1. User Engagement in Influencer Branding as Research Objective

Social media—as a part of digital media—can be defined as a “group of internet-based applications that build on the ideological and technological foundations of web 2.0, and that allow the creation and exchange of user generated content”. Kaplan/Haenlein (2010), p. 61.

According to the latest long-term study of ARD/ZDF (2015, p. 18) regarding the topic “mass communication”, half of the respondents (n = 4,131; 14 + years) state to rely on the Internet while searching for information. Solely TV surpasses this number (63 percent). For the most accessible target group of social media communication—the so-called digital natives—the numbers are even clearer: 79 percent of the respondents (n = 837; aged between 14 and 29) state to use the Internet as a source of information; television is only named by 50 percent. Cf. Breuning/Engel (2015), p. 331.

Traditional advertising can be classified as above-the-line communication. Advertisements published via newspapers, magazines, television, radio, and cinema or out-of-home media belong to this category. Cf. Meffert/Burmann/Kirchgeorg (2015), p. 586.

Cf. Dinner/van Heerde/Neslin (2014), pp. 528 et seq.; Medienpädagogischer Forschungsverbund Südwest (2017), p. 13.

“Online communication refers to all communication activities between companies and consumers, as well as among consumer themselves, that affect the achievement of marketing and corporate objectives and are handled through an internet protocol (IP).” Burmann et al. (2017), p. 196 referring to Meffert/Burmann/Kirchgeorg (2015), p. 633.

Cf. Dentsu Aegis Network (2019a).

Cf. Dentsu Aegis Network (2019b).

Social networks, e.g. Facebook or Instagram, “include platforms that focus on relationships among people with shared interests”. Kleine-Kalmer (2016), p. 2.

Cf. Burmann/Kleine-Kalmer/Hemmann (2014), pp. 62 et seq.

Cf. Faktenkontor (2019), p. 54. The respondents were asked to state how much trust they place in information rerceived from commercial suppliers (e.g. companys) regarding the respective media sources. A further study conducted by PwC with 1,000 respondents (18 + years) confirms these results. Compared to traditional media channels (e.g. TV = 74 percent, radio = 72 percent), social media networks have the lowest level of trust in information (e.g. Twitter = 14 percent, Facebook = 18 percent). Cf. PwC (2018), p. 12.

Note: the sample size varied with the investigated communication channels; n = 2,933 (Radio, TV, Print, Online Print), 973 (Blogs); n = 2,581 (YouTube); n = 871 (Twitter); n = 1,291 (Instagram); n = 639 (Snapchat); n = 2,417 (Facebook)

Cf. Nielsen (2015). The latest survey of Nielsen’s “Global Trust in Advertising” was conducted in February and March 2015. More than 30,000 consumers in 60 countries throughout Asia-Pacific, Europe, Latin America, the Middle East, Africa and North America were polled. The sample claims to be representative for Internet consumers in the respective countries. A more recent survey conducted by trnd in April 2017 with 12,500 users of the company’s own research community confirms these results and states that even 87.4 percent of the polled users rely on recommendations of friends before a purchase. Cf. trnd (2017).

Cf. Agrawal (2016); Khamis/Ang/Welling (2017); SproutSocial (2018a).

The values indicate the search interest relative to the highest point in the graph over the period considered. The value 100 stands for the highest popularity of this search term. The value 50 means that the term was half as popular and the value 0 equals a popularity of less than 1% compared to the maximum. Cf. Google Trends (2020).

Corresponding to the changing consumer behavior during the christmas period, i.e. a shifted focus on christmas-specifc topics, search interest for the the term „influencer“ is likewise decreasing in the respective periods.

Cf. Chan/Misra (1990), pp. 54 et seq.; Flynn/Goldsmith/Eastman (1996), p. 138; Katz (1957), p. 77.

Rogers/Cartano (1962), p. 435.

Cf. Berger/Schwartz (2011), p. 870; Chan/Misra (1990), p. 53; Rogers (1983), p. 271.

Cf. Boyd/Golder/Lotan (2010), p. 1; Godes/Mayzlin (2004), p. 545; Kane et al. (2009), p. 45; Martin/Lueg (2013), p. 801.

