Influencer advertising on Instagram: product-influencer fit and number of followers affect advertising outcomes and influencer evaluations via credibility and identification

Abstract

Social media influencers are increasingly employed as product endorsers, and a growing body of academic research confirms that influencers are an effective advertising instrument. However, more research is needed on the specific influencer characteristics driving this success, and the processes responsible for these effects. In this study, we investigated to what extent product-influencer fit and number of followers interact in contributing to positive advertising outcomes and influencer evaluations. Moreover, we investigated to what extent perceived credibility of the influencer and identification with the influencer mediate these relationships. We conducted a 2 (poor vs. good product-influencer fit) X 2 (moderate vs. high number of followers) between-subjects experiment among 432 Dutch Instagram users. Participants were exposed to Instagram posts of health and fitness influencers who endorsed either a protein shake (good-fit) or ice cream (poor-fit). Results showed that fit and number of followers seem to work in tandem: although influencers with a high number of followers are liked more, and their endorsements result in a more positive attitude toward both the ad and product, and a greater likelihood to buy the advertised product as compared to influencers with a moderate number of followers, the endorsed product should fit the influencer’s self-branded image for these positive effects to occur. Moreover, effects were all mediated by perceived credibility and identification, which appear to be important drivers of influencer endorsement effects. Together, our findings contribute to a growing body of knowledge on the processes driving audience responses to influencer marketing, and provide clear guidelines for practitioners.

Keywords:

• Social media influencers
• influencer endorsement
• product-influencer fit
• number of followers
• credibility
• identification

Introduction

Companies are increasingly using social media influencers to promote their products or services (Mediakix 2017). Influencers, also called ‘micro-celebrities’, have gained fame through strategic self-presentation on social media (Khamis, Ang, and Welling 2017). These self-branded ‘fitfluencers’, ‘travelbloggers’, or ‘beautyvloggers’ have built large networks of followers on platforms like YouTube, TikTok and Instagram, and their online word-of-mouth has become an influential marketing instrument (Campbell and Grimm 2019; Chae 2018; Ki et al. 2020; Lin, Bruning, and Swarna 2018). In contrast to traditional celebrity endorsers, such as athletes and movie stars, social media influencers are perceived as more credible and relatable sources of consumer information, especially among a young audience (e.g. Lou and Yuan 2019; Schouten, Janssen, and Verspaget 2020; Trivedi and Sama 2020).

A growing body of research seems to confirm that influencers are effective brand endorsers and academics have started to investigate which factors specifically contribute to their success (e.g. Breves et al. 2019; Chapple and Cownie 2017; De Cicco, Iacobucci, and Pagliaro 2021; De Veirman, Cauberghe, and Hudders 2017; Djafarova and Rushworth 2017; Kim and Kim 2021; Lou and Yuan 2019; Schouten, Janssen, and Verspaget 2020; Torres, Augusto, and Matos 2019; Trivedi and Sama 2020). For academics as well as practitioners in the field of influencer marketing, it is relevant to know which influencer and endorsement characteristics are the main drivers of endorsement effectiveness, both in terms of advertising outcomes (such as more positive product attitudes and higher purchase intentions), and in terms of influencer evaluations (e.g. increased liking).

In prior work, different characteristics of influencers and endorsements have been suggested to contribute to endorsement effectiveness, such as influencer credibility, similarity, and attractiveness (Lou and Yuan 2019; Torres, Augusto, and Matos 2019), number of followers, type of product endorsed (De Veirman, Cauberghe, and Hudders 2017), and the fit between the influencer and the endorsed brand (Breves et al. 2019; De Cicco, Iacobucci, and Pagliaro 2021; Kim and Kim 2021; Schouten, Janssen, and Verspaget 2020; Torres, Augusto, and Matos 2019). However, as evidenced by a recent systematic literature review on influencer marketing effects (Vrontis et al. 2021), to date only a relatively small number of studies have experimentally tested the causal impact of such characteristics on advertising outcomes and influencer evaluations. We believe that a more systematic investigation of the predictive value of these factors, and especially their interactions, would advance our knowledge on the main drivers of influencer endorsement effects. In addition, this knowledge would offer practitioners more tangible tools for selecting successful influencer-brand collaborations.

Specifically, in the present study, we experimentally investigate the interplay between two factors that have been individually identified in previous research as potentially important drivers of influencer endorsement effects: product-influencer fit and the number of followers influencers have on a social media platform (e.g. Breves et al. 2019; Kay, Mulcahy, and Parkinson 2020; Kim and Kim 2021; De Veirman, Cauberghe, and Hudders 2017). As evidenced by the online promotion of influencer marketing agencies (e.g. Join 2021; Mooitheagency 2021; Onemedia 2021), and interviews with brand and digital agency representatives (Uzunoğlu and Kip 2014), the most important criterion for brands selecting influencers for their campaigns appears to be a good fit between the influencer and the endorsed brand. Academic studies indeed confirm that, in line with traditional celebrity endorsement effects (e.g. Kamins and Gupta 1994), a good fit between the characteristics of influencers and the products they endorse is an important determinant of endorsement success (Breves et al. 2019; De Cicco, Iacobucci, and Pagliaro 2021; Kim and Kim 2021; Schouten, Janssen, and Verspaget 2020; Torres, Augusto, and Matos 2019). However, good alignment between an endorser and product or brand characteristics may be more important for specific types of influencers than for others.

As recognized by a recent literature review on influencer marketing effects (Vrontis et al. 2021), there is a need for research into consumer responses to endorsements of different types of influencers, which are generally distinguished based on their follower count (Campbell and Farrell 2020). In the Netherlands, specifically, we distinguish ‘mega influencers’ (over 1 million followers), ‘macro influencers’ (100,000–1 million followers), ‘meso influencers’ (50,000–100,000 followers), ‘micro influencers’ (5000–50,000 followers), and ‘nano influencers’ (100–5000 followers; Haverkamp 2018; Joosten 2021). Although there is reason to believe that influencer endorsement effects, and specifically the impact of product-influencer fit, depend on the number of followers (e.g. De Veirman, Cauberghe, and Hudders 2017; Mooitheagency 2021), to our knowledge, their interplay has not yet been investigated. For academics, as well as for practitioners it would be valuable to know whether dealing with specific types of influencers requires better alignment of endorser and product characteristics than working with others.

Hence, to contribute to a better theoretical understanding of influencer endorsement effects, and provide practical advice for brands, influencers, and influencer marketing agencies that can be applied in future campaign development, we will investigate how product-influencer fit and number of followers together affect endorsement effectiveness. We will investigate these effects in the context of Instagram, one of the most popular platforms for influencer marketing (Mediakix 2017). To measure advertising outcomes, the present research builds upon existing (influencer) marketing literature by assessing consumers’ attitude toward the advertisement, attitude toward the product, and intention to purchase the endorsed product (e.g. Fink, Cunningham, and Kensicki 2004; Schouten, Janssen, and Verspaget 2020; Till and Busler 2000). In addition, since an influencer marketing campaign also affects the image of influencers, who need to manage their personal branding and maintain authenticity among their fans (Audrezet, Kerviler, and Moulard 2020), we will assess influencer likeability perceptions (cf. De Veirman, Cauberghe, and Hudders 2017).

