casaló et al., 2020; Martin and Sharma, 2022) and most SMIs also self-identify as such (Yaşa and Birsen, 2024). Audiences also tend to see SMIs as opinion leaders, regardless of their reach (Conde and Casais, 2023; De Veirman et al., 2017). Additionally, SMIs generally match central opinion leader criteria: In line with the original opinion leadership traits (Katz, 1957), SMIs first commonly represent certain values (e.g. pro-environmental values, Dekoninck et al., 2023), hold central positions in large networks (Uzunoğlu and Kip, 2014), and address mass audiences—meaning large, dispersed, and mostly anonymous groups of people. While not necessarily having formal expertise, they demonstrate autodidactically acquired topical knowledge (Dekoninck et al., 2023), corresponding with conceptualizations of opinion leaders as topically invested individuals who are distinguished from formal “leaders” and professionals (Jungnickel, 2018). Overall, and in line with prior research (Casaló et al., 2020; Farivar et al., 2021; Uzunoğlu and Kip, 2014), we thus argue that SMIs generally match attributes of opinion leadership.

Two key aspects to distinguish between opinion leaders

Although SMI meet traits indicative of opinion leadership (Katz, 1957), this term is also related to ordinary news-sharing individuals on social media (e.g. Turcotte et al., 2015), who diverge from SMIs in reach and are embedded in more personal communication contexts than SMIs. In addition, it has also been used for traditional celebrities (Bastien, 2009; Stehr et al., 2015), who not only have a different origin of fame (Lee et al., 2024), but are often also described as less communicative than SMIs (Jin et al., 2019). Based on these differences, we argue that two key opinion leader characteristics should be distinguished to differentiate between opinion leader types: their status tier and their level of personalized communication.

Status tier

A key idea in the literature is that opinion leadership involves horizontal influence among individuals of similar socioeconomic status, education, or professional achievement, distinguishing it from vertical influence exercised by people with exceptional status or power (Lazarsfeld et al., 1948; Weimann, 1994). While network size primarily reflects people’s gregariousness—an important trait of opinion leaders (Batinic et al., 2016)—it can also act as a symbol of status (Conde and Casais, 2023), and determine whether influence exertion takes place horizontally or vertically. Substantial network size and—at its highest end—public popularity distinguish potential opinion leaders from opinion followers, who typically lack sizable networks (e.g. extending beyond 1000 followers; Berne-Manero and Marzo-Navarro, 2020). Yet, even those people with a celebrity status—who are termed “mass opinion leaders” in the context of this work—can appear relatable to the regular person dependent on how they have built their network. Driessens (2015) highlights how stardom has shifted from an exclusive domain to one that appears accessible to anyone, as ordinary individuals today have opportunities to present themselves to public audiences online, relying on self-branding rather than institutional mediation in the process. Accordingly, persons who have reached celebrity status (largely) without the involvement of third parties (Piehler et al., 2022) may also appear more “within reach” for audiences, allowing some mass opinion leaders to develop near-horizontal influence despite their large networks (Bause, 2021). Consequently, we establish three different status tiers, which are informed not only by network size (personal network vs public popularity) but also by the origin of said network (self-built vs third-party-curated), leaning on literature surrounding public personalities’ origin of fame (e.g. Driessens, 2015; Khamis et al., 2017).

Status tier I

Opinion leaders in the first tier are well connected, but their networks can best be described as personal or local rather than public (Dubois and Gaffney, 2014). Networks are created and maintained by the opinion leaders themselves. Denoted as a chief opinion leader characteristic (Batinic et al., 2016), their gregariousness enables them to efficiently gain and maintain contact with opinion followers. Opinion leaders in this category are thus also most similar to their opinion followers, given that they lack (continued) public visibility and manage their social networks independently. When judging status by network size and network origin, their influence on others is therefore truly horizontal (Weimann, 1994).

