Introduction

Social media and social networking have risen exponentially among consumers (Lou and Yuan 2019; Lee et al. 2023). In this scene, personal information sources capable of influencing consumption decisions gain weight. The term social media influencer has risen in popularity among scholars to describe “online personalities with large numbers of followers, across one or more social media platforms (e.g. YouTube, Instagram, Snapchat, or personal blogs), who influence their followers” (Lou and Yuan 2019 p. 58). Influencers have also been referred to as popular social media users, social media stars, or micro-celebrities (Hudders et al. 2020). This figure has played a relevant role in sports, fashion, and food.

The study of influencers started seriously in 2018, but, as Taylor (2020) has remarked, influencer marketing seems to be reentering during COVID-19 due to more time spent at home, leading to more internet time and less social interaction. Diverse studies have partially reviewed the birth and evolution of this term. For example, Hudders et al. (2020) did a search using the Scopus database with a total of 154 peer-reviewed academic publications focussing on virtual influencer marketing, published mainly in the last 3 years (2018–2020). One year earlier, three other studies started to be interested in reviewing the evolution of virtual influencer marketing. First, Nafi and Ahmed (2019) focussed on the ethics of influencers’ marketing in the tourism industry based on 49 different articles, internet columns, conferences, and book reviews. Second, Sundermann and Raabe (2019) analysed 39 papers about the role of influencers in strategic communication. And third, De Veirman et al. (2019) developed a review of influencing marketing focussed on the children segment.

However, the revisions on these themes are scarce and incomplete. Although research on celebrity endorsement is old, the study of social media influencers remains underinvestigated (Lou and Yuan 2019). So, a deep bibliometric analysis is needed to cover research gaps and identify future lines of research in this field.

In this framework, the present paper has been developed with two main objectives: (i) to carry on a performance analysis to measure the visibility/impact of the scientific production on influencers’ marketing (most cited authors, journals, and themes) and (ii) to visually present the scientific structure by topics of research in influencers marketing as well as its evolution along time in industries such as tourism (sports and leisure) or food among others. A sample of 1703 indexed papers was considered: (i) from 1997 to 2012 (649 papers), (ii) from 2013 to 2016 (540 papers), and (iii) from 2017 to 2021 (514 papers). The methodology used (bibliometric analysis based on SCIMAT software) has been used previously to obtain interesting results to analyse the evolution of other topics from a scientific approach: product packaging (Vila-Lopez and Küster-Boluda 2021), product placement (Vila-Lopez and Küster-Boluda 2023), sharing economy (Vila-Lopez and Küster-Boluda 2022), and complaints in b2b (Küster-Boluda et al. 2023). SciMAT is unique in expressing the evolution of the discipline and theme (Xie et al. 2020). The usefulness of using cognitive mapping to investigate research marketing trends can also be appreciated in the work of Petrescu and Krishen (2023).

Compared to previous works, this research adds value as long as a multidisciplinary approach is adopted, including works from business and management, computer science, communication, food, and sociology journals. In addition, three additional contributions are made compared to previous biometric analyses on this topic. Compared to Vrontis et al. (2021) review (based on 68 articles from 29 Chartered Association of Business Schools), this work follows an interdisciplinary approach. Second, compared with the revision of Tanwar et al. (2022) in the influencer marketing field (based on 76 articles retrieved from the Scopus Data since 2011), this paper follows a longitudinal approach, showing differentiated results for three different periods defined because two turning points were identified in 2013 and 2017 (from 1997 to 2012, 649 papers; from 2013 to 2016, 540 papers; and from 2017 to 2021, 514 papers). Third, only indexed papers in the Web of Science base in many disciplines were considered to ensure the rigour of the articles analysed. The suitability of this procedure can be seen in works published in high-impact journals (Rodríguez-López et al. 2019, Xie et al. 2020, or Küster-Boluda et al. 2023 among others). Fourth, as previous works have demonstrated (Petrescu and Krishen 2023), dynamic scientific maps were visually shown using perceptual representations to clearly present the outstanding topics, their relationships, and their evolution.