Moreover, to gain more insight in the underlying processes, we investigate the role of two potential mediators of the effects of fit and number of followers: perceived endorser credibility and identification with the influencer. Specifically, previous research has shown that endorsers are perceived as more credible and followers are more likely to identify with them when they advertise products that fit their expertise (e.g. Breves et al. 2019; Djafarova and Rushworth 2017; Lee and Koo 2015; Schouten, Janssen, and Verspaget 2020), and number of followers also seems to significantly affect both credibility and identification (e.g. Djafarova and Rushworth 2017; Lin, Bruning, and Swarna 2018; Uzunoğlu and Kip 2014). In turn, both credibility and identification have been shown to affect endorsement outcomes (e.g. Bond and Drogos 2014; Fink, Cunningham, and Kensicki 2004; Djafarova and Rushworth 2017; Lou and Yuan 2019; Uzunoğlu and Kip 2014), further increasing the relevance of investigating their potentially mediating roles.

Literature review

Explaining the success of influencer advertising

The use of social media platforms like YouTube, TikTok, and Instagram keeps rising. In the Netherlands, the video-sharing social network TikTok has 1.7 million users, and its use increased by 149% since 2020 (Newcom 2021). Photo- and video-sharing social networking site Instagram is also still growing, with 5.9 million users of 15 years and older in the Netherlands alone (Newcom 2021). Given the popularity of these social media platforms, it is not surprising that influencers have become the new opinion leaders among youngsters, serving as brand ambassadors for a diverse range of products and services, and an important source of consumer product information (Lin, Bruning, and Swarna 2018).

Mirroring the upsurge of influencer endorsements on social media, a growing number of academic studies has demonstrated the positive impact of these endorsements on advertising outcomes. For example, Lee and Watkins (2016) showed that fashion vloggers who promote luxury brands on their YouTube channels positively affected consumers’ perceived brand value and purchase intentions. In addition, an interview study by Djafarova and Rushworth (2017) revealed that Instagram users regularly bought or recommended products reviewed by ‘Instafamous’ personalities. The success of social media influencers as product endorsers is also evidenced by more positive evaluations of influencer endorsements as compared to company-sponsored promotions and ‘traditional’ celebrity endorsements. For example, in a study by Colliander and Dahlén (2011), blog posts about fashion brands resulted in higher brand attitudes and increased purchase intent in comparison to company-sponsored online magazine articles. Recent studies comparing influencer endorsements to traditional celebrity endorsements showed that influencers are more effective endorsers than traditional celebrities (such as fashion models and professional athletes), resulting in more positive brand attitudes and purchase intentions (Schouten, Janssen, and Verspaget 2020; Trivedi and Sama 2020).

Influencers thus seem to be an effective advertising instrument, and research has started to identify the factors that contribute to their success. Specifically, the persuasive power of influencers seems to originate from their unique positioning as authentic, relatable, and accessible ‘superpeers’. Not only do consumers actively seek out influencer content (as compared to traditional marketing communication which is often considered intrusive), but influencers are also perceived as credible and relatable endorsers (Chapple and Cownie 2017; Djafarova and Rushworth 2017). Even though most consumers are aware that influencers profit from their endorsements, this does not seem to negatively affect credibility perceptions (Chapple and Cownie 2017; Dhanesh and Duthler 2019). As compared to traditional celebrity endorsers, influencers are perceived as more trustworthy, consumers feel more similar to them, and are more likely to aspire to be like them. These processes, in turn, positively affect advertising outcomes (Lou and Yuan 2019; Schouten, Janssen, and Verspaget 2020). In sum, both perceived endorser credibility (which consists of the subcomponents trustworthiness and expertise) and identification with the influencer (consisting of perceived similarity and wishful identification) seem to be important drivers of influencer endorsement effects.

Despite these increasing insights, much remains to be explored to better understand the value of influencers as an instrument for marketing communication. To extend current theorizing and advise practitioners, the present study aims to investigate to what extent credibility and identification underlie the effects of two key influencer endorsement characteristics that practitioners use to select influencers for a commercial campaign: the fit between influencers and the products they endorse, and the number of followers influencers have on a social media platform.

Product-influencer fit

The ‘fit’ (or ‘match-up’) between the image or personality of an endorser and the product he or she endorses is well-established in the traditional advertising domain, where it is considered one of the most important aspects influencing brand attitudes (Fink, Cunningham, and Kensicki 2004; Kamins and Gupta 1994; Till and Busler 2000). Originally, fit was defined as the match-up between the physical attractiveness of an endorser and the degree to which the endorsed product was attractiveness-related (Kamins 1990). In later research, match-up was defined along other dimensions, including the expertise of the endorser regarding the type of product (Till and Busler 1998), congruency between brand personality and endorser personality (Mishra, Roy, and Bailey 2015), shared cultural values (Choi, Lee, and Kim 2005), and the general symbolic meaning that the endorser may transfer onto the brand (McCracken 1989). In general, fit refers to congruence between certain characteristics of the endorser and the characteristics of the products endorsed (Erdogan 1999; Kamins and Gupta 1994).

According to Schouten, Janssen, and Verspaget (2020), fit may even be more important for influencer endorsements than traditional celebrity endorsements. Influencers brand themselves as representative or expert within a particular niche (e.g. ‘fitfluencer’), so their influencer brand personality may more naturally match brands and products endorsed within their niche. Recent studies demonstrated that indeed also in the context of influencer endorsements, consumers evaluate a product or brand more positively and show more interest in a purchase when an influencer’s characteristics match those of the product endorsed (Breves et al. 2019; Kim and Kim 2021; Schouten, Janssen, and Verspaget 2020; Torres, Augusto, and Matos 2019). Specifically, in a survey study, Torres et al. found positive relationships between influencer-brand congruence and attitudes toward the endorsement, brand, and purchase intentions. Breves et al. (2019) found positive fit effects on brand evaluations and intentions to recommend the endorsed brand to friends. Kim and Kim (2020) found that well-fitting influencer endorsements (e.g. a health influencer promoting organic cleansing juice) resulted in more positive product attitudes than ill-fitting endorsements (e.g. a health influencer promoting a high-calorie milk shake), and Schouten, Janssen, and Verspaget (2020) demonstrated more positive ad attitudes and higher purchase intentions when influencer characteristics corresponded to the endorsed product (e.g. a food influencer promoting a stand mixer vs. a food influencer promoting a watch).

The present research aims to replicate and experimentally test the robustness of the fit-effect on advertising outcomes, in the context of influencer endorsements on Instagram. Moreover, as fit may also be an important condition for influencers to accept a brand partnership (Audrezet, Kerviler, and Moulard 2020), we aim to investigate to what extent fit affects perceived influencer likeability. To our knowledge, only one study to date investigated whether the positive effect of fit transfers to influencer evaluations. Specifically, De Cicco, Iacobucci, and Pagliaro (2021) showed that influencer-product congruence positively affected people’s attitude towards the influencer and intention to follow the influencer in the future. We will investigate whether this effect replicates for influencer likeability. Specifically, we will test the following hypotheses:

H1a: Influencers endorsing a product that fits well with their image will ensure more positive advertising outcomes (measured as attitude toward the ad and product, and purchase intention) as compared to influencers endorsing a product that poorly fits their image.