Status tier II

Status tier II designates people who are at the center of large dispersed networks but have constructed their networks in a “do-it-yourself” manner (Driessens, 2015: 371). They have created their networks online via content creation (Khamis et al., 2017), attention labor (Brooks et al., 2021), and/or self-branding (Piehler et al., 2022); therefore, they use means that are practically available to any person. In this way, opinion leaders in this category are able to maintain a status of “ordinariness” despite the extraordinary popularity that they have reached (Hearn and Schoenhoff, 2016). They thus share some similarities with their opinion followers, suggesting that their influence may be more horizontal than vertical (Bause, 2021; Weimann, 1994).

Status tier III

Status tier III is assigned to opinion leaders who have access to large dispersed networks which have significantly been curated by third parties (such as TV stations or agencies; Hearn and Schoenhoff, 2016). Some people reach public popularity because they become well known for a talent they possess or a specific job they have, and are in the public spotlight due to the visibility that is granted to “their” media industries (e.g. film, sports; Khamis et al., 2017; Piehler et al., 2022). Aside from being reflected in the qualities or aspects that make them well known (e.g. being an outstanding footballer), the exceptionalism of opinion leaders in this tier is therefore highlighted by the complexity of their road to fame, which involves significant “institutional support” (Gräve and Bartsch, 2022: 596), attention from legacy media (Hou, 2019), and the continuous involvement of “gatekeepers” like producers and agents (Khamis et al., 2017: 198). The impact of opinion leaders in this category corresponds with vertical influence (Weimann, 1994), since they diverge from opinion followers not only in terms of network size but also with regard to their reliance on third parties to achieve public visibility.

Implications

We argue that opinion leaders’ status tier affects how they are perceived by opinion followers: Viewing potential implications through a social identity lens (e.g. Hogg, 2001), opinion leaders who belong to status tier I or II may be more readily accepted as part of opinion followers’ in-group than more psychologically distanced opinion leaders (i.e. those in status tier III). Correspondingly, a “lower” status tier (I or II) should reflect higher perceived similarity or homophily (Looi et al., 2023; Shehzala et al., 2024), which can in turn increase trust in opinion leaders’ advice (e.g. Naderer et al., 2021). In contrast, opinion leaders assigned to status tier III would likely be perceived as rather heterophilic (Shehzala et al., 2024), given that their status would then be fully discrepant from the status of the average opinion follower. Differences in status may also rub off on perceptions of opinion leaders’ attainability, wherein categorization in status tier I or II would translate to higher perceived attainability of the opinion leader, describing the extent to which people think they can become like their idols (Harff et al., 2025). This assumption is based on prior literature, which finds that factors such as smaller follower or network size can make opinion leaders seem not only more authentic (e.g. Park et al., 2021) but also more attainable than widely known personalities (Conde and Casais, 2023; Jin et al., 2019). Similarly, research suggests that public personalities who build their networks independently may seem more “within reach” than those relying on third parties for this purpose (Shehzala et al., 2024). Since opinion leaders are sought at a horizontal level (Weimann, 1994), information-seeking through “attainable” opinion leaders may be preferred.

Personalized communication

Besides taking into account status similarities, it is also important to consider communication strategies that opinion leaders (can) employ in changing media environments. On social media platforms, mass and interpersonal communication increasingly converge (O’Sullivan and Carr, 2018), allowing for intimate communication even between regular users and public figures. Here, we focus on categorizing opinion leader types rather than message types. Communicators share the same affordances when using a medium, but their use of these tools can vary. The degree of “personalized communication” reflects how effectively an opinion leader exploits a medium’s potential, making it a meaningful criterion to distinguish between opinion leader types. Still, the communication channel influences the level of possible personalization: streaming platforms (e.g. Twitch), for instance, enable interactivity and synchronization, while mass media have limited possibilities for personalization. Nevertheless, different types of communication are possible for different outlets, as many platforms nowadays support both mass and personalized communication simultaneously (McQuail, 2010; O’Sullivan and Carr, 2018).