Theoretical framework

Industries

A review of research on influencers in marketing shows that various industries have been investigated. The food industry features prominently, with influencers often referred to as “foodies”. The pioneering work of Cheung et al. (2008) focussed on users of an online community (OpenRice.com) about food and restaurants. They concluded that restaurant managers could join Openrice.com to provide comprehensive information (like pictures, menu, location, and price) to stimulate responses. Many times, the use of influencers in this industry has been criticised because children’s audiences wanted to be reached to encourage their consumption of not fully healthy foods (De Veirman et al. 2019).

The film sector has also been investigated in the leisure industry from the influencers’ perspective. In this scene, influencers are often referred to as “critics”. As Basuroy et al. (2003) explain, they are influencers, but they can also be considered predictors if they merely predict consumers’ decisions. As these authors recommend, the power of critics could be better exploited by other industries such as theatre and performance arts, book publishing, recorded music, and financial markets.

In addition, other industries, such as tourism (Leong et al. 2021) or politics, have also been deeply investigated from the influencing marketing approach. For example, in tourism, sports celebrities have strongly influenced different audiences, sometimes helping to reinforce the image of the country of origin they come from (Wu 2015). From another approach, Nafi and Ahmed (2019) concluded for the tourism industry that before making any purchase decisions, most hospitality consumers use e-servicescape platforms to find reviews and rely on consolidated influencers’ opinions. In the political industry, the role of influencers as “opinion leaders” has also gained weight in recent years (Lotan et al. 2011), being Twitter an outstanding network to this end.

In sum, influencer marketing seems to work for tangible products and a wide range of services. These influencers are often crucial to explaining the image of their country of origin insofar as there is a subjective source effect that transfers the perceptions/attitudes developed towards the person to the perceptions/attitudes created towards the place they come from. As the pioneering work of Eliashberg and Shugan (1997) stated, the role of critics and influencers is not negligible in entertainment industries (movies, theatre, music, television, and other recreational services such as sports).

Influencers’ theories

Not only have the industries analysed been varied, but the starting theories used to support the use of influencers as a marketing tool have also been varied. First, the Subjective Source Effect Theory postulates that an appropriate source of a message encourages word-of-mouth marketing (WOMM) (Vrontis et al. 2021). This is defined by Kozinets et al. (2010) as the intentional influencing of consumer-to-consumer communication. The source is named “key influencer” in the work of Cheung et al. (2008), and the main point is that an influencer always receives any kind of compensation (free products or money); otherwise, it cannot be considered an influencer. (De Veirman et al. 2019). As previous lines have remarked, the Source Effect often connects the image of one person with the image of the place this person comes from, which is key in shaping destination images (Wu 2015).

Second, De Veirman et al. (2019) used the Identification Theory to explain how a stronger identification between the audience and the message’s source will increase the likelihood of adopting source beliefs, attitudes, and behaviours. A meaning transfer from the source to the audience is below this identification.

Third, this author also used the Social Learning Theory to explain how the audience will likely imitate the source of the message behaviour, including the products they use (De Veirman et al. 2019). This social learning will increase if the audience develops a parasocial relationship with the source. From this approach, Lotan et al. (2011) used the Two-Step Flow Communication Theory to explain information-sharing behaviour. As these authors point out, the information diffusion models help explain how an ideological contagion encouraged by influencers is used to target consumer responses, known as “information cascades”. Linked to the former, Lou and Yuan (2019) mention the Communication-Persuasion Theory to explain the effectiveness of this communication tool based on different elements, such as the source, the message, the channel, the receiver, and the destination. What an influencer says, does, or wears is contagious, provoking love for this person and often for their country of origin.

Finally, the relationship between the message’s source and the audience has been analysed many times from the Attribution Theory perspective. According to this theory, the audience will respond and accept the message if they do not feel that the influencer is taking advantage; that is because, from this theory, the role of a given cause in constructing a given effect is discounted if other probable causes are also present (Stubb et al. 2019).

