Datascience in the music industry

Decisions in the music industry are often based on gut feeling. Organisers book a band they personally like in the hope that the audience shares their opinion. managers give record deals to bands they believe will break through. Festival organisers have to make decisions about catering, security, staff, logistics and so on. Unfortunately, gut feeling often falls short. In the music sector, one misstep is often enough to kill an SME; a disappointing festival edition quickly leaves a deep pit of debt. Therefore, many players in our sector would benefit from an application that supports their decisions based on objective data. Fortunately, times have changed. The rise of social media, streaming services and technological advances in data collection and processing have radically increased the number of data sources to base decisions on. In other sectors such as automotive and telco, we see a large number of applications based on the newly available data collection and processing techniques. The music sector in Flanders for the time being fails to integrate this new methodolgy into their daily operations. A survey we conducted showed (n=30) that 90% of respondents from the sector recognised the importance of data to support decisions. Only 15% said they had the necessary knowledge, time or resources to apply this technology. This creates a huge untapped potential within our music sector. In a PWO trajectory, the research group PXL-Music Research joined forces with Ancienne Belgique and Clubcircuit to develop a proof of concept that allows organisers to start working with their data themselves. Thanks to qualitative focus discussions with the partners combined with applied data science research, we developed a methodology to let organisers make more objective decisions. The music sector in Flanders does not have a tradition of scientific research. With this Tetra project, we want to take a first step in connecting science and practice based on the research results of the PWO project. In this way, we hope to strengthen this interaction between science and the sector in the future.

Decisions in the music industry are often based on gut feeling. Organizers book a band they personally like in the hope that the public shares their opinion. A&R managers give record deals to bands they believe will break through. Festival organizers have to make decisions about catering, security, personnel, logistics and so on. Unfortunately, gut instinct often falls short. In the music industry, one misstep is often enough to kill an SME; a disappointing festival edition quickly leaves a deep pit of debt. Therefore, many players in the music industry would benefit from an application that supports their decisions based on objective data. Thanks to the rise of social media, streaming services and technological advances in data collection and processing, the number of data sources on which to base decisions has increased radically. The music industry in Flanders is not, for the time being, succeeding in integrating this new methodology into their daily operations. A survey we carried out (n=30) showed that 90% of the respondents from the sector understood the importance, but only 15% said they had the necessary knowledge, time or resources to apply data science.

Within this TETRA project, PXL-Music Research wants to help the music sector take those first steps. We identified 5 pillars that we believed could help the sector: data collection, processing, analysis, visualization, analysis and prediction.

We convinced stakeholders from the music sector to join forces: from non-profit organizations like Publiq, mediarte, VI.BE to management associations PlayRight and Sabam, record companies Universal Music and News, Ultratop, Clubcircuit (organization of 14 Flemish concert halls like AB, Het Depot, MOD, etc.) and artists were represented by Greenhouse Talent and Gentle Management.

Back in 2017, we started collecting publicly available data and M.I.D.G.E. was developed. Our music intelligence data gathering engine has been happily foaming the worldwide web for artist data (e.g. metadata, social media handles, data from streaming platforms and radio airplay) ever since. The starting point was MusicBrainz, an open music encyclopedia, where artists have a unique ID.

Gradually, however, we discovered that lesser-known and/or local artists were missing, an issue that similar (international, commercial) services such as Chartmetric and Soundcharts also experience. What starts moving quietly in Flanders doesn't show up on their radars. Enter VI.BE, also user-generated, but the platform where local artists register at an early stage in order to participate in talent shows such as Humo's Rockrally or Studio Brussel's Nieuwe Lichting.

Besides this more granular approach, this project distinguishes itself from other initiatives in bringing together the publicly available data and the private data of our partners. The closed data is obviously only fed back to the partners, while the insights and analyses - either anonymous and aggregated - can be used by the entire sector.

The results are unlocked to the sector in 3 ways:

The data collected is made available via API. This is not full-blown with the latest specifications, but in the same spirit of approachable and accessible open source software (R, Python). Using paid software was deliberately avoided to ensure easy and economical implementation.
Throughout the project, in an iterative and demand-driven process with the above-mentioned partners, we developed several proof-of-concepts. These were bundled as modules in 2 web applications:
On the one hand A.D.A.M. visualizing the publicly available data and on the other V.I.P. reserved for partners who granted us access to their ticketing data.
This project was largely synchronized with the corona crisis that further exacerbated the challenges of the music industry. After 2 years of forced closures and temporary unemployment, we believe even more strongly in the future of data science in the music industry.
After all, necessity is an underestimated stimulus for innovation. Or, never waste a good crisis.