Dörr (2016) conducts thirteen semi-structured interviews with natural language (NLG) service providers, which are based on structured data. His objective is to study NLG applied to journalism and thus concludes that NLG in journalism has begun to develop and expand considerably.

From a methodological point of view, the most relevant contributions are Sánchez and Sánchez’s (2017) and Túñez’s, Tou-ral’s and Cacheiro’s (2018). On the one hand, Sánchez and Sánchez (2017) use documentary analysis, direct observation, and interviews to delve into the study from Polibot on audiences’ perspectives. This is how they study the bots with con-versational applications such as Messenger or Telegram, identifying that they are perceived in a positive light based on the quality of the information and the emotional connection to it. Finally, Túñez, Toural, and Cacheiro (2018) survey profes-sional bodies, associations, and trade unions whose aim is to create a world reference map which does not only include media and agencies but also companies using news automation. The study also provides an insight into the professional perception of the phenomenon in Spain, while at the same time highlights the lack of knowledge about the influence that AI has on the profession of journalism.

Target audiences are undoubtedly one of the keys to the robotized, computerized, and computerized machinery. For that reason, we must not lose sight of the audience and the investigative process in this phenomenon. Therefore, it is also es-sential to analyze and be attentive to potential changes generated in audiences just as Napoli did (2012). An alternative is to adopt Pineda-de-Alcázar’s (2018) approach, who investigated the relationship between AI and human being’s commu-nication. Thus, from Pineda-de-Alcázar’s point of view, AI research needs to be complemented with social studies linked to ethics, semiotics, ecology and human behavioral sciences. From the research team’s perspective, the audience’s perception must not be the only one to be taken into account, but also the views of professionals of rigorous and quality journalistic information. Salaverría (2017) indicates that new forms of automated content production are emerging, which require modifications to the ways the information is produced and disseminated.

2. Methodology

Like Karlsen & Stavelin (2014), Lindén (2017), and Salazar (2018), this research uses a dual methodology. On the one hand, it uses a case study of the quantitative analysis of BeSoccer data and its comparison with other organizations in the sector; on the other hand, it uses qualitative interviews that were carried out with the organization’s founder and CEO and BeSoc-cer’s director of marketing and business development. According to Monge (2010), by using case studies we can obtain significant results and information that can not be found through quantitative methods, making it extremely valuable for decision making in companies. In any case, it is necessary to differentiate the concept of case studies as a whole from the type of case study at hand, which is for academic research. According to Yin (1994), it is clear that opinions from case studies are not statistically generalizable, nor do they represent a sample of the population; instead, they are a theoretical proposal in which researchers support the broadening or generalization of theories. As a research tool, the aim is to un-derstand how they interact and what the characteristics of a system are to be able to apply generic conclusions, as Hartley points out (1994). In short, by using case studies we can try to find correlations from a single case and apply it to others in a generalized way.