Classification of influencers’ research: source, message, and audience

The Communication Model for Advertising, reviewed by Hudders et al. (2020), was a starting point to base the studies on influencer marketing on three pillars: source, message, and audience. These three exes represent the three critical entities stated by Lou and Yuan (2019) to explain the effectiveness of this tool: influencers (source), brands (message), and consumers (audience).

Studies focussed on the source of the message

As the theories mentioned in previous lines support, with the irruption of influencers, communication has evolved from “organic” (it occurs between one consumer and another without direct prompting, influence, or measurement by a marketer) to “linear”, which uses opinion leaders who smart marketers could target and influence (Kozinets et al. 2010). These agents are paid to amplify the message to be expanded.

However, as previous literature has demonstrated, diverse influencer profiles exist (foodies, sportsmen/sportswomen, politicians, etc.). Each company must choose the one that better fits its values (Leong et al. 2021). For example, Kozinets et al. (2010) provide some blogger archetypes: the citizen journalist, the tell-it-like its mother, the satirical exhibitionist, and the making-ends-meet professional blogger, among others.

A good influencer must be “credible”, which means perceived to be believable, competent, and trustworthy by information recipients (Cheung et al. 2008). More specifically, credibility can be unfolded on two main dimensions: source expertise and source trustworthiness (De Veirman et al. 2019). Therefore, the underlying property of every influencer should be its prominence due to their work on social media platforms only. In comparison to celebrities, they can alter the initial advertising message to make it more authentic (Sundermann and Raabe 2019).

Furthermore, the source’s physical appearance or “attractiveness” plays a major role in reinforcing the endorsers’ credibility (De Veirman et al. 2019). According to literature reviews, source attractiveness is driven by factors such as the source’s perceived similarity, familiarity, and likeability.

On the contrary, their previous formation has not been taken into consideration by the literature. Influencers often do not have proper training (Nafi and Ahmed 2019), meaning they create and share information but follow their own criteria because of the lack of formation.

Studies focussed on the content of the message

Regarding the content, different types of effective messages have been proposed in the form of a video (i.e. YouTube video, Instagram story), a written post (i.e. blog entry or Twitter), or a mixture of a picture and a text (Instagram post) (Sundermann and Raabe 2019). But, in all the cases, the common denominator is that advertisers have control over the content, which also comprises simple final approval of the output or general instructions regarding the output (De Veirman et al. 2019). Sometimes, celebrities are mere professionals in their field of work whose success at work makes them a benchmark, as would be the case for the sports industry.

Kozinets et al. (2010) conclude that written messages (in blogs) will succeed if they are prepared following four social media communication narrative strategies: (i) evaluation message (focussed on the product and/or on the community); (ii) explanation message (focussed on the source); (iii) embracing message (focussed on personal needs using professional language); and (iv) endorsement message.

In any case, Cheung et al. (2008) summarised that a quality message must be elaborated following four main recommendations: relevance, timeliness, accuracy, and comprehensiveness. Quality and perceived usefulness of the message are two faces of the same coin to enhance or improve one’s performance. Therefore, if the content of a message is of quality, it will be perceived as useful and will work. For Lou and Yuan (2019), a quality message is a mix of its perceived formative and entertainment values.

Another point to consider when defining the content of the message to seem transparent is the need to remark that the content is sponsored. As previous literature has reviewed (Stubb et al.. 2019), a sponsorship compensation justification disclosure in the transmitted message will generate more positive effects than a simple sponsorship disclosure or when no disclosure is present (messages have been also analysed from a sentimental approach). In this case, the effects of positive and negative messages provided by influencers have been compared (Basuroy et al. 2003), expecting negative reviews to hurt box office performance more than positive reviews help. Interesting exceptions have been found in the sports industry.