While O’Sullivan and Carr (2018) consider personalization the degree to which communication is tailored to an individual rather than to a group, collective, or the public, we argue that communicators’ content can still be highly personal, even when they do not address specific individuals. We identify two key strategies opinion leaders can employ in this context:

1. Self-disclosure: Defined as revelations by communicators about their personal lives (Ferchaud et al., 2018), self-disclosure fosters emotional connections with opinion leaders (Hosman and Tardy, 1980) and characterizes an inviting communication style. For mass opinion leaders, self-disclosure offers followers “backstage” access to their lives, revealing personal aspects beyond their professional personae (Marwick and Boyd, 2011).

2. Reciprocal communication: Symmetrical mediated or interpersonal interactions are less common in the case of mass opinion leaders, who cannot partake in continued one-on-one exchange with individual opinion followers. However, reciprocity between popular personalities and their followers can also be observed at a more collective level, where the former interact with their communities in chats or via polls (Kowert and Daniel, 2021; Lou, 2022). This “collective reciprocity” (Lou, 2022) signals accessibility and even allows uninvolved audience members to vicariously engage with public figures (Hartmann, 2023).

Implications

We argue that the level of personalized communication determines the level of relational proximity or closeness between opinion leaders and opinion followers. Social relationships, where both parties know details about each other (Baek et al., 2013), are more emotionally intense than parasocial relationships (Eyal and Cohen, 2006), which lie at the low end of the proximity spectrum. Social relationships arise from continued reciprocal exchange and the sharing of personal details (Baek et al., 2013), while parasocial relationships result from one-to-many communication—as found in mass media—and lack actual reciprocity (Dibble et al., 2016). Social media, meanwhile, enable hybrid relationships combining parasocial and social elements due to increased interactions and reciprocity at a collective level. Lou (2022) proposes the term transparasocial relations for bonds between communicators and their communities, which are “marked by . . . collectively reciprocal, (a)synchronously interactive, and co-created interactions and relation development” (Lou, 2022: 6). These relationships involve audience interactions with mass opinion leaders, who respond to followers’ requests, comments, and feedback (Hartmann, 2023), forging the group context in which opinion leadership can best be exercised (Katz and Lazarsfeld, 1955). Relative to purely parasocial relationships, transparasocial relations can increase individual feelings of closeness to mass opinion leaders (Hartmann, 2023) and make them seem more accessible. This logic should in principle also apply to followers who are not actively involved with mass opinion leaders, who will expectedly also “feel more intimate” with the respective media persona by observing follower-leader interactions (Hartmann, 2023: 60). Overall, we thus contend that greater reciprocity strengthens followers’ attachment to opinion leaders by fostering a more personal relationship. Similarly, self-disclosure enhances knowledge about the other, supporting stronger relational ties (Hosman and Tardy, 1980). Higher relational proximity—a consequence of higher degrees of personalization in their communication style—should ultimately facilitate influence exertion via opinion leadership.

Toward a novel categorization of opinion leaders

Drawing from these two criteria, three different types of opinion leaders can be derived: (1) the opinion leader in the traditional sense (TradOL), who is assigned to status tier I and whose communication tends to be characterized by high personalization; (2) the distant mass opinion leader (DisMOL), who is lowest in both relational and status proximity, given a low propensity to engage in personalized communication and a classification in the most “unattainable” status tier; and (3) the proximal mass opinion leader (ProMOL), who lies in between these two types on both dimensions (see Figure 1). As we will illustrate below, different variations of status tier and personalized communication are possible within these categories, highlighting that the relationship between both dimensions is not strictly linear. We also acknowledge that there may be special cases of opinion leaders, who cannot clearly be assigned to one category, such as those whose success stems from both social and legacy media. Nonetheless, the following categorization should apply to the overwhelming majority of opinion leaders who can be found in and outside of (social) media.

Figure 1.

Categorization of opinion leaders along two axes.

