Introduction and status of the issue

The Broadband Society (Fondevila-Gascón, 2013), social media and the Internet are changing the relationship between businesses and consumers. In fact, online communication surpasses offline communication, led by social media. Instagram surpasses Facebook in the number of active users by 15%, and it is also the first choice when developing “branding” and “engagement” marketing strategies (CECABLE, 2019).

The communications sector, particularly advertising, public relations and marketing, goes to great lengths to make the most of the options provided by the Internet to move and attract the audience’s attention. If we analyse an individual that sees an ad, we can control at least three variables: attention, emotion and memory (Torreblanca et al., 2012). Attention and emotion can be measured through Sociograph. This is called registering the Electrodermal Activity (EDA).

The essential goal of this research is to measure the psychophysiological signals of attentional and emotional levels, through Sociograph, of subjects when they see Instagram Stories of real influencers that have different amounts of followers. This allows us to verify the amount of (emotional and attentional) influence generated over the follower and to determine if their activation is positive or negative, in addition to the potential activation differences according to the gender of the subject that sees them. We also suggest getting to know user consumption habits or what an “influencer” is to them to determine their perception.

Material and methods

This research, performed during 2019, follows a quantitative methodology. To collect the necessary data, we got in touch with four influencers of different types according to range of followers (IAB, 2019), gender and category (Table 1 and Table 2).

We requested a 15-second video in vertical format from the influencers, with the same communication traits as their usual character on social media. This was the way to make the result similar to an influencer marketing strategy within a conventional campaign. We obtained a minimum sample of 10 individuals for each influencer video, differentiating between the subject’s gender, so that each video was seen by at least five men and five women. Each influencer could use any kind of audio-visual tactics (visual effects, filters, audio effects, etc.), in this favourable environment. These variables were not controlled in the study.

The video’s objective was to transmit how good the promoted product was, which, in this case was an unbranded bottle of water, a neutral product, uninfluenced by feelings and of vital necessity. Before registering the EDA, we supplied a questionnaire that, together with the one filled out during the registration, was used to determine the sample’s qualitative data (type of emotion, range of influence, etc.). The data extracted from the EDA registration offers quantitative data regarding the activation level of EDL and EDR. Once the EDL and EDR signal is registered, it runs through an analogue-digital converter (PowerLab) to process the EDA through Labchart software. This application allows us to convert a graphic signal into a numerical signal to store it in Excel. This file shows columns with time, EDL, EDR and a fourth column which is manually coded by inserting the cuts written on the registration sheets. The following step is to convert the Excel files into plain text (.txt), in order to treat the signal with Electro software. Finally, this software generates the results that are viewed in output files (.res) and are transferred to statistics software (SPSS or XLSTAT). This is the end of the signal processing route and the explanation of research instruments.

The subjects’ experimentation process began by providing an initial questionnaire to provide demographic, social and behavioural data on Instagram, and continued by accessing the Social Psychology Laboratory at the University of Barcelona with the individual. Individuals were placed in front of a screen to proceed with registration. During the registration with Sociograph, several activity modules were developed for the individuals to perform. Once the Activity modules were completed, the EDA registration with the subject was concluded. We registered a total of 42 individuals. The final sample, after processing the EDA, was 37, generating a total of 187,336 records (156.11 minutes registered). Anova or analysis of variance was applied to the Sociograph results with a risk of p=5.

The five records that did not pass the EDA processing were removed from the qualitative information. The entire research sample consisted of adults (between 18 and 24 years of age). Prior experience with Instagram was required and their socioeconomic level was not taken into account. Sample members were homogenous regarding the age variable (men with an average age of 22 and women with an average age of 21). Most men had a university degree (87.50%; 44.12% of women, more with non-university secondary studies, 52.94%). There was a mismatch between men (n=8) and women (n=34) which affected and minimised conclusions regarding the gender variable. A research ethics code was applied, and participants were called individually, chosen randomly to avoid influences between them that could affect the results.

EDA was only measured once per individual for the study. Each subject viewed a different video with a distribution that included the records rejected after EDA processing. The hypotheses of the study were the following:

• H1: Men activate more emotion and attention in women in Stories marketing campaigns.

• H2: In marketing campaigns, the most-viewed elements on Instagram are Stories.

• H3: The influence of advisers on the consumer is limited.

Analysis and results

General results per type of influencer (low, medium and high level) (Figure 1) revealed significant F (2, 18731) = 1010.06 greater targeted attention (EDL) in the high level, and less in the medium level with similar levels of distraction in the data. The level of significant emotional response (EDR) F (2, 18731)=65.07 was greater in the high level and lower in the medium level. Low-level influencers have generated more heterogeneous emotional responses in comparison to the medium and high levels.

The general reactivity result according to influencer (Figure 2) showed a greater emotional distraction (EDR) for influencer Aroa, which was significant F (3, 17670)=24.90. No significant F (3, 17670)=338.41, emotional (EDR) or attentional (EDL) reactivity differences were observed in comparison to Figure 1. The global data of the signal required a more specific analysis according to activity modules.