Studies focussed on the audience of the message

The audience of influencers’ messages is expected to react subjectively in terms of engagement, persuasion, and interaction instead (Kozinets et al. 2010). As Cheung et al. (2008) remark, engagement occurs when a process of information adoption starts. These desired effects are simplified as information adoption. Stubb et al. (2019) separated the desired effects of sponsored content into attitude improvement and more positive consumer responses. In addition, and linked to the former, the audience is expected to react objectively in monetary terms (Basuroy et al. 2003), for example, when visiting a tourism destination.

On the contrary, many times, the effects are adverse. That is because the audience feels disappointed when influencers promote consumption without making it clearly explicit. In this case, the content shown on social media might hurt audience reactions (Stubb et al. 2019).

Concerning the length of time of the audience’s reactions, diverse studies have been worried about the durability of the opinion leader’s influence over time. For example, Basuroy et al. (2003) stated that an influencer is expected to have the most effect in the early phases of a film’s run before word of mouth has a chance to spread. The same results were obtained in the work of Eliashberg, and Shugan (1997) regarding film critics’ influence on box office revenues.

In sum, we can conclude that the research on virtual influencer marketing has been developed from different approaches. Following previous bibliometric analysis on this field (Kim and McMillan 2008; Montero-Díaz et al. 2018), the two main objectives presented in the introduction were concretised in three research questions. The first two questions addressed our first objective (measuring the visibility/impact of the scientific production), and the last question addressed the second objective (identifying the scientific structure and evolution of the principal themes):

  • RQ1: Which authors have obtained the highest visibility and impact in the field of influencer marketing in the Web of Science? That is, who are the most cited authors?

  • RQ2: Which journals have obtained the highest visibility and impact in the field of influencer marketing in the Web of Science? That is, which are the most cited journals?

  • RQ3: Which themes have obtained the highest visibility and impact in the field of influencer marketing in the Web of Science? That is, what are the most cited product placement papers? Have they evolved?

Methodology

Software

Our bibliometric analysis allowed a quantitative analysis of all scientific publications indexed in the Journal Citation Reports containing the keywords “virtual influencers”, or “online influencers”, or “foodies”, or “bloggers”, or “Youtubers”, or “peer endorsement”, or “virtual endorsers”, or “online endorsers”. To this end, SciMAT software was used (Cobo et al. 2013). The use of the resources of Web of Science to carry on a bibliometric analysis of generic communication themes can be shown in Muñoz-Leiva et al. (2015) and Montero-Díaz et al. (2018), among others. As Muñoz-Leiva et al. (2015) have explained, this software uses co-work analysis to identify the interests and aspirations of academic researchers. “This technique reduces a large set of descriptors (or keywords) to a set of network graphs that effectively illustrate the strongest associations between descriptors” (p. 682).

Data collection and data set

We retrieved the information from the ISIWeb of Science. A final set of 1740 papers containing these keywords was analysed. Citation counts to April 1, 2021, were considered (15,773 citations within those papers were recorded and analysed). The period embraced was from 1999 (when the first indexed paper was published) until 2021.

Data collection was based on Cobo et al. (2013), applying a de-duplicating process over the keywords. The author’s keywords and the Keywords Plus were considered to group words representing the same concept. Some keywords meaningless in this context, such as stop words or words with a very broad and general meaning (Cobo et al. 2013), were removed (for example, the generic themes “influencer” or “communication”).

To answer RQ1 and RQ2, citation analysis was used. This is a method of tracking publishing patterns based on the assumption that a heavily cited author, paper, or book is considered important by many scholars in a discipline (Kim and McMillan 2008). To solve RQ3, three scientific maps were obtained and compared for three consecutive periods.

To avoid subjectivism, a preliminary analysis was presented at a Marketing Trends Conference in a work-in-progress session (Vila et al. 2023). More specifically, two steps were followed. In the first one, two anonymous reviewers of our proposal provided feedback about the content’s relevance before accepting it. In the second one, when the work in progress was presented, three skillful marketing doctors in the room helped us with the methodology to, for example, guarantee the correct grouping of the keywords to represent the same concept and to interpret the graphics properly.

