Journal of Marketing Research Scholarly Insights are produced in partnership with the AMA Doctoral Students SIG – a shared interest network for Marketing PhD students across the world.
Who wouldn’t agree that music connects with your soul? Knitting together the past, present, and future of eclectic human communities into one fabric called humanity, music has been a common link— economically, socially, and technologically—between different aspects of society throughout history. According to a report by BBC music correspondent Mark Savage, the release of albums by well-known artists like Taylor Swift, Adele, Beyoncé, and BTS caused the year 2021 to see the fastest revenue growth rate in the global music market since the 1990s. Accounting for 65% of the cumulative revenues, music streaming is becoming the new engine in driving the popularity of music recordings.
Although the music industry is growing at an unstoppable pace, so is the rate of failure. While the monetary figures suggest the lucrative profitability of investing in the music industry, experts warn that these profits are not homogeneously distributed. In the same BBC report, Paulo Pacifico, CEO of the Association of Independent Music said, “It will remain very difficult to reconcile the positive numbers in the market as a whole with the hard reality of making a life in music by an individual.” In other words, it is only the rarest of the rare musicians who are fortunate enough to succeed in the industry. This leads us to the question: What drives the commercial success of music?
Ask any music lover and they will tell you that a playlist or album is not merely a list of songs; rather, it is a meaningful curation of music that depicts a theme, tells a story, or sets a mood for the listener. Since the success of creative products such as albums depends on consumers’ felt experience, it is often difficult to predict beforehand.
However, in a recent Journal of Marketing Research study, Khaled Boughanmi and Asim Ansari have developed a novel machine-learning framework that uses acoustic features of the song (e.g., levels of energy, liveness, danceability), metadata, and user-generated textual data to predict the success of music albums. Their model helps in forecasting success and in recommending and designing albums and playlists. They have demonstrated the superiority of their framework by using it to predict the success of musical products over the past five decades. The findings of this research highlight the significance of different acoustic features in album success and yield key insights into the evolution of popular music. The article also provides a vivid summary of listeners’ experiences and perceptions of different musical genres and sub-genres.
The findings of this research highlight the significance of different acoustic features in album success and yield key insights into the evolution of popular music.
We had a chance to contact the authors to learn more about their study and gain additional insights. Have a read of the following Q&A with the authors to quench your curiosity further:
Q: What was your motivation for pursuing the study on experiential determinants of musical success?
A: We explored this area for two main reasons. First, this is an area of personal interest. As music lovers, we wanted to study a field for which we have a genuine and personal connection. Second, there was little work in marketing around preferences for music from a content perspective. This was due principally to the lack of data and unavailability of the statistical and mathematical tools that would make such analysis possible. Our ability to scrape data from multiple online sources that are currently available, combined with our knowledge of appropriate methodology, provided us with the perfect opportunity to work on music dynamics.
Q: What challenges did you face while conducting this research?
A: We faced three main challenges. The first challenge was data collection. As we planned to study the multifaceted nature of music, we needed acoustical data as well user-generated descriptors to study the determinants of musical success. The second challenge was methodological. We had data of different modalities and needed to develop a methodology capable of fusing different types of data. Finally, we needed to incorporate dynamics in our modeling framework. Since we took a semiparametric route, this task was not straightforward.
Q: While several marketing applications of the proposed model have been showcased in the article, it would be great to know which other realms you believe could benefit from the findings of the study.
A: While our methodology was applied to a music context, practitioners and academics in several other industries could leverage our proposed approach. For instance, in any setting where structured and unstructured data on products are available, researchers could derive insights by modeling the inter-temporal evolution of success factors, as we do. In addition, several aspects of our methodology are novel, and researchers can use the different components of our framework in their modeling efforts.
Q: Conducting such a study must have required sound knowledge of music and the music industry (for instance, the intricate use of musical notes, pitch, tempo, etc.), which is a matter of subject expertise. Therefore, do you possess any background in music?
A: We do not possess an academic background in music. However, we acquired necessary knowledge to study the topics. This was easier because the article was part of Khaled’s dissertation, which necessarily requires a deep dive into a topic. This was done through reading relevant papers published in music journals and books that detailed the evolution of music in the United States. Also, discussions with colleagues that had musical expertise helped us in shaping our research.
Q: What is your advice to scholars who would like to extend this research to their work?
A: We encourage other researchers to work in the consumption of experiential goods. New technologies are allowing us to model experiential goods like music, games, and art using machine learning and deep learning-based representations.
These areas are understudied in quantitative marketing and offer new and interesting avenues for research. We would encourage our peers to bring new data and new types of modeling to the marketing field. Studying these areas will also usher in new insights about consumer behavior in experiential domains.
Read the full article:
Khaled Boughanmi and Asim Ansari (2021), “Dynamics of Musical Success: A Machine Learning Approach for Multimedia Data Fusion,” Journal of Marketing Research, 58 (6), 1034–57. doi:10.1177/00222437211016495