H1b: Influencers endorsing a product that fits well with their image will be perceived as more likeable as compared to influencers endorsing a product that poorly fits their image.

Number of followers

We aim to investigate the effects of product-influencer fit in relation to influencer type (Vrontis et al. 2021). Influencers are generally distinguished based on their number of followers, ranging from nano influencers with several hundreds of followers, to mega influencers with more than a million followers (Boerman 2020; Campbell and Farrell 2020). According to the ‘popularity principle’ on social media, quantity as expressed in terms of likes, friends, or followers has value, because the larger the influencer’s network, the larger the reach of a message (Van Dijck 2013). Consumers also tend to use number of friends or followers as a heuristic cue for judging the merits of the source and message content (e.g. Djafarova and Rushworth 2017; Jin and Phua 2014; Utz 2010). In line with these observations, De Veirman, Cauberghe, and Hudders (2017), who compared user responses to Instagram influencers with a moderate number of followers (2100) and a high number of followers (21,200), found that the last were perceived as more popular and more likeable. In addition, in earlier work focusing on traditional celebrities, Jin and Phua (2014) demonstrated that consumers were more willing to buy a product endorsed by a celebrity on Twitter when this person had a high number of followers, as opposed to a low number of followers.

Research findings thus seem to imply that influencers with a large follower base are more successful product endorsers. However, to our knowledge, the impact of number of followers on advertising outcomes, such as brand attitude and purchase intentions, has not yet been investigated. A recent study did show that an influencer Instagram post with a high number of Likes resulted in higher intentions to purchase the endorsed product than the post with a low number of Likes (Kay, Mulcahy, and Parkinson 2020). Although a large reach seems intuitively appealing, the effects of follower count may not be as straightforward as they seem. Dutch market research showed that consumers nowadays are more likely to trust product endorsements of influencers that operate on a smaller scale (Reputatiefabriek 2019), which is also acknowledged by influencer marketing agencies (e.g. Mooitheagency 2021). In addition, Westerman, Spence, and Van Der Heide (2012) showed that both too many and too few followers can negatively affect people’s perceptions of a social media account, as compared to a moderate number of followers.

Because of these mixed findings, we believe it is important to explore the effect of number of followers on advertising outcomes, as well as test to what extent the positive effect on influencer likeability, as found by De Veirman, Cauberghe, and Hudders (2017), still holds in the dynamic field of influencer marketing. Specifically, we aim to answer the following research question:

RQ1: What is the effect of the number of followers on advertising outcomes (attitude toward the ad and product, and purchase intention) and perceived influencer likeability?

Credibility and identification as underlying processes

To gain more insight in the underlying processes of the effects of product-influencer fit and number of followers on both advertising outcomes and influencer likeability, we will investigate the mediating role of perceived influencer credibility and identification. Influencer credibility is the extent to which consumers view the influencer as a reliable expert with regards to a product or service (Djafarova and Rushworth 2017), and has been previously measured with the two subcomponents of source credibility: trustworthiness and expertise (Sternthal, Phillips, and Dholakia 1978, see Schouten, Janssen, and Verspaget 2020). More specifically, trustworthiness entails perceptions of integrity, honesty, and believability of an endorser and expertise refers to the relevant skills, knowledge, or experience an endorser is thought to possess (Erdogan 1999). The extent to which one identifies with an influencer, on the other hand, has been previously measured with the subcomponents perceived similarity, the extent to which individuals perceive to have things in common with the endorser, and wishful identification, which entails the desire to be like the other person (Hoffner and Buchanan 2005; see Schouten, Janssen, and Verspaget 2020).

Traditional celebrity endorsers are generally perceived as more credible when they advertise products that fit their expertise (Dwivedi and Johnson 2013; Lee and Koo 2015). In the influencer domain, Breves et al. (2019), De Cicco, Iacobucci, and Pagliaro (2021), and Schouten, Janssen, and Verspaget (2020) recently demonstrated that influencers endorsing products that fit well with their image (e.g. a food influencer promoting a stand mixer) are also perceived as more trustworthy and as having more expertise than influencers endorsing ill-fitting products (e.g. a food influencer promoting a watch). Although Schouten, Janssen, and Verspaget (2020) did not find fit effects on identification, interviewee responses in Djafarova and Rushworth’s study (2017) suggest that people are more likely to identify themselves with influencers that are not just promoting anything to make money (which could be inferred from ill-fitting endorsements), but are honest and authentic individuals who carefully select their brand partnerships (which could be inferred from good-fitting endorsements).

In turn, credibility and identification are expected to result in more positive attitudes toward both the endorser and the endorsement. Credibility has long been an important factor in explaining endorsement effectiveness (e.g. Fink, Cunningham, and Kensicki 2004; Djafarova and Rushworth 2017; Lou and Yuan 2019). Moreover, individuals are generally more interested in the content posted by influencers that they identify with (Bond and Drogos 2014; Uzunoğlu and Kip 2014). In sum, we hypothesize that:

H2: Perceived credibility of the influencer and identification with the influencer mediate the relationship between product-influencer fit and advertising outcomes.

Whether credibility and identification also mediate an effect of product-influencer fit on perceived influencer likeability (as hypothesized in H1b), will be explored in this study.

RQ2: To what extent do perceived influencer credibility and identification mediate the effect of product-influencer fit on perceived influencer likeability?

Previous work suggests that the more followers influencers have, the more credible they are perceived and the more people identify with them (Djafarova and Rushworth 2017; Lin, Bruning, and Swarna 2018; Uzunoğlu and Kip 2014). Influencers with more followers have a much larger network size and reach, which makes them seem more trustworthy compared to influencers with fewer followers. Furthermore, influencers with a high number of followers are usually also seen as experts in their respective fields (Lin, Bruning, and Swarna 2018). Moreover, once influencers gather large numbers of followers on social media, they become known to their followers as someone they admire, associate with, and aspire to be (Djafarova and Rushworth 2017). In contrast, recent trends in influencer marketing seem to suggest that consumers are more likely to trust and relate to influencers that operate on a smaller scale, such as nano and micro influencers, who tend to have a more devoted follower base than macro and mega influencers (Brewster and Lyu 2020; Haverkamp 2018; Maheshwari 2018; Mooitheagency 2021; Reputatiefabriek 2019). To investigate these contrasting expectations, we have formulated the following research question:

RQ3: To what extent do perceived influencer credibility and identification mediate the effects of number of followers on advertising outcomes (attitude toward the ad and product, and purchase intention) and perceived influencer likeability?