Periods of time

Based on the number of papers per year, three main periods were identified (Fig. 1): (i) from 1997 to 2012 (649 papers), (ii) from 2013 to 2016 (540 papers), and (iii) from 2017 to 2021 (514 papers). The idea was to analyse a comparable number of papers per period. In 2013, an augment of documents was observed until 2016 because technologies advanced and social networks multiplied; then, in 2017, a second turning point was identified.

Fig. 1
figure 1

Number of papers per year

Full size image

Results

Most prolific authors (RQ1)

Figure 2 shows that the most cited work is that from Kozinets et al. (2010) published in the Journal of Marketing (about how bloggers generated Ewom through mobiles). Second, we find other interesting references: the work of Lotan et al. (2011) in the International Journal of Communication (about how information flows during the 2011 Tunisian and Egyptian revolutions using Twitter), the work of Lewandowsky et al. (2013) (related to how NASA faked the moon landing), and the work of Yarkoni (2010) (about how bloggers personality was linked to differences in linguistic style in their blogs).

Fig. 2
figure 2

Most cited works

Full size image

Based on the number of documents, Fig. 3 shows the outstanding position of Nitin Agarwal, from the Information Science Department at the University of Arkansas at Little Rock, USA (Hindex = 25, Hi10 = 59), Distinguished Professor of Information Science. His research was mainly focussed on bloggers and their influence on a community. Other authors with extensive research in this field are Amelia Burke-Garcia, Digital Health Communicator actually in charge of digital strategy in NORC at the University of Chicago; Khan Hikmat Ullah, PhD.(Computer Scienc`e) from COMSATS at the University Islamabad, an expert in computer science and data mining and Kostas Zafiropoulos, Professor of Quantitative Methods at the University of Macedonia (Greece) and specialised in the political arena.

Fig. 3
figure 3

Authors and number of papers. Note: The top 25 authors have selected

Full size image

So, answering RQ1, we can conclude that the study of influencers has been approached from different perspectives, the dominant ones being marketing in general, communication, and psychology. Then, this can be understood as a multidisciplinary topic of research. Top researchers are not concentrated only in the USA; also, relevant contributions have been developed in Europe (i.e. Greece) and Asia (i.e. Pakistan).

Most prolific journals (RQ2)

As Figs. 4 and 5 show, the most prolific journals in articles on this topic belong mainly to three research areas: communication (such as Journalism, New Media Society or International Journal of Communication), computers science (such as Lectures in Computer Science, Computers in Human Behaviour or Journal of Computer Mediated Communication), and information science (such as Information Communication Society or Online Information Review).

Fig. 4
figure 4

Journals and number of papers. Note: Only the top 25 journals are shown

Full size image
Fig. 5
figure 5

Areas of research

Full size image

Answering RQ2, we can conclude that research on influencers has been published in journals related to the specific field of communication, and new information technologies. Therefore, other areas, such as Social Sciences in general, Psychology, Business Management, Tourism, etc., remain less explored.

Answering RQ2, we can conclude that most prolific journals with articles on this topic belong mainly to three research areas: communication (such as Journalism, New Media Society, or International Journal of Communication), computer science (such as Lectures in Computer Science, Computers in Human Behaviour or Journal of Computer Mediated Communication), and information science.

Trending topics in the field of influencer marketing and evolution (RQ3)

The key topics in these three periods of time have not always been the same, as Fig. 6 shows.