Henning-Thurau et al. (2004), p. 39.

Kaplan/Haenlein (2010), p. 61.

Contrarily, brand-generated content (BGC) encompasses online content created and published by brands in order to reach commercial goals. Cf. Arnhold (2010), p. 31; Eilers (2014), p. 47.

Arnhold (2010), p. 331. Moreover, brand-related UGC can further be subdivided into sponsored and non-sponsored brand-related UGC. Cf. Burmann et al. (2018), p. 253.

Cf. Flynn/Goldsmith/Eastman (1996), p. 137; Sinnig (2019), p. 30.

Cf. e.g. Burmann et al. (2018), p. 255; de Veirman/Cauberghe/Hudders (2017), pp. 798 et seq.; Erkan/Evans (2016), p. 52; Killian (2017), p. 64.

Cf. Audrezet/de Kerviler/Moulard (2018), p. 2; Nee (2016), p. 4; Sinnig (2019), p. 30.

Cf. e.g. Cotter (2019); de Veirman/Cauberghe/Hudders (2017); de Veirman/Hudders (2019); Djafarova/Trofimenko (2019); Evans et al. (2017); Freberg et al. (2011); Jaakonmäki/Müller/vom Brocke (2017); Jin/Muqaddam (2019); Lee/Watkins (2016); Lu/Chang/Chang (2014); Sinnig (2019); Uzunoḡlu/Kip (2014).

Cf. Uzunoḡlu/Kip (2014), pp. 592, 595.

Lu/Chang/Chang (2014), p. 259.

The authors define sponsored recommendation posts as “a blog article written by a blogger which is sponsored by the producers of any product or by a marketing agency that reviews and promotes products or services on their blog”. Lu/Chang/Chang (2014), p. 259.

Cf. Lu/Chang/Chang (2014), p. 259; Zhu/Tan (2007), p. 2.

Lee/Watkins (2016), p. 5753.

Cf. Lee/Watkins (2016), p. 5753.

Cf. e.g. de Veirman/Cauberghe/Hudders (2017), p. 798; de Veirman/Hudders (2019), p. 2; Evans et al. (2017), p. 139; Jaakonmäki/Müller/vom Brocke (2017), p. 1153; Jin/Muqaddam (2019), p. 522.

Jaakonmäki/Müller/vom Brocke (2017), p. 1153.

Cf. Jaakonmäki/Müller/vom Brocke (2017), p. 1153.

Cf. de Veirman/Hudders (2019), p. 2.

Several authors, who recently published in the field of influencer branding, correspondingly apply this approach, e.g. Daniel/Crawford Jackson/Westerman (2018); Sinnig (2019); Xiao/Wang/Chan-Olmsted (2018).

Cf. Burmann et al. (2018), p. 253. In the following the words social media influencer and influencer are used interchangeably throughout this dissertation.

Cf. BVDW (2019), p. 9. The survey was conducted in March 2019 with n = 1,051 German respon-dents aged between 16 and 64 years. Note: weekly contact with an SMI implies both active as well as coincidental contacts, i.e. seeing an SMI or his/her content in social media.

Cf. Sinnig (2019), p. 2; Webster/Ksiazek (2012), p. 40.

Cf. Forbes (2019); Sinnig (2019), pp. 3 et seqq.

Similar results are shown in a recent study by mScience (2019, p. 25 et seq).

Cf. e.g. Boerman (2020); de Veirman/Cauberghe/Hudders (2017); Domingues Aguiar/van Reijmersdal (2018); Klein/Becker (2018).

Cf. Alassani/Göretz (2019), p. 252 et seq.; Influicity (2017), p. 4; van der Nolk van Gogh (2017).

Cf. Boerman (2020), p. 202; Domingues Aguiar/van Reijmersdal (2018); Lammers (2018), p. 110.

Cf. e.g. Khamis/Ang/Welling (2017); Loroz/Braig (2015); Engh (2006), Thomson (2006).

Cf. Djafarova/Trofimenko (2019); Khamis/Ang/Welling (2017).