The emotional response and targeted attention data of the different activities according to influencer level aimed to identify average differences in activity modules, mainly for viewing and task. Greater and significant F (4, 4913)=552.24 EDR averages were observed in the high level for viewing and task. Standard deviation was homogenous. Regarding attention reactivity (EDL), we observed a mild but greater average in low level during viewing, although with a more significant F (4, 11928)=8837.15 heterogeneous distraction.

Regarding EDA psychophysiological responses in activities for each influencer, we observed significant EDR differences (Aroa F (4, 48883)=367.98 and Diego F (4, 7041)=0473.91) between both low-level influencers compared to Figure 3. Influencer Diego revealed a greater average emotional reactivity during viewing than Aroa (her distraction was heterogeneous). Aroa produced greater targeted attention, which was significant F (4, 4883)=6064.71. For the viewing activity, Abi generated less attention, significantly F (4, 5985)=2474.14. Diego registered a more heterogeneous attention distraction, significantly F (4, 7041)=0473.91. For the task activity, we observed significant F (4, 4913)=2242.73 greater targeted attention in the subjects that watched Aroa and Alberto, with similar averages. Task attention distractions were significantly heterogeneous in all four influencers. Moreover, emotional reactivity was quite similar in all four influencers (heterogeneous standard deviations).

Regarding the averages and standard deviations for the viewing module per influencer according to the subject’s gender, for EDR F (7, 1304)=148.51 we detected a large emotional range in men for influencer Abi (homogeneous standard deviation). Regarding EDL F (7, 1304)=107.92, we observed greater targeting in the men that watched Alberto. Aroa generated greater distraction for emotion registration (EDR). For influencer Diego, no significant ranges were observed beyond targeted attention in men.

The viewing modules according to individual influencer aimed to observe significant differences between men and women for the same influencer. We observed a mild attention range in men (standard deviation was greater than for women). We did not observe significant differences for emotion. The high heterogeneity in men for registering emotion was noteworthy. We noted a mild significance of attention in men together with a more homogeneous standard deviation, which was heterogeneous in women.

Viewing according to gender for influencer Abi revealed the clear range of emotion in men, which was homogeneous. For women, the signal was quite similar in all ranges and no major differences were observed.

When watching Alberto, we observed a greater attention span in men. The dispersion is slightly higher (more heterogeneous) than in women. This confirmed H1 (Men activate more emotion and attention in women in Stories marketing campaigns), with an emphasis on attention.

In the task activity, there were significant ranges in terms of targeted attention. We observed significant attention averages F (7, 8952)=137.03 for influencer Aroa with male subjects (standard deviation was more homogeneous than in women). The average attention for Diego among men was one of the most significant averages, with a lower standard deviation in comparison to other influencers. The attention for influencer Alberto among women was significant, but with a heterogeneous standard deviation in comparison to the aforementioned groups. Regarding EDR F (7, 8952)=30.21, all standard deviations were quite heterogeneous, except for Diego among men, who presented a homogeneous standard deviation and a low average range. The highest EDR average was observed in Alberto for men, but his standard deviation was very heterogeneous.

The psychosocial data extracted from the questionnaires reflected the trend of Instagram users (type of content viewed or perception of the Instagrammer according to the sample). Subjects classified influencers as content creators or celebrities. There were multiple content categories. The sample’s most followed trends were related to friends, humour, music, animals and food. Regarding the most-viewed elements within the app, Stories took first place, with posts in second place and Instagram TV last. This confirmed H2 (In marketing campaigns, the most-viewed elements on Instagram are Stories). Regarding the sample’s perception of the influencers’ influence, it tended to be limited.

Regarding the data on each influencer obtained from the post-viewing questionnaire, when it came to Diego, the men who watched him would have dismissed his Stories; nevertheless, the women, including exceptions, would have gone to the influencer or product profile. 20% of women would watch more of Diego’s content, although the influencer was unknown by both genders. The emotions most aroused in the subjects were, for men, amusement and annoyance and, for women, positive emotions, although also negative emotions (frustration or boredom). The most remarkable thing regarding attention in Diego’s Story was the way he communicated and sounded (probably his accent). Diego’s level of influence over the subjects was low except in women, with slightly more impact. We observed that none of the subjects would have purchased the product.

Regarding influencer Aroa, men left the Story, whereas women timidly accessed the product profile for more information. The emotions expressed by men regarding Aroa’s video were negative; however, women expressed positive emotions (amusement, friendliness or emotion). The most striking thing about Aroa’s Story was herself and the way she communicated. Regarding Aroa’s level of influence over the subjects, we considered it limited, except for the case of women, which were more inclined to buy the product.