Fig. 6
figure 6

Evolution of the trending topics on influencers marketing (h-index)

Full size image

Topics concerning the “media” and the “industry” have evolved to analyse new media and different industries. That is, strong topics in period 1 declined in period 2 and disappeared in period 3 (BLOGS, HEALTH, and LANGUAGE), while new topics concerning the media and the industry have replaced them, as the h-index corroborates. More specifically, (i) BLOGS declined because other communication tools have irrupted and flourished over the last 5 years, such as INSTAGRAM or SOCIAL MEDIA; (ii) HEALTH industry lost its relevance because other industries have become comparatively more important in the field of influencers marketing such as TOURISM as far as influencers from different sub-fields (such as sports or foodies) become relevant to position a tourism destination; and (iii) LANGUAGE also declined because another theme, IDENTITY, has replaced it. For example, the identity of a tourism destination can be reinforced, thanks to virtual ambassadors who speak well of it or who bring him success, for example, in terms of interesting restaurants to visit.

Topics concerning the “message” have remained more or less stable during the whole period (CONTENT). This means that the relevance of the message is a constant key point in influencer marketing.

Topics concerning the “source of the message” emerged in period 3 (CELEBRITIES and WOMEN), becoming trendy during the last 5 years.

Topics concerning the “audience of the message” gained weight in period 2, becoming trendy in period 3, that is, during the last 5 years. That is the case of COMMUNITIES.

Topics related to the “strategy/objectives” of influencer marketing have evolved. More specifically, the trendy topic INTENTION in period 1 evolved to DISCLOSURE in period 2, and this evolved to PERSUASION in period 3. Over the last 5 years of influencer marketing, an increased persuasive intent has become apparent.

The last row of Fig. 6 shows an increase in the keywords analysed (from 70 in period 1 to 79 in period 7). The shared keywords between periods also augment, from 69 (between period 1 and period 2) to 78 (between period 2 and period 3). The stability index (in parenthesis) has remained constant over time, which indicates that the community has been strengthening its vocabulary to describe published documents. So, a consolidated body of research on influencer marketing can be confirmed with our analysis.

To better understand the evolution of the trending topics during the three periods, Fig. 7 shows the strategic diagrams for each period with indicators used to analyse the centrality and density of each theme in this period: ‘centrality’ measures the degree of interaction of a network with other networks and ‘density’ measures the internal strength of the network (Cobo et al., 2014). Based on these two indicators (centrality and density), each diagram includes four groups of themes (for quadrants in the diagram): (i) motor themes, (ii) basic and transversal themes, (iii) emerging or declining themes, and (iv) highly developed and isolated themes. Sphere size is proportional to the number of published documents associated with each research theme (Cobo et al., 2014). In.

Fig. 7
figure 7figure 7

Strategic Diagrams based on h-index

Full size image

(i) Motor themes (high density and centrality): upper-right quadrant

The outstanding themes were BLOGS, IDENTITY, and INTENTION in period 1; BLOGS, HEALTH, TOURISM (foodies), and COMMUNITIES in period 2; and SOCIAL MEDIA, PERSUASION, WOMEN, and CELIBRITY in period 3. That is, motor themes on virtual influencer marketing research have evolved from pioneer social forums (blogs) to more modern ones (social media). Also, concern for health and tourism has given way to concern for women. In addition, the focus has shifted from intention to persuasion.

(ii) Declining or emerging themes (low density and low centrality): lower-left quadrant

Declining themes were in period 1 HEALTH; in period 2 CONTENT and LANGUAGE; and in period 3 CONTENT, IDENTITY, and INSTAGRAM. The health industry becomes an old one to be investigated, as previous lines have anticipated, while the topic content (the message) rises in relevance with time.

(iii) Basic-transversal themes (high centrality but low density): lower-right quadrant.

The basic themes in periods 1, 2, and 3 were CONTENT, DISCLOSURE, and COMMUNITIES, respectively. That is, a lot of trending topics in previous periods became merely basic topics in the next period.

(iv) Highly developed but isolated themes (high density but low centrality): upper-left quadrant

Those isolated themes investigated separately in periods 1, 2, and 3 have been LANGUAGE, PUBLIC RELATIONS, and TOURISM (leisure and sports), respectively. That is, research on these topics has followed its way.

