In the Minds of Internet Users TPMP: The Power of Digital Media Engagement

Statistical analysis and sentiment analysis of the engagement generated by 1362 TPMP publications on Twitter, Facebook and Instagram.

Since its creation, Touche Pas à Mon Poste (TPMP) has captivated a large audience thanks to an active presence both on television and on social networks . This study of their digital strategy offers a visual of the engagement generated on Facebook, Instagram and Twitter for a month of publication. It reveals an overview of trends in Internet users' behavior and highlights the effective strategies deployed by Bolloré's media.

Power BI Dashboard

The TPMP show strategically uses the specific features of social networks to promote its content and cultivate an active virtual community. The analysis of publications shows sustained activity, especially on Twitter (791 publications/month), where a diversity of content (photos, videos, texts, links, retweets, polls, gifs) enriches the overall engagement, with in addition a live communication. Facebook (469 publications/month) is used for longer formats without editing and contests are organized on Instagram (102 publications/month).

The analysis of political publications reveals that they generate particularly high engagement despite their rarer presence, demonstrating the electrifying effect of political debates on the audience. On the other hand, the correlation matrix between engagement and political or non-political content is not verified. To deepen this analysis, other tests could be carried out, including regression analysis to quantify the relationship between content type and engagement, and Pearson or Spearman correlation tests to measure the strength of this relationship. In addition, an analysis of variance (ANOVA) could compare engagement means between content categories, and chi-square or Student tests could determine significant differences. Finally, a clustering approach would allow identifying natural groups of publications with similar engagement behaviors, thus providing a more detailed and nuanced view of the impact of political content on engagement.

Examining the emotions expressed in popular comments shows a polarization of opinions , but with overall neutral sentiments , highlighting that we could find a complexity of emotional engagement. On the other hand, this result is calculated with VADER (Valence Aware Dictionary and sEntiment Reasoner) and TextBlob, Python libraries used for natural language processing (NLP). Both give similar results and the dashboard presents the results of the TextBlob method. These libraries are mainly designed to analyze texts in English, which can pose several inaccuracies.

Semantic analysis reveals the use of striking and sensational words such as " upsetting ", " violent ", " shock ", " tense " in the publications. At the same time, the show uses terms such as " revelations ", " politics ", " testimonies ", " debate ", " affair ", " justice ", highlighting its potential desire to appear as a serious media outlet . TPMP could thus influence the construction of Internet users' perceptions and beliefs, including in their political orientations. The dichotomous staging and agenda setting reveal a marked preference for themes such as violence, discrimination, governance and harassment. It is essential to note the short period of one month of content analysis, limiting the generalization of the results. However, previous research, notably that of Claire Sécail and Stéphane Encel, seems to corroborate these observations.

Through this digital exploration we get a glimpse into the workings of a successful engagement strategy. TPMP seems to demonstrate that a thoughtful and diversified presence on social networks, coupled with a keen understanding of emotional and political dynamics, can generate unprecedented enthusiasm.