Cf. Chen (2013); Labreque/Markos/Milne (2011).

Cf. Audrezet/de Kerviler/Moulard (2018), p. 2; de Veirman/Cauberghe/Hudders (2017), p. 798; Krömer/Borchers/Enke (2018), p. 119 et seq.; Sinnig (2019), p. 176.

Cf. Burmann et al. (2018), p. 253; Sinnig (2019), p. 40.

Cf. de Veirman/Cauberghe/Hudders (2017), p. 801; Dhanesh/Duthler (2019), p. 3; Evans et al. (2017), p. 138; Jaakonmäki/Müller/van Brocke (2017), p. 1153; Lou/Yuan (2019), p. 58; Sinnig (2019), p. 40.

Note: Since the focus of this work lies within B2C markets the term “consumer” will be used throughout this dissertation. Nevertheless, all explanations on the basic model of influencer branding are also generally applicable in the context of B2B markets.

Cf. Boerman (2020); Sinnig (2019), p. 46.

Cf. Schade (2012), p. 77; Sinnig (2019), pp. 71 et seqq.; Sirgy (1982), p. 288 et seq.; Sirgy (1985), p. 195. Accordingly, a positive impact on brand attitudes or purchase intention is strongly affected by a high similarity between a consumer’s self-concept and the image of a brand. Cf. Graeff (1996), p. 13; Parker (2009), p. 175 et seq.; Pradhan/Duraipandian/Sethi (2016), p. 457; Schade (2012), p. 77.

Sinnig (2019) for example refers to psychological constructs such as identification theory, homo-phily, or self-esteem when explaining consumer characteristics and their relationship with SMIs. For a detailed examination of the concepts cf. Sinnig (2019), pp. 49 et seqq.

For a detailed examination of different consumer-related concepts cf. e.g. Casalo/Flavian/Ibanez-Sanchez (2019); Dhanesh/Duthler (2019); Ki/Kim (2019); Lou/Yuan (2019); Schouten/Janssen/Verspaget (2019).

Cf. Burmann et al. (2018), p. 255; Flynn/Goldsmith/Eastman (1996), p. 137; Granados (2017), p. 5; Killian (2017), p. 64.

Cf e.g. Jaakonmäki/Müller/van Brocke (2017); Klein/Becker (2018); Lou/Yuan (2019); Schouten/Janssen/Verspaget (2019); Sinnig (2019).

Cf. Klein/Becker (2018), p. 6. The authors conducted a 3 (influencer type) × 3 (advertising disclosure) × 2 (multiple product endorsements) online experiment with n = 1,238 respondents.

Cf. e.g. de Jans/Cauberghe/Hudders (2018); de Veirman/Hudders (2019); Ki/Kim (2019); Lim et al. (2017); Lou/Yuan (2019); Schouten/Janssen/Verspaget (2019); Sokolova/Kefi (2019); Stubb/Colliander (2019); Stubb/Nyström/Colliander (2019); Xiao/Wang/Chan-Olmsted (2018). For a detailed examination of the source credibility model see Section 2.​3.​1.​1.

In academia, the term fit represents the subjective evaluation of a relationship between a brand and another image object, in this case the SMI. Cf. Baumgarth (2000), p. 48; Nitschke (2006), p. 28. The fit-construct will be further specified in Section 2.​3.​1.​2.

Corresponding to the fit-construct, the match-up hypothesis was subject of analysis within celebrity endorsement research. Cf. e.g. Kamins (1990); Till/Busler (2000).

Cf. Lienemann (2020); Sinnig (2019), p. 43.

Cf. e.g. Bergkvist/Zhou (2016); Fink/Cunningham/Kensicki (2004); Kahle/Homer (1985); Kamins (1990); Kamins/Gupta (1994); Solomon/Ashmore/Longo (1992); Till/Busler (2000).

To the best of the author’s knowledge solely the publications of Breves et al. (2019); Evans et al. (2017); Lienemann (2020); Schouten/Janssen/Verspaget (2019); and Torres/Augustos/Matos (2019) to some extent included the SMI/brand-fit as an influential (moderating) variable.