Answering RQ3, we can observe that (i) new social media (i.e. Instagram) have replaced old ones (i.e. blogs); (ii) the health industry has lost weight in favour of the tourism industry; and (iii) researchers pay more attention to influencers identity than to its language resources. Women and children’s influencers are becoming increasingly important. The persuasive tone of the message gains prominence over other approaches. Content, communities, and disclosure are basic themes in the three periods.

Theoretical implications, managerial recommendations, and future research lines

The following lines show some theoretical implications, managerial recommendations, and future lines of research. As the annexed details, we will show some trending topics during the last 5 years, representing promising future research lines.

Theoretical implications

First, regarding the audience, and as De Veirman et al. (2019) support, our results confirm, from a theoretical perspective, the relevance of developing research focussed on minor audiences that remain unexplored, as previous literature has also demonstrated (Lee et al. 2023), for example, children. Recent studies identified in our analysis show that influencer marketing can reach out to niche demographic audiences more effectively than any other media (Nafi and Ahmed (2019). In the same vein, our results support the trend in favour of gender studies. As Su et al. (2021) summarise, females are relationship‐oriented and likely to communicate implicit doubts about online shopping in the absence of interpersonal activities (Dai et al. 2019). On the contrary, male consumers are focussed on objects (Pease and Pease 2003) and the functional benefits of online shopping (e.g. convenience); therefore, they are less sceptical about the attributes of digital platforms. Our results have highlighted the relevance of topics such as gender, women, and family, as the annex shows.

Second, regarding the content, we support the findings of Ghorbani et al. (2021, p. 23) as “it looks like the research hotspots of digital marketing have changed during the last two decades and researchers are increasingly focussed on new topics”. More specifically, our results demonstrate a growing line of research favouring ethics in business, particularly in marketing. Nafi and Ahmed (2019) remark that the power to formulate mass opinion is providing superfluous power to many untrained, biassed, and unethical persons, so “scientific literature related to the ethical considerations to be a social influencer is still hard to come by” (p.4).

Third, regarding the media, our proposal corroborates the relevance and evolution of influencer marketing through new social networks. As in the work of Ghorbani et al. (2021), our research corroborates a growing interest in digital marketing during the last decade. Previous studies identified in our proposal show how virtual influencer marketing has been developed on specific media in the past, such as YouTube (Stubb et al. 2019), Instagram, blogs, Facebook, Twitter, etc. Recent studies show a significant lack of research on this topic, and even more on specific networks such as TikTok (De Veirman et al. 2019), which frequently shows interesting restaurants to visit or tasty cooking recipes.

Regarding the source of the message, our results support the Subjective Source Effect Theory, which postulates that an appropriate source of a message encourages word-of-mouth marketing (WOMM) (Kozinets et al. 2010) as the intentional influencing of consumer-to-consumer communication. That is, the success of the message will depend on the source.

Regarding the methods, our results also confirm the recognition of Netnography as a necessary methodological innovation for research, as far as it overcomes the limitations of other qualitative methodologies, such as ethnography, action research, and qualitative case studies (Discetti and Anderson 2023).

Managerial implications

Regarding the audience, we recommend brand managers explore the role of children as influencers and peer receivers of the message. The annexe shows the connections between this target, CHILDREN, and other topics such as YOTUBE, GAME, or ENDORSEMENT. So, gamification can be used to promote brands focussed on young people through influencing marketing.

Second, regarding the content, managers should be aware of ethical requirements to avoid legal problems. In this line, several legal measures are being adopted in different industries, particularly in the food industry. For example, The Federal Trade Commission in the USA has remarked that any type of sponsored content by a third-party source must clearly reveal sponsorship information to avoid confusing the audience (Stubb et al. 2019). As these authors have demonstrated, managers need to configure the message, including sponsorship compensation justification disclosure.

Third, regarding the media, networks should continue to be included in a company’s communication strategy. Still, the type of social networks used should be updated, taking into account the target. In this area, a promising future is foreseen for influencer marketing through growing influencer networks such as TikTok.