Regarding influencer Abi, there were no differences between men and women when it came to the amount of time spent on the platform, 50% of subjects recognised the influencer. There was a low perception of the influence of opinion leaders and ample use of Instagram. We observed that the main reaction to Abi’s video was to skip the Story, although a percentage of women clicked through to the influencer’s profile. The emotion aroused in the subjects was negative in men (associated to boredom) but positive in women, who described her as a source of amusement (66.67%) and surprise (11.11%), although also peculiarity (11.11%). The factors that most stood out about Abi’s Story were the product (50% of men, 22.22% of women) and how she communicated (half of men and 66.67% of women). Regarding the influence described by subjects regarding Abi’s Story, it was modest, except for 12.5% of women who considered it average. This limited influence diverges from the 25% of women who would have purchased the product.

Regarding influencer Alberto, we observed no remarkable differences in the years of Instagram use between men and women. He was recognised by men, although they do not follow him on Instagram. Women spent more time on Instagram, and the perception of the control influencers had was average. The sample’s reaction to Alberto’s Story for both genders was to skip it or leave it, except for a small percentage of men who would have swiped up to see more content about the product. The emotion and attention aroused in the subjects, for both men and women, reflected positive and negative feelings. Men were split 50/50 between amusement and indifference, while women, despite the prevalence of boredom (44.44%), also thought of friendliness or amusement, without forgetting about indifference. What stood out the most to both genders were Alberto’s communication style and (in the case of women) his character and the product. Alberto’s range of influence over subjects was low for both genders, although half of the men and 25% of the women stated that they would have perhaps consumed the promoted product. Based on the results, we confirm that the most-viewed marketing campaigns on Instagram are Stories, that the influence of advisers on the consumer is limited and that the degree of influence generated over followers is proportional to the activation of emotion and attention. Therefore, H3 (The influence of advisers on the consumer is limited) was confirmed.

Discussion and conclusions

The research reflects the relevance for businesses to analyse attention and emotion in the use of marketing campaigns on social media, Instagram in this case. We confirm that the most-viewed elements on Instagram are Stories, in line with Brandmanic (2018). The most followed content is created by friends or about humour, music or fashion. The sample confirms that an “influencer” is a celebrity or a content creator, in contrast with what Campillo (2016) states, and the influence of advisers over platform users is considered limited.

Regarding the levels of emotion and attention activation, they are different according to the influencer’s range of influence. The more influence generated over the individual, the greater emotion and attention activation observed. However, the lower the influence, less emotion and attention activation. This confirms the influencer marketing strategy (Augure, 2018; Alonso et al., 2018; Santamaría-de-la-Piedra & Meana, 2018), its impact in terms of advertising (Fondevila-Gascón et al., 2015; Tapia-Frade & Martín-Guerra, 2017), and sector (Jaromír et al., 2017), and the effectiveness of EDA (Aiger et al., 2013) as measurement technology.

So, we observed activation differences according to the influencer. Attentional activation was higher in influencer Aroa (low level) and emotional activation, in influencer Alberto (high level). We observed differences according to the subject’s gender. Men revealed greater emotion and attention activation, especially regarding attention, with all the circumspection of an extendable sample of individuals and a scalable experiment.

We detected differences according to influencer and the subject’s gender. So, arranged from lesser to greater emotional activation registered, influencer Aroa did not manage to generate emotion in the subject, although she did attract the attention of both genders. This activation became negative emotions, dismissal of the Story, lack of interest in the act of purchasing the product or, in men, viewing more of her content. Nevertheless, a small percentage of women sympathised with the content. Influencer Diego did not manage to arouse emotion, and only a higher level of attention in men. Women interacted with his Story and spoke more positively about it. No purchases or interest in his content were achieved from either gender. Influencer Abi raised strong emotions in men and less in women, and he/she did not manage to attract the subject’s attention. Women showed more interest in her, expressed positive emotions and made the purchase. In men, although activation was greater, it did not convert into a purchase. Influencer Alberto generated emotion in both genders and attention only in men. Activation translated into positive emotions and produced a likely potential purchase in both genders. Therefore, we observed fluctuating ranges of attention and emotion.

The application of the registration with Sociograph to describe the nature of emotion and attention activation (EDA) in Instagram Stories presented various research limitations. Thus, the sample was limited, and male participation could be improved. Likewise, the number of influencers could also be optimised. Other variables could be taken into account, such as sound effects, clothing or visual effects in the influencers’ videos, factors that can affect emotion and attention and which have been standardised for this study. Therefore, delving into the videos’ visual standards and obtaining a larger and more varied sample of influencers are guidelines to follow in future studies.

In this sense, the influence and emotional load generated by social media content on the population suggests different types of research: sectoral, according to type of campaign, on different social media (even comparing them synchronously), on different media outlets, such as interactive HbbTV (Fondevila-Gascón et al., 2015), at different ages and different moments of the customer’s journey. The use of other tools associated to neuromarketing (EEG or Eye tracker) can help to define the source of emotion and attention, fundamental for future decisions that actors can take within the field of advertising, especially interactively (HbbTV), where the emotion-attention pairing is triggered by the dialogue between the audience and brands. Establishing categories and types of emotional and attentional impact through ads is both a stimulating and monetizable challenge for companies. 1