Cf. Audrezet/de Kerviler/Moulard (2018), p. 5; Russell/Belch (2005), p. 74.

Cf. e.g. Brennan/Dubas/Babin (1999); d’Astous/Chartier (2000); Gupta/Lord (1998); Karrh (1998); Law/Braun (2000); Russell (2002); Yang/Roskos-Ewoldsen (2007).

Cf. Jin/Muqaddam (2019). The authors conduct a 2 (source: brand vs. SMI) × 2 (brand placement: brand-only vs. SMI with brand) between-subjects experiment with n = 304 females. Results suggest that a concurrent presentation of both SMI and brand leads to higher perceived SMI credibility (i.e. trustworthiness and expertise) and brand attitude when compared to a brand-only presentation. Cf. Jin/Muqaddam (2019), pp. 528 et seqq.

Cf. e.g. Cheung/Thadani (2012); Ducoffe (1996); Kumar et al. (2016); Lee/Hosanagar/Nair (2018); Lin/Lu/Wu (2012); Tellis (2004).

Cf. e.g. de Vries (2019); Evans et al. (2017); Hughes/Swaminathan/Brooks (2019); Jaakonmäki/Müller/van Brocke (2017). Hughes/Swaminathan/Brooks (2019) for instance could prove that the integration of giveaways as well as content publication on the weekend positively affects user engagement, i.e. the number of likes, on Facebook. Cf. Hughes/Swaminathan/Brooks (2019), p. 88. The impact of differing disclosure languages—i.e. #SP (an abbreviation of sponsored), #Sponsored and #PaidAd—used in an SMI post was investigated by Evans et al. (2017). Their results suggest that using explicit, non-ambiguous disclosure language such as #PaidAd or #Sponsored have the strongest (indirect) negative impact on attitude towards the brand. Cf. Evans et al. (2017), pp. 142 et seqq. For a detailed examination of relevant control variables in the context of user engagement in influencer branding see Section 2.​3.​4.

Cf. e.g. Hughes/Swaminathan/Brooks (2019), p. 78; mediakix (2019).

With regard to the definition of influencer branding, an influencer branding campaign can be described as the actual manifestation of a communication message within a social media platform of an SMI. Cf. Bruhn (2013), p. 7; Meffert/Burmann/Kirchgeorg (2015), p. 586.

Cf. Gräve (2019), p. 1. Several studies emphasize the importance of measuring influencer branding campaigns. Launchmetrics (2018) is unveiling that the measurement of the effectiveness of influencer branding campaigns is perceived as one of the main challenges (28.5 percent). In total, 600 professionals in marketing, communications and PR from fashion, luxury and cosmetic industries answered the question “What do you consider as your main challenge when it comes to implementing your company’s influencer marketing programs?”. Cf. Launchmetrics (2018), p. 37. A similar result is shown in the previous year’s report for influencer branding in general. Accordingly, 43 percent of the surveyed marketers consider the measurement of the effectiveness of influencer branding campaigns as the main challenge. In total, 600 PR, Marketing and Communication professionals answered the question “What do you consider as your main challenge when it comes to implementing your company’s influencer marketing programs?” Cf. Launchmetrics (2017), p. 15.

Cf. Burmann et al. (2018), pp. 263 et seqq.; Kamps/Schetter (2018), p. 7.

Cf. Enke/Borchers (2018), p. 194 et seq.

Cf. here and in the following Burmann et al. (2018), pp. 264–269 referring to Tuten/Solomon (2015), p. 301.

Cf. Zerres/Litterst (2017), p. 194 et seq.

Cf. Zerres/Litterst (2017), p. 204 et seq.

Cf. Gräve/Greff (2018), p. 295.

For an overview of further quantitative KPIs in social media advertising cf. Burmann et al. (2018), pp. 264 et seqq.; BVDW (2016), pp. 10 et seqq.; Zerres/Litterst (2017), pp. 198 et seqq.

Cf. SproutSocial (2018b).

Cf. BVDW (2016), p. 36.