Interaction between product-influencer fit and number of followers

Finally, the most important goal of our study is to explore the role of fit for different influencer types, by testing potential interaction effects between product-influencer fit and number of followers on both advertising outcomes and perceived influencer likeability. Since no earlier research has investigated their interplay, we can only speculate on the ways in which they may interact. Fit and number of followers may work in tandem: influencers with a high number of followers may be perceived as highly credible, and endorsing products in their domain of expertise may have more impact than when influencers with a low number of followers endorse them. Alternatively, fit could be more important for smaller influencers than for more established endorsers. Influencers with a smaller follower base, such as nano influencers, often have a close connection with their followers and advertising an ill-fitting product may be perceived as selling out (Campbell and Farrell 2020; Khamis, Ang, and Welling 2017). In comparison, followers of macro influencers are generally aware of the commercial interests of these endorsers and may expect them to advertise a larger diversity of (both good- and ill-fitting) products as a source of income (Campbell and Farrell 2020; Friestad and Wright 1994). On the other hand, if this commercial attitude is indeed becoming more and more off-putting to consumers (Reputatiefabriek 2019), number of followers may also weaken a positive fit effect.

Taking a different perspective, number of followers may only be an important determinant of endorsement effects when there is a good fit between the influencer and the endorsed product, and the effect could be weakened or even absent for ill-fitting endorsements. To explore potential interaction effects between product-influencer fit and number of followers on endorsement effectiveness, we pose the following research question:

RQ4: Are there any interaction effects between product-influencer fit and number of followers on advertising outcomes (attitude toward the ad and product, and purchase intention) and influencer likeability, and to what extent are they mediated by perceived credibility of the influencer and identification with the influencer?

Method

Participants

We tested our conceptual model (visualized in Figure 1) in an online experiment among a convenience sample of 432 Dutch Instagram users (226 female, 206 male) between the ages of 18 and 28 (M = 19.87, SD = 2.00). This age group is the most common demographic on Instagram, with over 64% active users (Pew Research Center 2018). Currently, Instagram has 5.9 million users of 15 years and older in the Netherlands (Newcom 2021), and influencer profiles are among the accounts most followed on Instagram (Starngage 2021).

Figure 1. Conceptual model.

Participants in our sample used Instagram on average 6 days a week (M = 6.06, SD = 1.67), with 68.1% indicating daily use. On a typical day, participants used Instagram for about 46 minutes (M = 45.45, SD = 39.59). On average, participants followed 425.81 (SD = 237.10) accounts and had 443.21 (SD = 395.52) followers. 68.5% of participants followed one or more influencers on Instagram, and of those participants 93.3% indicated to check the posts of these influencers multiple times per week or more. Of our participants, 25.2% were still in high school, 11.8% (27.3%) were currently enrolled in or finished lower (higher) vocational training, and 35.6% attended or finished university.

Design and procedure

The study used a 2 (Product-influencer fit: poor fit vs. good fit) × 2 (Number of followers: moderate vs. high) between-subjects design. Participants were randomly assigned to one of the four experimental conditions. Within each of the conditions, participants received a short introduction in which the concept ‘social media influencer’ was explained to them (cf. Kim and Kim 2021), after which they were presented with the Instagram bio and two posts of an influencer in the domain of health and fitness. This domain is one of the most popular social influencer niches according to academic studies (e.g. De Veirman, Cauberghe, and Hudders 2017; Schouten, Janssen, and Verspaget 2020; Sokolova and Perez 2021; Zou, Zhang, and Tang 2021), as well as popular sources (e.g. IZEA 2019; Socialbuzzhive 2020). Participants’ gender was matched to the gender of the Instagram influencer, such that female participants were presented with a female influencer and male participants with a male influencer (cf. De Veirman, Cauberghe, and Hudders 2017).

To ensure ecological validity, we used two existing influencer profiles as stimulus materials, which we adapted to fit our manipulations. Specifically, we selected a male health and fitness influencer from France and a female health and fitness influencer from Austria. To prevent existing attitudes toward these influencers affecting our results, we selected micro influencers, whose audience (as compared to macro influencers) tends to be more localized to their geographic base (Campbell and Farrell 2020), and we only included participants unfamiliar with these influencers in our sample.¹ We maintained the names and profile pictures of both influencers, but manipulated the number of posts and number of ‘followees’ to be constant for both of them. Underneath the Instagram bio of each influencer, two posts were shown. The first post contained a picture of the influencer, in which the comments on the original posts were blurred.² The second post contained a picture of a product ostensibly endorsed by the influencer, accompanied by the caption ‘The perfect weekend doesn’t exi- @[brand name]’ (see Figure 2).

Figure 2. Experimental condition showing the female influencer endorsement with a high number of followers and poor product-influencer fit. Blacked out areas were visible to participants in the experiment but are not visible here for legal & copyright reasons.

Product-influencer fit

In the good-fit conditions, the product endorsed by the influencers was a protein shake (cf. Schouten, Janssen, and Verspaget 2020), and in the poor-fit conditions, the influencers endorsed an ice cream brand. An additional sample of 66 Dutch Instagram users (described in footnote 2) rated six products (i.e. pizza, salad, ice cream, protein shake, chocolate chip cookies, and a protein bar) on perceived fit with a health and fitness influencer (1 = very poor fit; 5 = very good fit). A protein shake was perceived as the best match with influencers propagating a fit and healthy lifestyle (M = 4.82, SD = .52), and as a significantly better match than ice cream (M = 1.50, SD = .71; p < .001).

Number of followers

The bio of the influencer with a high number of followers reported a number of 578k (578,000) followers. We contrasted this high number of followers with a condition with a moderate number of followers, which was set at 5039, since the follower base of an influencer should be of at least moderate size to be considered an influencer (cf. De Veirman, Cauberghe, and Hudders 2017). With these numbers of followers, one is considered a macro influencer and a nano influencer, respectively (Campbell and Farrell 2020; Haverkamp 2018). To ensure that participants would indeed perceive 578k followers as a high number and 5039 as a moderate number of followers, we explained in the introduction of the experiment that influencers on Instagram have an average number of 50,000 (50k) followers. The number of ‘followees’ was kept constant and was set to 321 in all conditions (cf. De Veirman, Cauberghe, and Hudders 2017). See the Appendix for a complete overview of the stimulus materials in all experimental conditions.

Measures

Credibility

Perceived influencer credibility was measured with six 7-point semantic differential scales (α = .89) adopted from Ohanian (1990), assessing trustworthiness (‘I think this influencer is….’ untrustworthy – trustworthy, insincere – sincere, and unreliable – reliable) and expertise (not an expert – an expert, unknowledgeable – knowledgeable, and unqualified – qualified).

Identification

Participants subsequently indicated to what extent they could identify themselves with the influencer on eight Likert-scale items (1 = totally disagree; 7 = totally agree; α = .89) adapted from Hoffner and Buchanan (2005), measuring wishful identification (e.g. ‘Sometimes I wish I could be more like this influencer’) and perceived similarity (e.g. ‘This influencer is similar to me’).

Influencer likeability

We measured influencer likeability with four items on a 7-point Likert scale (1 = totally disagree; 7 = totally agree) adapted from the social attraction dimension of the interpersonal attraction scale by McCroskey and McCain (1974). The items were ‘I like this influencer’, ‘I think the influencer could be a friend of mine’, ‘I would like to meet this influencer’, and ‘The influencer makes me feel comfortable, as if being with friends’ (α = .82).