Type 1: the traditional opinion leader (TradOL)

With the term traditional opinion leader (TradOL), we refer to the classic opinion leader type (Katz and Lazarsfeld, 1955; Lazarsfeld et al., 1948) who engages in reciprocal exchange with opinion followers and holds social relationships with them. TradOLs’ influence exertion is not limited to the offline context; they can also act as opinion leaders for personal contacts or in smaller groups/communities on social media (Bergström and Jervelycke Belfrage, 2018; Dubois and Gaffney, 2014; Turcotte et al., 2015). TradOLs are assigned to status tier I, as these opinion leaders lack (significant) public visibility and rely on self-built private or local networks to exert their influence (Dubois and Gaffney, 2014; Turcotte et al., 2015). They notably do not reach large dispersed audiences; instead, they benefit from being well positioned in the smaller social groups that they are part of. Although some people may be more inclined toward this role than others (Batinic et al., 2016), TradOLs typically assume this function contextually.

Examples of TradOLs include close friends, who expectedly engage in highly personalized communication. Good acquaintances who are popular on social media would also score high on this dimension, but have a slightly more disparate status due to their popularity. Extended family members who mostly act as opinion leaders through communication in group chats would again fall into the status tier I but likely communicate in a less personalized manner than other TradOLs. Small-scale activists in online groups or local communities and organizations can also be seen as TradOLs: While their influence may then reach beyond intimate social circles (i.e. friend groups, family), it is restricted to an audience that is private in the sense that it is still locally concentrated and thus not dispersed (Dubois and Gaffney, 2014; Keller and Berry, 2003).

Type 2: the distant mass opinion leader (DisMOL)

The distant mass opinion leader type builds upon Stehr et al.’s (2015) parasocial opinion leader, describing personae in legacy media, such as athletes and TV presenters. Interactions and relationships with them likely remain largely parasocial, since these actors are described as less responsive than SMIs (e.g. Lou, 2022; Xu et al., 2023). However, DisMOLs can in some cases also be relatively close with their audiences, specifically when they are more active on social media. Yet, research finds that only very few people report actually receiving responses to their comments from traditional celebrities whom they follow online (e.g. Bond, 2016). As a second criterion that influences the level of personalized communication, traditional celebrities have been found to showcase comparably low levels of self-disclosure in their messages (Eyal et al., 2020).

Also with regard to status, DisMOLs distance themselves from opinion followers: Since they are famous due to their extraordinary achievements or skills (Khamis et al., 2017)—leading us to classify them into status tier III—they may be perceived as out of reach (Jin et al., 2019). Similarly, Lee and colleagues (2024: 3) contend that “celebrity endorsers transfer their existing fame to social media, which arguably can make them feel less relatable” than personalities like SMIs. This argument highlights the notion that DisMOLs’ origin of network—being well-known outside of social media—may result in the perception that they are out of touch with the ordinary person and discrepant in status from opinion followers (see also Shehzala et al., 2024).

Examples of DisMOLs include TV anchors, whose high status and potentially lacking social media presence limits personalized communication with opinion followers. Reality TV stars also count as DisMOLs may have a more personalized output due to their social media activities and verge more closely on status tier II, as they are “ordinary people” selected by media industries. Some celebrities (who could be dubbed “SMI-inspired celebrities”) have also increasingly adopted micro-celebrity techniques, borrowing from the influencer culture, which might lead to a higher degree of personalized communication, while they nonetheless remain in status tier III due to their origin of fame.

Type 3: the proximal mass opinion leader (ProMOL)

The proximal mass opinion leader (ProMOL) can be placed between the DisMOL and TradOL on both dimensions and describes personalities like SMIs, who become well-known through self-branding on social media. ProMOLs reach a mass audience and thus have a bigger network than TradOLs. Yet, compared to DisMOLs, ProMOLs are more prone to adopting techniques to help build and maintain a close relationship with their audiences (Brooks et al., 2021), such as focusing on self-disclosure or engaging in regular exchange with their followers (e.g. by reacting to comments; Ferchaud et al., 2018; Lou, 2022). Ouvrein (2024) has also advanced the notion that these two aspects of personalized communication constitute key features in SMIs’ communication that allow them to build social capital. Providing further evidence for this point, Xu et al. (2023) showed that followers perceive SMIs as significantly more responsive to audiences’ comments and feedback than traditional celebrities. Additionally, SMIs are known to actively involve followers in the creation of content (Lou, 2022) and are likely to share aspects of their personal lives with followers (Ferchaud et al., 2018). In consequence, ProMOLs’ communities may form relationships with them that oscillate between social and parasocial (Kowert and Daniel, 2021; Lou, 2022). Defined by partial reciprocity, ProMOLs’ communities resemble the “primary groups” of TradOLs (Katz and Lazarsfeld, 1955), with ProMOLs taking central positions in closely affiliated networks (Kowert and Daniel, 2021). However, we highlight here that not all relationships between ProMOLs and their audience members are necessarily strictly “transparasocial.” ProMOLs who have ventured into other industries (such as music or fashion) and reduce their focus on social media activities may personalize their communication less than others, rendering relationships with them rather parasocial.