Cf. Hootsuite (2019a). Note: The terms engagement and interaction are often used interchangeably. In the following the more common term engagement will be used.

Burmann et al. (2017), p. 57 referring to Aaker (1991).

Cf. Zerres/Litterst (2017), p. 204.

For an overview of further qualititative KPIs in social media advertising cf. Burmann et al. (2018), pp. 264 et seqq.; BVDW (2016), pp. 10 et seqq.; Zerres/Litterst (2017), pp. 198 et seqq.

Cf. here and in the following Burmann et al. (2018), pp. 264–269 referring to Tuten/Solomon (2015), p. 301.

Burmann et al. (2017), p. 56 referring to Foscht/Swoboda (2011); Trommsdorff (2011).

Kleine-Kalmer (2016), p. 93 referring to Park et al. (2010), p. 2.

Note: The numbers in brackets indicate the absolute frequencies of mentions within the survey.

Cf. SimplyMeasured (2017), p. 13.

In former studies of Linquia engagement is operationalized with the key performance metrics number of likes, number of comments, and number of retweets or number of shares. Cf. Linqia (2018).

Further studies focussing on influencer branding in different industries support these results, cf. e.g. InfluencerMarketingHub (2020), p. 34; Launchmetrics (2017), Launchmetrics (2018).

Cf. van Doorn et al. (2010), p. 254; Rossmann/Ranjan/Sugathan (2016), p. 542. The derivation of the definition of user engagement as well as a distinction of common user engagement metrics will be further specified in Section 2.​2.

Cf. here and in the following Boerman (2020), p. 200; Domingues Aguiar/van Reijmersdal (2018); Khan (2017), p. 236.

Such algorithms “make sense of user behavior based on underlying assumptions about how users will behave and what that behavior signifies”. Cotter (2019), p. 899 referring to de Laat (2017).

Cf. Cotter (2019), p. 895; Instagram (2016); Instagram (2018); SproutSocial (2020a).

Cf. Cotter (2019), p. 896; Rader/Cotter/Cho (2018), p. 1 et seq.

Cf. Agung/Darma (2019), p. 744 et seq.; Cotter (2019), p. 902 et seq.; Hallinan/Striphas (2016), pp. 117 et seqq. Further dimensions of the algorithm are assumed to involve content-driven functions, e.g. image recognition algorithms which automatically categorize users’ interests regarding topics and related SMIs as well as a penalty system which aims to detect fake SMIs. Unfortunately, concrete information on how the respective algorithms work are not provided. Cf. SproutSocial (2020a).

Cf. Luarn/Lin/Chiu (2015), p. 507 referring to Brodie et al. (2013); Pham/Avnet (2009); Schau/Munz/Arnould (2009).

Cf. Jahn/Kunz (2012); Piehler et al. (2019).

Cf. Jahn/Kunz (2012); Pöyry/Parvinen/Malmivaara (2013).

Cf. Casalo/Flavian/Guinaliu (2007), Hollebeek (2011); Kang/Tang/Fiore (2014).

Cf. Chan/Li (2010); Jahn/Kunz (2012); Kang/Tang/Fiore (2014).

Cf. Andersen (2005); Bowden (2009); Casalo/Flavian/Guinaliu (2007); de Vries/Carlson (2014); Jahn/Kunz (2012).

Cf. Bowden (2009); Kumar et al. (2010); Neff (2007).

For a more detailed examination of the outcomes of user engagement see Section 2.​2.​3.

Cf. Hughes/Swaminathan/Brooks (2019), p. 78; Jaakonmäki/Müller/vom Brocke (2017), p. 1153; Sokolova/Kefi (2019), p. 2.

Cf. Boerman (2020), p. 200; Cvijikj/Michahelles (2013), p. 845; Hughes/Swaminathan/Brooks (2019), p. 79; Jaakonmäki/Müller/vom Brocke (2017), p. 1152; Piehler et al. (2019), p. 1834.

The authors define the engagement rate as “the quantity of responses and interactions that content on social media generates from users”. Jaakonmäki/Müller/vom Brocke (2017), p. 1152.