Attitude toward the ad, product, and purchase intention

Next, participants evaluated the sponsored Instagram post and the product endorsed by the influencer. Participants indicated their attitude toward both the ad and the product on five 7-point semantic differential scales developed by Spears and Singh (2004): ‘I think the advertisement/product is…’ unappealing – appealing, bad – good, unpleasant – pleasant, unfavorable – favorable, and unlikeable – likeable (α = .88; α = .90, respectively). Purchase intent was measured with one item: ‘The next time you are looking to purchase this type of product, how likely are you to buy this product?’ (1 = very unlikely; 7 = very likely).

Manipulation checks

As manipulation check of product-influencer fit, we asked respondents whether the influencer fits the endorsed advertisement on a 7-point Likert scale (1 = totally disagree; 7 = totally agree). Similarly, we asked respondents to indicate (1) whether they thought the influencer was popular and (2) whether the influencer had many followers (r = .80, p < .001).

Results

Manipulation checks

The manipulation of product-influencer fit was successful, with perceptions of product-influencer fit being higher in the good fit condition (M = 5.67, SD = 1.43) than the poor fit condition (M = 2.37, SD = 1.54), t(430) = 23.07, p < .001, d = 2.22. Participants also considered the influencer with a high number of followers as more popular (M = 5.88, SD = 1.03) than the influencer with a moderate number of followers (M = 4.14, SD = 1.62), t(430) = 13.39, p < .001, d = 1.27.

Main analyses

To test our hypotheses and answer our RQs, we first conducted ANOVAs on our mediating and dependent variables with product-influencer fit (poor fit vs. good fit) and number of followers (moderate vs. high) as between-subject factors. This also allowed us to test whether there are any interaction effects between product-influencer fit and number of followers on any of the mediating or dependent variables. Means and standard deviations are presented in Table 1.

Table 1. Means and standard deviations (in parentheses) for all mediating and dependent variables for the good-fit and poor-fit conditions, as a function of number of followers.

	Poor fit		Good fit
	Moderate number of followers	High number of followers	Moderate number of followers	High number of followers
	(n = 116)	(n = 105)	(n = 106)	(n = 105)
Credibility	3.69^a (1.07)	3.81^a (1.01)	4.23^b (1.03)	4.57^c (1.06)
Identification	2.67^a (1.18)	2.65^a (1.11)	2.75^ab (1.20)	3.07^b (1.13)
Influencer likeability	2.29^a (1.04)	2.42^ab (1.11)	2.32^ab (1.12)	2.69^b (1.22)
Attitude toward the advertisement	4.35^a (1.33)	4.48^a (1.08)	4.57^a (1.12)	4.99^b (1.06)
Attitude toward the product	5.11^a (1.26)	5.00^a (1.10)	4.55^b (1.15)	4.80^ab (1.22)
Purchase intention	3.57^a (1.62)	3.30^ab (1.58)	2.91^b (1.52)	3.06^ab (1.55)
Manipulation check fit	2.29^a (1.49)	2.46^a (1.59)	5.52^b (1.49)	5.82^b (1.34)
Manipulation check number of followers	4.14^a (1.64)	5.79^b (1.09)	4.15^a (1.60)	5.98^b (0.97)

Note. Different superscripts indicate significant differences (p < .05) between conditions (across rows, Tukey’s B correction applied).

To test H2, and answer RQs 2 and 3, we conducted moderated mediation analyses using PROCESS 3.1 (Hayes 2013). To test H2 and answer RQ2, we conducted four separate analyses, one for each of the four dependent variables (attitude toward the ad, attitude toward the product, purchase intention, and influencer likeability) with product-influencer fit as independent variable, and credibility and identification as mediators. We included number of followers as a moderator in the relationship between fit and the mediating variables and between fit and the dependent variables (model 8, 10.000 bootstrap samples). The analyses for answering RQ3 were similar, except number of followers served as the independent variable, and product-influencer fit was entered as the moderator.

Direct effects of product-influencer fit and number of followers on advertising outcomes and influencer likeability

Attitude toward the ad was more positive in the good fit than in the poor fit condition, F(1, 428) = 10.60, p = .001, η² = .024, ω² = .021. However, attitude toward the product, F(1, 428) = 11.02, p = .001, η² = .025, ω² = .023, and purchase intention, F(1, 428) = 9.07, p = .003, η² = .021, ω² = .018, were less positive in the good fit than in the poor fit condition, indicating that people prefer ice cream over a protein shake. Attitude toward the ad was more positive when the endorser had a high number of followers than when the endorser had a moderate number of followers, F(1, 428) = 6.01, p = .015, η² = .014, ω² = .011. Product attitude, F(1, 428) = 0.38, p = .540, and purchase intention, F(1, 428) = 0.14, p = .710, were not affected by the number of followers of the endorser. In addition, there were no interaction effects on attitude toward the ad, F(1, 428) = 1.61, p = .205, attitude toward the product, F(1, 428) = 2.65, p = .104, and purchase intention, F(1, 428) = 1.89, p = .170.

Product-influencer fit did not affect likeability of the endorser, F(1, 428) = 1.88, p = .172. However, participants did like the endorser with a high number of followers better than the endorser with a moderate number of followers, F(1, 428) = 5.49, p = .020, η² = .013, ω² = .010. No interaction effect was found between product-influencer fit and number of followers on influencer likeability, F(1, 428) = 1.12, p = .283.

In sum, H1a was only confirmed for the positive effect of product-influencer fit on attitude toward the ad. In response to H1b, there was no direct effect of fit on influencer likeability, and in response to RQ1, the number of followers positively affected both attitude toward the ad and influencer likeability.

Direct effects of product-influencer fit and number of followers on credibility and identification

Product-influencer fit significantly affected credibility, F(1, 428) = 42.47, p < .001, η² = .090, ω² = .087, with credibility of the influencer deemed higher in the good fit than the poor fit condition. Influencers with more followers were also seen as more credible than influencers with fewer followers, F(1, 428) = 5.16, p = .024, η² = .012, ω² = .009. There was no interaction effect between product-influencer fit and number of followers on credibility, F(1, 428) = 1.20, p = .274.

Identification was higher in the good fit than in the poor fit condition, F(1, 428) = 5.10, p = .024, η² = .012, ω² = .009. However, number of followers had no effect on identification, F(1, 428) = 1.85, p = .175, and no interaction effect was observed between product-influencer fit and number of followers, F(1, 428) = 2.37, p = .125.

Moderated-mediation analysis: the effect of product-influencer fit on advertising outcomes and influencer likeability via credibility and identification, moderated by number of followers

Attitude toward the ad

Credibility, b = .435, t(426) = 8.67, p < .001, and identification, b = .236, t(426) = 5.23, p < .001, both significantly predicted attitude toward the ad. Product-influencer fit did not directly affect attitude toward the ad when controlling for the mediators, b = −.034, t(426) = 0.25, p = .800. Credibility mediated the relationship between product-influencer fit and attitude toward the ad, both for influencers with a moderate, b = .237, 95% BCI [.112, .377], and for influencers with a high number of followers, b = .332, 95% BCI [.191, .496]. Identification mediated the relationship between product-influencer fit and attitude toward the ad for influencers with a high number of followers only, b = .100, 95% BCI [.026, .192], and not for influencers with a moderate number of followers, b = .019, 95% BCI [−.054, .103].