With regard to status tier, proximal mass opinion leaders (ProMOLs) can uniformly be conceptualized as ordinary people who have gained an extraordinary status (i.e. public visibility) through self-branding on blogs or social media platforms (Khamis et al., 2017). ProMOLs’ source of public visibility stands in contrast with the primary reason for DisMOLs’ success: ProMOLs gain popularity largely without a dependency on being selected in industries with high media attention (such as film, sports; Driessens, 2015).Instead, ProMOLs’ public networks are largely self-curated (Driessens, 2015) and the result of continuous “attention labor” (Brooks et al., 2021: 529), allowing them to bypass “the gatekeeper role of media and entertainment industries” (Hou, 2019: 535). This alternative and ostensibly accessible road to fame may make them seem more relatable than DisMOLs (Jin et al., 2019; Lee et al., 2024; Shehzala et al., 2024). Therefore, ProMOLs can be assigned to a “lower” status tier than DisMOLs, aligning with literature suggesting that audiences perceive SMIs as “normal” people (e.g. Lehto, 2022; Martensen et al., 2018) with whom they share a reality (Dekoninck et al., 2023). This ordinariness is further communicatively constructed, as SMIs commonly emphasize having “the same needs and motives as their followers” (Giles, 2018: 158). Being closer to opinion followers in status may explain why audiences seem to consider SMIs to be more similar to themselves (Gräve, 2017; Schouten et al., 2020) and more attainable (Harff et al., 2025; Martensen et al., 2018) than traditional celebrities.

We note that the classification on the status tier dimension may slightly diverge between nano-, micro-, meso-, and macro-SMIs. Arguably, smaller SMIs will be perceived as more ordinary and attainable, given that they are themselves still in the process of gaining celebrity status and have a lower network size (Jin et al., 2019). In such cases, it may even be difficult to draw boundaries between ProMOLs and TradOLs, since nano-SMIs, for instance, only have comparably small reach. Here, it becomes important to consider whether these personalities personally know and/or engage with the majority of their followers; if so, they better qualify as TradOLs, given that their influence potential then primarily rests in interpersonal connections, rather than the targeting of a public network. In contrast, micro-SMIs’ networks (⩾10,000 followers) are arguably large and dispersed enough to qualify as “public.”

Examples within the ProMOL category include various types of SMIs who can rank slightly differently on the two dimensions. That is, micro-SMIs with a focus on interactive content will rank high on the personalized communication dimension (due to their emphasis on interactive content) and verge on status tier I due to their comparably small network size. Macro-SMIs will generally edge closer to status tier III due to their high network size. Meanwhile, regarding personalized communication, we would also expect macro-SMIs to be responsive and share personal insights with audiences, as SMIs regardless of their size are known for inviting collective reciprocity and co-creation as well as for self-disclosing to followers (Ferchaud et al., 2018; Lou, 2022). Yet, more “achieved” macro-SMIs who have possibly ventured out into other media industries would likely personalize their content to a lesser extent than others. Bloggers may also be counted as ProMOLs, since their success is equally ascribed to self-promotion (Duffy and Hund, 2015). We mention them in addition to SMIs, because their online presence is typically based on a personal blog or a website rather than social media.