Fourth, regarding the source of the message, companies must increasingly focus on the influential power of children and young people. As De Veirman et al. (2019) have reviewed, the number of child vloggers is flourishing. What is named “kid-influencers”. Linked to the message’s source, the study of children as potential influencers is gaining momentum (De Veirman et al. 2019).

Fifth, regarding the industry, our results show that influencer marketing is highly recommended for entrepreneurs in tourism, food, wine-alcohol, or fashion sectors. On the contrary, communication tool has lost weight in the health industry. Celebrities in the tourism and sports industry need to be better utilised.

Lastly, regarding the methods, this paper corroborates, in line with Petrescu and Krishen (2023), the usefulness of using cognitive maps to detect relevant insights regarding a specific topic and to present future trends for researchers. In this vein, we recommend managers be aware of new technologies’ advantages to measure the results of the influencer communication strategy.

Future research lines

First, regarding the audience, influencing marketing based on celebrities represents a growing trend to reach minority audiences and show them the right to eat and grow. So, more investigation on this communication tool focussing on specific niche audiences represents a promising line of research. On the other hand, and linked to the audience, gender studies represent a promising line of research as our results have confirmed, because our annexe show some topics investigated jointly during the last 5 years (GENDER, WOMEN, FAMILY, and AESTHETICS).

Second, regarding the content, another promising line of research is the study of influencers’ social marketing from an ethical approach. That is, questions such as photoshopping, fake news or unreal followers could represent novel research in the influencers’ marketing arena. Recent investigations illustrate this trend as our annexe shows by connecting CONTENT, ETHICS, SENTIMENTS, and PUBLIC RELATIONS.

Regarding the media, a comparative study of the impact of a particular source and/or message through diverse social networks at the same time remains underinvestigated. As the annex shows, other promising topics are SOCIAL-NETWORK, FACEBOOK, TWITTER, and BLOGS.

Regarding the source of the message, more research is needed on which influencer profiles work best for each target and in each culture. As Lotan et al. (2011) remarked, more work is needed to understand better how information flows among sources, highlighting that information cross linguistic barriers could exist. As results show, the persuasive power over young followers of influencing marketing can become a problem that needs further research, as far as many times a subconscious persuasion occurs. The annex corroborates this, showing related topics involving the message’s source which are growing increasingly: PERSUASION, CREDIBILITY, EXPERTISE, ENGAMENT, WORD of MOUTH, and PARASOCIAL.

Also, and linked to the message, future research seems to be directed towards the identity constructed by the influencer through his discourse. From this approach, some terms appear connected: IDENTITY, DISCOURSE, LANGUAGE, and MUSLIM. From this approach, a line of research based on the use of influencers to encourage healthy food is a promising opportunity.

Fifth, regarding industry, recent work seems to focus on sectors such as TOURISM, FOOD, WINE-ALCOHOL, or FASHION. This last industry investigated using INSTAGRAM social-network. In other words, these areas have been a good breeding ground for the proliferation of influencers in recent years.

Lastly, regarding the method, some new tools such as NETNOGRAPHY are gaining weight to investigate COMMUNITIES and INTERNET.

Limitations

In the present study, only the data from the WOS were retrieved. Some important studies from other databases or non-indexed journals might be missed, potentially limiting the completeness of the analysis. The study is based on published papers, which might suffer from publication bias. Negative results or unpublished studies may not be considered, leading to an incomplete view of the research landscape. On the other hand, this work is based on many journals, often from very remote fields/specialisations. The results would probably be completely different if the analysis were limited to only a few more related categories. Consequently, research not included in this database is outside the scope of the current study. Therefore, future studies could include additional information sources or limit the focus to a specific field of research. In addition, more keywords could be used in future studies to enrich these results. Lastly, a qualitative analysis of a subset of the papers in their dataset could be conducted in the future to get a deeper understanding of the research dynamics in the field.