Attitude toward the product

Product-influencer fit had a direct negative effect on attitude toward the product, after controlling for the mediators, b = −.723, t(426) = −4.80, p < .001. Credibility, b = .255, t(426) = 4.53, p < .001, and identification, b = .234, t(426) = 4.61, p < .001, both positively predicted product attitude. For both influencers with a moderate number of followers, b = .139, 95% BCI [.055, .242], and a high number of followers, b = .195, 95% BCI [.085, .327], credibility positively mediated the relationship between product-influencer fit and product attitude. Moreover, identification mediated the relationship between product-influencer fit and product attitude for influencers with a high number of followers, b = .099, 95% BCI [.025, .194], but not for influencers with a moderate number of followers, b = .019, 95% BCI [−.055, .098].

Purchase intention

Product-influencer fit had a direct negative effect on purchase intention, after controlling for the mediators, b = −.802, t(426) = −3.92, p < .001. Credibility, b = .208, t(426) = 2.72, p = .008, and identification, b = .314, t(426) = 4.57, p < .001, both positively affected purchase intention. Again, credibility was a significant mediator in both the moderate, b = .113, 95% BCI [.019, .238], and high number of followers conditions, b = .159, 95% BCI [.029, .328], and identification was only a significant mediator in the high number of followers condition, b = .139, 95% BCI [.032, .261], and not in the moderate one, b = .025, 95% BCI [−.071, .140].

Influencer likeability

Finally, both credibility, b = .223, t(426) = 5.26, p < .001, and identification, b = .546, t(426) = 14.31, p < .001, significantly predicted influencer likeability. Although product-influencer fit did not directly affect influencer likeability, b = −.133, t(426) = −1.18, p = .240, credibility mediated this relationship both for influencers with a moderate, b = .121, 95% BCI [.054, .200], and high number of followers, b = .160, 95% BCI [.087, .267]. Identification, on the other hand, only mediated the relationship between product-influencer fit and likeability for influencers with a high number of followers, b = .231, 95% BCI [.065, .406], and not for influencers with a moderate number of followers, b = .044, 95% BCI [−.121, .228].

To conclude, H2 was confirmed for both mediators on all dependent measures, since both credibility and identification mediate the effects of product-influencer fit on advertising outcomes and perceived influencer likeability. However, as questioned by RQ4, for identification, this effect was found for influencers with a high number of followers only. Figure 3 presents an overview of all moderated indirect effects.

Figure 3. Moderated indirect effects of product-influencer fit via credibility and identification on all dependent variables. Figure depicts significant direct effects at p < .05. Coefficients are unstandardized b’s.

Moderated-mediation analysis: the effect of number of followers on advertising outcomes and influencer likeability via credibility and identification, moderated by product-influencer fit ³

Attitude toward the ad

Number of followers did not directly affect attitude toward the ad when controlling for the mediators, b = .132, t(426) = 0.64, p = .520. In the good-fit condition, credibility mediated the relationship between number of followers and attitude toward the ad, b = .147, 95% BCI [.022, .289], but not in de poor-fit condition b = .051, 95% BCI [−.067, .176]. This was similar for identification, which mediated the relationship between number of followers and attitude toward the ad in the good-fit condition, b = .076, 95% BCI [.002, .161], but not in the poor-fit condition b = −.005, 95% BCI [−.079, .068].

Attitude toward the product

Number of followers did not directly affect attitude toward the product after controlling for credibility and identification, b = −.142, t(426) = −0.96, p = .340. Again, credibility mediated the relationship between number of followers and attitude toward the product when fit was good, b = .086, 95% BCI [.012, .184], but not when fit was poor, b = .030, 95% BCI [−.042, .102]. This was also the case for identification, which was a significant mediator in the good-fit condition, b = .075, 95% BCI [.000, .160], but not in the poor-fit condition, b = −.005, 95% BCI [−.080, .067].

Purchase intention

Number of followers did not affect purchase intention after controlling for the mediators, b = −.283, t(426) = −1.40, p = .161. When there was good product-influencer fit, credibility mediated the relationship between number of followers and purchase intention, b = .070, 95% BCI [.004, .171], but not when there was poor product-influencer fit, b = .025, 95% BCI [−.038, .090]. This was also the case for identification, which was a mediator when product-influencer fit was good, b = .101, 95% BCI [.002, .216], but not when it was poor, b = −.006, 95% BCI [−.105, .096].

Influencer likeability

Finally, number of followers did not directly affect influencer likeability, after controlling for credibility and identification, b = .122, t(426) = 1.09, p = .277. Credibility significantly mediated the relationship between number of followers and likeability but only when there was good product-influencer fit, b = .075, 95% BCI [.013, .156], and not when there was poor product-influencer fit, b = .026, 95% BCI [−.035, .091]. This was similar for identification, which only mediated the relationship between number of followers and influencer likeability when product-influencer fit was good, b = .176, 95% BCI [.007, .153], and not when it was poor, b = −.011, 95% BCI [−.172, .162].

In response to RQ3 we can conclude that both credibility and identification mediate the effects of number of followers on advertising outcomes and perceived influencer likeability. However, as questioned by RQ4, in all analyses this effect was found for good-fitting endorsements only. Figure 4 presents an overview of all moderated indirect effects.

Figure 4. Moderated indirect effects of number of followers via credibility and identification on all dependent variables. Figure depicts significant direct effects at p < .05. Coefficients are unstandardized b’s.

Discussion

The present study experimentally tested the effects of influencer endorsements on Instagram. We systematically investigated to what extent product-influencer fit and number of followers interact in affecting advertising outcomes and influencer likeability, and examined to what extent perceived credibility of the influencer and identification with the influencer mediate these effects. Our results contribute to theorizing on the factors driving influencer advertising success, contribute to a deeper understanding of the processes underlying these effects, and provide guidelines for practitioners.

The first hypothesis posed that influencers endorsing a product that fits well with their image ensure more positive advertising outcomes as compared to influencers endorsing a product that poorly fits their image (H1a). Additionally, we investigated the effect of product-influencer fit on perceived influencer likeability (H1b) and the mediating roles of perceived credibility and identification with the influencer (H2 and RQ2). First, as expected in H1a, people were more positive about the ad when the endorsement fit the influencer than when the endorsement did not. In contrast, attitude toward the product and purchase intention were lower in the poor-fit conditions than in the good-fit conditions, which is a methodological artifact resulting from the products endorsed: people generally like ice cream (the ill-fitting product) more than a protein shake (the well-fitting product). Even though these effects were negative, credibility and identification positively mediated the relationship between fit and all three measures of advertising effectiveness (confirming H2), although identification was only a mediator when the influencer had a high number of followers. In sum, even though people like protein shakes less than ice cream, endorsing a protein shake made people perceive health and fitness influencers as more credible and aspirable than when the influencers were endorsing ice cream, which resulted in greater appreciation for the ad and the product, and a higher intention to buy the product endorsed. Confirming earlier work (Breves et al. 2019; Kim and Kim 2021; Schouten, Janssen, and Verspaget 2020; Torres, Augusto, and Matos 2019), advertising products that fit the endorser’s image is of utmost importance for the endorsed brand, especially when brands hire highly popular macro influencers. As previous studies implied (Djafarova and Rushworth 2017; Till and Busler 2000), the success of these endorsements is fueled by both trust and similarity.