ProMOLs and multistep flows of communication

With the conceptualization of ProMOLs as a new opinion leader type comes the need to remodel communication flows between these opinion leaders and other actors in their environment. Figure 2 illustrates that the two-step flow of communication is generally upheld (see bold arrows), since ProMOLs, much like TradOLs, function as individuals who relay media messages to recipients (e.g. Martin and Sharma, 2022). However, on social media, and specifically in the case of ProMOLs, the original two-step flow of communication is extended in several ways. Though the model shares some assumptions with existing work on multistep flows of communication (Davis, 2009; Ognyanova, 2017), it includes several novel considerations, such as the potential of ProMOLs to engage in reciprocal exchange not only with opinion followers, but also with actors at the public level (e.g. journalists/the media, parties etc.; Allgaier, 2020). While the model may in part apply to other opinion leaders, we want to emphasize that many of the model’s aspects may be specific to ProMOLs, such as the role of co-creation in their content production.

Figure 2.

Flows of communication among ProMOLs, their communities, and actors in the public sphere.

Like in the original two-step flow of communication, the model describes opinion leaders as intermediaries of messages who convey information from mass media to recipients (Martin and Sharma, 2022). However, ProMOLs do not only contextualize and relay news, but also fulfill certain functions for followers when they transmit content. Following Stehr et al. (2015), our model highlights opinion leadership functions such as providing orientation, simplification, and interest arousal. These functions are also central to the role that their followers ascribe to ProMOLs (Wunderlich, 2023), fulfilling them in the context of a perceived “peer-to-peer-relation” (Enke and Borchers, 2019: 266).

We also acknowledge in our model that affordances of social media have changed dynamics of social curation. In “reactions” on YouTube and “stitches” on TikTok, ProMOLs can reply to content from other actors, making audiences witness their seemingly authentic and spontaneous responses to media messages (Harff and Schmuck, 2024b; Vizcaíno-Verdú and Abidin, 2023). On Twitch, some ProMOLs even watch news together with their audiences (Grayson, 2022), thereby “live-curating” content. Such reactions transform the original understanding of the two-step flow of communication, in that followers simultaneously come into contact with original media messages and opinion leaders’ thoughts on the presented topics (Harff and Schmuck, 2024b).

Additionally, opinion leaders can today draw from a multitude of sources when discussing topics. They do not only pass on content from legacy media to opinion followers, but also directly comment on messages from domain-specific actors, the mass media, and other mass opinion leaders and their communities (Turcotte et al., 2015). Our model also recognizes that the flow of communication between both parties is not unidirectional; ProMOLs can also respond directly to actors in the public sphere. When ProMOLs share content from other actors in the digital public sphere, they—due to their continuous reach (which TradOLs lack)—make their attitudes toward this content visible to these primary sources. For example, they can hold these actors accountable, potentially prompting content or PR strategy adjustments (Allgaier, 2020).

Yet, collaborations between ProMOLs and other actors in the public sphere are also often established. ProMOLs can be strategically used as opinion leaders by other actors in the digital public sphere to spread their messages, for example, in the context of get-out-the-vote campaigns (e.g. Goodwin et al., 2020).

Moreover, social media have allowed opinion leaders to (further) interact with opinion followers. As previously argued, ProMOLs likely make use of this opportunity more than DisMOLs do. Interactions are in part characterized by reciprocal accountability practices: ProMOLs hold their followers accountable for how they behave within and outside of their communities, for example, criticizing their followers and other social media users in posts about online hate when they have become the target of uncivil comments (Ouvrein et al., 2023). Meanwhile, in reactions to ban appeals, ProMOLs comment on requests of followers to get unbanned who were excluded from their channels due to misconduct (Thach et al., 2022), which establishes group norms and may thus shape followers’ perceptions of values and actions that are (in)tolerable. On the other hand, followers can also hold ProMOLs accountable for their actions and posts (Lehto, 2022).

In line with Lou (2022), ProMOLs moreover push collective interaction, describing “reciprocity . . . in a collective sense, namely accommodating to collective demands from the followers” (p. 12). Thereby, ProMOLs show their willingness to take audiences’ suggestions and interests into account. Similar to enabling collective reciprocity, ProMOLs let their audiences co-create content by actively involving them in the production of posts, for example, when their comments are shared in ProMOLs’ content.