After extracting all relevant features the authors obtain a dataset consisting of 768 features or single words for the 13,369 posts analysed. Cf. Jaakonmäki/Müller/vom Brocke (2017), p. 1155.

Cf. Jaakonmäki/Müller/vom Brocke (2017), p. 1152. The authors started with a total set of 140,000 Instagram posts published by German-speaking bloggers that was then reduced randomly.

Cf. Jaakonmäki/Müller/vom Brocke (2017), p. 1152.

Cf. here and in the following Jaakonmäki/Müller/vom Brocke (2017), p. 1157.

Cf. Audrezet/de Kerviler/Moulard (2018), p. 5

Cf. e.g. de Vries/Gensler/Leeflang (2012); Gavilanes/Flatten/Brettel (2018); Lee/Hosanagar/Nair (2018).

Cf. Jaakonmäki/Müller/vom Brocke (2017), pp. 1155 et seq.

Cf. Jaakonmäki/Müller/vom Brocke (2017), pp. 1158.

Cf. Hughes/Swaminathan/Brooks (2019), p. 78.

Mommy bloggers are another common type of SMIs, who typically write about their children or issues surrounding their pregnancy. Cf. Archer (2019), pp. 47 et seq.; Lopez (2009), p. 734.

The initial data sample consists of n = 1,830 posts from 595 mommy bloggers across 57 different influencer branding campaigns published between September 2012 to December 2016. Due to reasons of data cleansing the data sample was reduced to n = 1,237 blog posts. Cf. Hughes/Swaminathan/Brooks (2019), p. 84.

Cf. Hughes/Swaminathan/Brooks (2019), pp. 84 et seq. User engagement on blogs was mea-sured via the total number of comments a blog post received; on Facebook the number of likes per post was used as engagement metric.

In their study, the authors measure the number of followers as the average number of all followers a blogger has on either Facebook or Twitter. Thus, this variable is assumed to represent the bloggers social media presence. Cf. Hughes/Swaminathan/Brooks (2019), p. 84.

The hedonic value of a post “refers to the enjoyment, emotions, and entertainment a consumer experiences from reading a post.” Hughes/Swaminathan/Brooks (2019), p. 84. On the converse, the authors define the functional value of a post as “the believeability and informativeness of a post”. Hughes/Swaminathan/Brooks (2019), p. 84.

Giveaways are any marketing actions which are “designed to generate specific responses and engagement from customers”. Cf. Hughes/Swaminathan/Brooks (2019), p. 82.

Cf. Hughes/Swaminathan/Brooks (2019), p. 88.

Cf. Hughes/Swaminathan/Brooks (2019), p. 86.

Cf. Hughes/Swaminathan/Brooks (2019), p. 88. Further IVs and control variables, i.e. day of posting, and a functional value of a post, had no significant impact on user engagement on blogs.

The number of Facebook posts indicates the total number of Facebook posts of a blogger per campaign. Cf. Hughes/Swaminathan/Brooks (2019), p. 84.

Cf. here and in the following Hughes/Swaminathan/Brooks (2019), p. 88.

The authors argue that the positive impact of giveaways for blog post comments implicitly shows a cannibalization effect for Facebook post likes. Cf. Hughes/Swaminathan/Brooks (2019), p. 92.

Cf. Hughes/Swaminathan/Brooks (2019), p. 89.

Cf. Hughes/Swaminathan/Brooks (2019), pp. 84 et seq.

Cf. e.g. de Vries/Gensler/Leeflang (2012), p. 84; Piehler et al. (2019), p. 1834; Rossmann/Ranjan/Sugathan (2016), p. 543.

Although the possibilities of liking a blog post are sometimes limited, the authors could have compared blog post comments and Facebook post comments in order to ensure comparable results within their two models.

Cf. Hughes/Swaminathan/Brooks (2019), pp. 84 et seq.

Hughes/Swaminathan/Brooks (2019), p. 84.

Cf. Hughes/Swaminathan/Brooks (2019), p. 86.