Second, product-influencer fit also appeared to be of significant importance for self-branding as an influencer (H1b). Fit did not affect influencer likeability directly, but this relationship was mediated by both credibility and identification (RQ2), with the latter only when the influencer had a high number of followers. In line with previous findings (Breves et al. 2019; De Cicco, Iacobucci, and Pagliaro 2021; Djafarova and Rushworth 2017; Schouten, Janssen, and Verspaget 2020), influencers endorsing a product that does not fit their image are seen as less credible, and people identify with them less, which results in less positive endorser evaluations. In line with De Cicco, Iacobucci, and Pagliaro (2021), who showed that influencer-product congruence positively affected people’s attitude towards the influencer, the present study demonstrates that advertising products that fit your image is of significant importance to be respected as an influencer, especially when you have already established an impressive (macro-level) follower base. Both trust and similarity are the foundation of a good influencer-follower relationship (Djafarova and Rushworth 2017; Uzunoğlu and Kip 2014).

Our first and third research question investigated the effect of number of followers on advertising outcomes and perceived influencer likeability, and to what extent perceived influencer credibility and identification mediate these effects. Number of followers positively affected attitude toward the ad (RQ1), and positive effects on attitude toward the ad, attitude toward the product and purchase intention were mediated by credibility and identification (RQ3), albeit only when the endorsement fit the influencer. People thus seem to value endorsements of a macro influencer more than those of a nano influencer, because they perceive them as more credible and identify with them more. These results do not align with recent trends that consumers are more likely to trust product endorsements of influencers that operate on a smaller scale (Reputatiefabriek 2019), but once more seem to demonstrate that quantity on social media has value (Van Dijck 2013). Previous work suggested that influencers with a large number of followers have a greater social influence (e.g. Djafarova and Rushworth 2017; Kay, Mulcahy, and Parkinson 2020), and speculated that this may be linked to how credible they are perceived by their audience. The findings of our study confirm these assumptions.

For perceived influencer likeability, similar effects were found. Similar to De Veirman, Cauberghe, and Hudders (2017), we found that participants liked (macro-)influencers with a high number of followers more than (nano)influencers who have a moderate number of followers (RQ1). In addition, we demonstrated that this can be explained by increased credibility perceptions and identification (RQ3), albeit only when the influencers endorsed a product that fit their image. Consumers indeed seem to use number of followers as a heuristic to judge whether they like the influencer (e.g. Jin and Phua 2014), but number of followers alone is not enough: only if influencers with a high number of followers endorse a product that fits their image they are perceived as more credible and aspirable than influencers with fewer followers, resulting in higher likeability. In sum, our study has not only shown the importance of both credibility and identification as underlying mechanisms in the effect of number of followers on endorsement effectiveness, but has also shown that influencers should stick to their image in their endorsements, as fit seems to be a prerequisite for these positive effects to be observed.

Finally, our exploration of interaction effects between product-influencer fit and number of followers (RQ4) revealed no direct interaction effects on any of the dependent variables, but as discussed above, several mediated interaction effects were found. All interactions converge in the conclusion that fit and number of followers seem to work in tandem: when both conditions are met, influencers are seen as credible experts, with whom one can identify and look up to, which positively affects evaluations of both the endorser and their endorsements.

Theoretical and practical implications

Together, our findings contribute to a growing body of knowledge on the processes driving audience responses to influencer marketing. The present work demonstrates the robustness of product-influencer fit as an important predictor of successful influencer endorsements. In the context of Instagram, endorsements in which influencers adhere to their self-branded image result in more positive outcomes for both the promoted brand and the influencer. Product-endorser fit has long been seen as an important factor in endorser advertising (e.g. Kamins 1990), and is considered even more important for influencers than more traditional celebrities (Schouten, Janssen, and Verspaget 2020). The present research confirms the importance of fit for influencer advertising, and adds that endorsing fitting products are even more important for macro influencers, who have an established influencer career, than for content creators who are just beginning to grow their follower bases.

Whereas our study confirms the importance of congruence between influencer and product characteristics (Erdogan 1999; Kamins and Gupta 1994), congruity theory and research has shown that mild incongruence can sometimes result in more favorable brand evaluations (Groza, Cobbs, and Schaefers 2012; Mandler 1982). This is often linked to a novelty effect, where the unexpected has positive effects (Yoon 2013), however only when the incongruities can successfully be resolved (Groza, Cobbs, and Schaefers 2012). The incongruities in the poor-fit conditions in our study were not easily reconciled by the participants, which confirms the importance of fit in influencer advertising.

Our study is one of the first to experimentally investigate whether consumers respond differently to endorsements of different types of influencers, as well as the processes underlying these effects. A large number of followers seems not only to provide an advantage when it comes to likeability perceptions (De Veirman, Cauberghe, and Hudders 2017), but a large follower base also ensures more positive ad outcomes. However, findings imply that a large follower count alone does not guarantee positive endorsement outcomes. Good product-influencer fit is a prerequisite for a large number of followers to result in more positive ad outcomes and increased likeability perceptions. Important to note is that what is considered a large amount of followers may depend on the medium on which the product is endorsed. In contrast to our findings, Westerman and colleagues (2012) showed that a moderate number of followers on Twitter (7000) was more positively perceived than a high amount (70,000) of followers.

Furthermore, the present research aimed to shed light on the roles of perceived credibility (cf. Ohanian 1990) and identification with the endorser (cf. Basil 1996) in explaining the effects of number of followers and fit. Whereas Schouten, Janssen, and Verspaget (2020) identified these processes as important drivers of fit effects on endorsement effectiveness, the present research shows that these effects also drive effects of number of followers, and are especially relevant for macro influencers who have established a large follower base. Moreover, the present work is the first to demonstrate that credibility perceptions and identification do not only affect advertising outcomes, but also play an important role in our evaluations of the influencer-as-a-brand. Influencers are appreciated more when their endorsements fit their image than when they do not, because they are perceived as trusted experts. Especially when influencers have built a large follower base, people feel more connected to those who endorse fitting products and stick to who they are instead of endorsing products seemingly only for commercial reasons.

Finally, these insights are not only important from a scientific perspective, but also provide guidance for advertisers and influencers when deciding which partnerships to pursue and which to avoid. Although product endorsements by nano and micro influencers appear to be on the rise (e.g. Maheshwari 2018; Reputatiefabriek 2019), our findings imply that macro influencers are more influential endorsers, because consumers relate to them more and are more likely to trust their content. Macro influencers thus remain attractive partners for a marketing campaign, and ensuring a large follower base seems aspirable for influencers wanting to become successful and admired product endorsers. However, ensuring a good match between influencer and product is an important prerequisite for both commercial partners, and this alignment is even more important when dealing with macro influencers.