Through collective reciprocity and co-creation (Lou, 2022), ProMOLs furthermore encourage interaction among community members, for example, in chats or comment sections, stimulating discussion among followers (Harris et al., 2023). ProMOLs may also be able to contact TradOLs among their audience members, who in turn exert influence within their personal networks. Last, ProMOLs’ communities possess potential for collective action, which may target actors in the public sphere. In this context, ProMOLs can also function as mouthpieces of their communities (Harris et al., 2023).

Summary and agenda for future research

In the digital media environment, where boundaries between types of communications and media formats increasingly blur, our novel conceptualization allows for a systematic placement of opinion leaders along two dimensions. The categorization into three opinion leader types helps to structure existing literature, which has used the concept of opinion leadership to describe very different groups of people who are active advice-givers, ranging from SMIs (e.g. Casaló et al., 2020) over celebrities (e.g. Bastien, 2009) to regular people (e.g. Bergström and Jervelycke Belfrage, 2018). Our conceptualization also enables scholars to categorize actors like virtual influencers (e.g. LilMiquela) or virtual pop stars (e.g. Hatsune Miku) as opinion leaders, who, despite not being human, offer differing degrees of personalized communication, enabling followers to build (trans)parasocial relationships with them, and can be distinguished from each other based on network size and origin.

Most importantly, this typology highlights that, despite their reach, SMIs exhibit two strengths originally associated with “TradOLs” only: first, due to their focus on highly personalized communication, relationships developed with ProMOLs may become increasingly personal (Lou, 2022), and their opinion leadership thus shares some conditions with the context of interpersonal influence (Katz and Lazarsfeld, 1955). Second, ProMOLs’ source of public visibility may make them seem attainable and similar to opinion followers in status (Enke and Borchers, 2019; Harff et al., 2025; Martensen et al., 2018), which is important against the background that influence on others’ attitudes and behaviors is often successfully achieved by equal-leveled peers (Lazarsfeld et al., 1948; Turcotte et al., 2015).

From this perspective, we encourage scholars to continue researching SMIs from the lens of opinion leadership, while being more precise regarding their differences to other opinion leader types. On social media, various communicators are present who may exert an influence on users’ opinion formation. In a first step, our typology should thus be employed to distinguish between communicators based on manifest characteristics, namely status tier and the level of personalized communication. This distinction is vital because it has implications for understanding opinion leaders’ potentially differential impact on opinion formation processes. In a second step, followers’ perceptions that are shaped by status tier (i.e. perceived similarity and attainability) and level of personalized communication (i.e. relational proximity) should be examined. These variables may help understand and explain the actual influence of SMIs/ProMOLs versus other communicators on opinion followers (Dekoninck and Schmuck, 2025; Harff and Schmuck, 2024a). Because our assumptions mainly stem from qualitative research, future research should strive to make systematic quantitative comparisons between the different opinion leader types to test, for example, whether ProMOLs indeed have a relative advantage over DisMOLs in changing attitudes and behaviors of opinion followers. Studies considering potential effects of being exposed to different opinion leader types could build on multi-wave data, so that relationships between variables can be investigated over time.

Panel studies would also be ideal to account for and examine the transactional nature of the communication flows outlined in this article. Building on our novel theoretical model, future research could also study how different communication flows impact ProMOLs’ self-presentation and content production, for example by drawing from in-depth interviews with ProMOLs. Case studies could be useful to investigate content creation involving both ProMOLs and their followers, or gain insights into reciprocal accountability practices between ProMOLs and actors in the public sphere (such as journalists or politicians; Allgaier, 2020). Meanwhile, collaborations with ProMOLs in the form of public campaigns can be further explored through content analytical research. Finally, (quasi-)experimental research may be suitable when wanting to explore the impact of novel formats—such as reactions or stitches—on receptivity to ProMOLs’ content and advice.