Cf. Hughes/Swaminathan/Brooks (2019), p. 84. It can be assumed, that the number of followers a blogger has on his/her Twitter account does not influence user engagement with published posts on Facebook.

Therefore, the authors analysed a sample of n = 264 participants via Amazon MTurc. The results suggest that Facebook is a rather low-involvement platform as it is characterised by a high degree of distraction and a comparably lower degree of specific content seeking. Cf. Hughes/Swaminathan/Brooks (2019), p. 81.

In respect of several debates regarding the topic of non-disclosed posts from SMIs, this topic has gained great attention in public. As posts from influencers promoting a brand are very similar to traditional product placements, the Federal Trade Commission (FTC) announced new regulations for these specific posts, including a requirement to disclose the partnership between brand and influencer. Cf. Ad Standards (2018); Federal Trade Commission (2017).

In the study user engagement is measured with the following six items on a seven-point scale (likeability of approval): “I would share the post with my friends on Instagram”, “I would comment on the post on Instagram”, “I would like the post on Instagram”, “I would share the post on Facebook”, “I would share the post via a private message on Instagram”, and “I would save the post on Instagram”. Cf. Boerman (2020), p. 203.; Boerman/Willemsen/van der Aa (2017), p. 87; Evans et al. (2017), p. 143.

Ad recognition indicates to what extent the participants perceive the Instagram post as advertisement. Cf. Boerman (2020), p. 203.

Cf. Boerman (2020), pp. 204 et seq. However, the author found a significant indirect effect of disclosure on user engagement mediated by ad recognition. In regard to the non-significant direct effect and the overall rather low engagement levels within the data sample, this indirect effect should not be given too much weight. Cf. Boerman (2020), p. 205.

Cf. e.g. de Vries/Gensler/Leeflang (2012), p. 84; Piehler et al. (2019), p. 1834; Rossmann/Ranjan/Sugathan (2016), p. 543.

In particular because previous research has proven significant negative effects of ad recognition on outcomes such as attitude towards a sponsored blog post, intention to spread eWOM or purchase intention. Cf. e.g. Evans et al. (2017), p. 144; Hwang/Jeong (2016), p. 532.

Cf. Jaakonmäki/Müller/vom Brocke (2017), p. 1154.

Cf. e.g. de Vries/Gensler/Leeflang (2012), p. 84; Gavilanes/Flatten/Brettel (2018), p. 8; Li/Xie (2020), p. 3; Piehler et al. (2019), p. 1834; Rietveld et al. (2020), p. 34; Rossmann/Ranjan/Sugathan (2016), p. 543; Tabellion (2019), p. 208.

Cf. Kopietz/Echterhoff (2016), p. 590 et seq.

Note: bold letters indicate a statistical significant impact on the measured DVs

Cf. Hughes/Swaminathan/Brooks (2019), p. 79; Lee/Hosanagar/Nair (2018), p. 2. Moreover, this type of data can be gathered in a cost-efficient manner and thus, enables the explanation of user engagement drivers at a larger scale.

Unfortunately, the authors do not provide a complete overview of all investigated variables. Solely the aforementioned SMI-related drivers could be identified in their publication.

In simple terms: Without the integration of a brand, a concept such as influencer branding could not be existent. Cf. Audrezet/de Kerviler/Moulard (2018), p. 4.

Unfortunately, the authors do not provide a complete overview of all investigated variables. Solely the aforementioned content-related drivers could be identified in their publication.

Cf. e.g. Boerman (2020), p. 206; de Veirman/Hudders (2019), p. 24; Hughes/Swaminathan/Brooks (2019), p. 78; Schouten/Janssen/Verspaget (2019), p. 19; Sinnig (2019), p. 201; Torres/Augusto/Matos (2019), p. 1275; Xiao/Wang/Chan-Olmsted (2018), p. 206.

The integration of a brand in SMIs post is comparable to traditional product placement in (online) advertisements. Cf. Audrezet/de Kerviler/Moulard (2018), p. 5; Jin/Muqaddam (2019), p. 523.

A similar procedure is conducted by the acknowledged publication of de Vries/Gensler/Leeflang (2012), p. 84.