For influencers, our research stresses the value of staying true to your image as a key element of a successful influencer career. Instagram influencers will be perceived as more attractive peer role models, as well as more skilled product endorsers, when they choose their endorsements wisely and do not simply endorse for commercial reasons only. When they collaborate with brands, they have to think carefully about which brands to endorse, since these endorsements should ideally not interfere with their carefully built-up image (cf. Audrezet, Kerviler, and Moulard 2020).

Limitations and suggestions for future research

A downside of running a controlled experiment on audience evaluations of Instagram influencers is the fact that one cannot invite actual followers of existing influencer accounts to participate. In contrast to our participants, who were confronted with a single Instagram post of a theretofore unknown influencer, ‘real’ followers have been able to experience parasocial interactions with these influencers over an extended period of time, and may have built parasocial relationships, perceiving the influencer as a close friend (Dibble, Hartmann, and Rosaen 2016). Processes such as perceived credibility and identification may therefore play a more significant role in real life than in our controlled environment. Moreover, whereas our study focused on credibility and identification, future research may profitably explore to what extent other processes, such as authenticity and realism perceptions, are responsible for the demonstrated effects. Furthermore, it remains to be investigated whether our findings can be generalized to behavioral outcomes, such as followers spreading word-of-mouth and actually buying the products that influencers endorse, although studies (and sales figures) imply this to be the case (e.g. Chapple and Cownie 2017; Djafarova and Rushworth 2017).

To be able to study the effects of fit and number of followers, other factors that could affect audience evaluations of influencers and their endorsements, such as number of followees and type of product, were kept constant. Future research may further investigate the boundary conditions for both number of followers and fit effects to occur. For example, De Veirman, Cauberghe, and Hudders (2017) demonstrated that popularity of Instagram influencers may turn against them if they only follow a few accounts themselves, and influencers with a high number of followers ensure less positive brand evaluations when a divergent product is being endorsed. Moreover, when people actually follow an influencer, other factors than number of followers may also determine the respect people have for these influencers, such as familiarity with the influencer and previous experience with endorsements of these influencers. Furthermore, influencer accounts with a moderate and high number of followers may differ in aspects other than just follower count. For example, influencers with a high number of followers may have more diverse posts on their Instagram profile or may advertise a wider array of products outside of their niche. As a result, posts from these influencers may have a poorer fit compared to influencers with a moderate number of followers, who are more likely to advertise products that fit within their nice. As said, future research may further investigate the boundary conditions of the effects of number of followers and fit, and explore their interplay with factors such as post diversity or number of endorsements.

Another direction for future research could be investigating influencer endorsement effects on other social media platforms. The current study focused on photo-based platform Instagram, which is still a relevant platform with a billion users worldwide and almost 6 million users in the Netherlands alone (Newcom 2021; Statista 2019). However, video-sharing platforms like YouTube and, more recently, TikTok are growing increasingly fast. We suggest that future studies consider these video-based platforms when investigating the effects of influencer endorsements, as well as compare endorsement effectiveness on photo and video-based platforms.

To ensure a basic level of perceived similarity with the influencer, in this study participants’ gender was matched to the gender of the Instagram influencer, such that female participants were presented with a female influencer and male participants with a male influencer (cf. De Veirman, Cauberghe, and Hudders 2017). Our participants could thus only choose between male or female, which does not represent an inclusive gender portrayal, and they may not have identified with either (Kuyper and Wijsen 2014). Although we do not expect this to have significantly affected our results (as described in footnote 2, in an additional sample only 1 out of 66 participants identified as ‘other’), we suggest future studies to include participants with a wider array of gender identities to allow for broader generalizations.

Finally, future research could further investigate the role of different influencer types. Although our choice for comparing macro and nano influencers was motivated by the literature (De Veirman, Cauberghe, and Hudders 2017), the demonstrated effects are bound to this specific comparison. Whether the positive effects of number of followers and fit also apply for influencers with other follower counts (i.e. micro influencers or mega influencers; Campbell and Farrell 2020) remains to be investigated. Influencers who are extremely successful and dominant in their business may have become such icons that they can get away with an endorsement gone wrong. However, the effectiveness of number of followers could also follow an inverted-U curve. Since more followers would make the influencer more of a celebrity than an influencer, this may result in followers finding the influencer less credible and authentic than less popular influencers (cf. Schouten, Janssen, and Verspaget 2020; Westerman, Spence, and Van Der Heide 2012). It would be interesting to investigate whether the influencer popularity-endorsement effectiveness relationship is linear or polynomial.

In conclusion

In their role of credible and relatable ‘superpeers’, social media influencers are uniquely positioned to persuade numbers of followers to admire their personas and the products they endorse. The present work contributes to our understanding of how to harness their power as an advertising instrument.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the corresponding author, L. Janssen, upon reasonable request.

Additional information

Notes on contributors

Loes Janssen

Loes Janssen is an assistant professor of Business Communication and Digital Media at Tilburg University. She applies her expertise in social psychology and communication science to understand consumer responses to persuasive communication. Specifically, she investigates the dynamics behind the (in)effectiveness of persuasive messages in the domains of marketing communication and health communication. Influencer (health) communication is one of her domains of expertise.

Alexander P. Schouten

Alexander P. Schouten is an assistant professor of Business Communication and Digital Media at Tilburg University. His research interests include social media use, online collaboration, and online impression management. Within these areas, he is specifically interested in how different media capabilities affect the way in which people and organizations can effectively use new media technologies to communicate, to market, to work together, and to present themselves.

Emmelyn A. J. Croes

Emmelyn A. J. Croes is an assistant professor of Communication and Technology at Tilburg University. Her research focuses on the impact of contemporary communication technologies on friendship formation and maintenance, ranging from computer-mediated communication technologies to human-machine communication. Her recent work studies the possibility of friendship formation between humans and chatbots.

Notes

1 Our original sample consisted of 435 participants, of which three participants were excluded because they indicated to be familiar with the male or female influencer.

2 An additional sample of 66 Dutch Instagram users (49 female, 16 male, 1 ‘other’, between the ages of 18 and 27, M = 19.23, SD = 1.69) rated the female and male influencer as similar in physical attractiveness, social attractiveness, and trustworthiness (F(1,63) = .69, p = .410, F(1,63) = .93, p = .340, and F(1,63) = .001, p = .981, respectively). All constructs were measured on 5-point semantic differential scales (Physical attractiveness: not sexy-sexy, ugly-beautiful, unattractive-attractive; Social attractiveness: unfriendly-friendly, unlikeable-likeable, unsympathetic-sympathetic; Trustworthiness: dishonest-honest, insincere-sincere, untrustworthy-trustworthy).

3 To avoid overlap in reporting, effects of the mediators (credibility and identification) on the dependent variables are not discussed again here.

Appendix.

Stimuli used in the experiment. Blacked out areas were visible to participants in the experiment but are not visible here for legal & copyright reasons

Table