Editor's Note: Tommaso Rocchi is a 2020 Master of Arts graduate of The Global Entertainment and Music Business program at Berklee College of Music in Valencia, Spain. As a former college radio Music Director at RadioBue.it, Rocchi focused on copyright law, new business models, and data analytics. You can listen to our podcast interview with Tommaso about his article below.
Data-driven A&R has been a buzzword for quite some time in the music industry, but also one of its most guarded secrets. Even before the acquisition of Sodatone by Warner Music Group, major and big indie record labels started to switch their mindsets and focus on the advantages of a data-driven approach.
Compared to the classic "gut-feeling" expertise of a senior A&R, data analysis allows today’s A&Rs to validate their intuition and justify talent acquisition with predictive modeling. According to the IFPI Global Music Report of 2019, record companies are investing more than one-third of their global revenues ($5.8 billion) in A&R and marketing each year. With such a significant share of recording company budgets invested into artists, their main objective is to minimize revenue loss, and data can help them make the right bets. The second reason why the industry is going full-data is volume: As stated by Spotify itself, nearly 40K tracks are uploaded on the DSP every day. With such an impressive number of records, it is impossible to use only a “gut-feeling” approach.
Due to these factors, record labels are extremely careful about how much they share regarding the technology and methods they are utilizing. Saying too much could mean losing a competitive advantage over another label and potentially arriving late to a signing.
New technological advancements have already changed the role of A&R in the music industry today, transforming the relationships between labels, artists, and managers. Still, data-driven A&R is largely a secret kept in plain sight, and very few people have a clear understanding of what A&R is and how it works in the music industry today.
What Is A&R in the Digital Age?
A&R, or Artists and Repertoire, really came into its own with the development of the recording industry. As the commercialization of the phonograph ramped up during the mid-20th century, the music industry was splitting from two sectors into three sectors: publishing, performing, and now, recording.
The technological disruption of recorded music at scale expanded the artist pool, because it opened up a whole other world of song interpretation and, consequently, consumption. The recording artist was born, and A&Rs were positioned to help them succeed.
Originally, the A&R role was tasked with matching artists to the right song and/or songwriter. Before the Beatles, after all, commercially viable artists who wrote their own material were pretty much unheard of.
As the album format reached full maturity with ‘60s/‘70s Rock, A&Rs were increasingly involved, in varying degrees, with every aspect of an artist’s development, from discovery to production, creative direction to career trajectory.
A&Rs weren’t just song pickers or talent scouts: They were the business-minded angels on artists’ shoulders, the bridge between the artist’s wildest creative impulses and the record label’s most stringent budgetary concerns.
A&R has changed since the dawn of the internet: Gone are the days when an A&R scout had to step inside a dark and smelly pub in South London and be the first to find the new Ed Sheeran. Even if this idea still has a lot of charm, these signings are extremely rare today. The job of an A&R scout is extraordinarily analytical and requires having a great understanding of the musical environment. Paul Samuels, Vice President of A&R at Atlantic Records UK, clearly remembers how hard it was to research what records “the cool kids were listening to”:
I am old enough to remember record shops and finding out what records were selling. My first gig with Craig Kallman [now CEO of Atlantic Records] was going around to record shops with records and trading them for independent ones. And that was my first research. We would go to all these shops and find out what people were listening to. So I'd go to Rough Trade Records in London on a Saturday and see what was happening. But most of the time, it was a bust, there was nothing.
The use of data and automation has certainly helped ease this process, and labels are investing many resources in this field as a result. If we look at this recent job posting by Motown Records, we can see a very different kind of professional figure than the usual A&R job description.
One profile that major labels are looking for today is extremely technical, involving extensive knowledge of SQL, programming languages such as Python or R, a background in statistics, and the ability to create predictive models. In this particular profile, there is no mention of any skill related to assessing talent, evaluating production and song structure, or vetting an artist’s branding. As stated by Samuels himself, it can be advantageous to have A&Rs make judgements based on data and not on their preconditioned tastes.
But data can’t solve every A&R problem. While we’ve become extremely good at creating algorithms that can outperform humans on very specific tasks, the job of an A&R involves so many different skills that it’s almost impossible for a machine to replace an A&R. First of all, numbers can lie and aren’t necessarily a universal sign of quality. An artist might just get a lucky playlist placement or strike the right viral chord on TikTok. As a result, successful signings that are purely data-driven are extremely rare.
And then there are the outliers. The fact that some artists don't have very big numbers at the beginning of their career doesn't necessarily mean that they don’t have potential. They could be doing something completely different from other musicians, and, most of the time, the artists who don't conform to trends are destined to become the most successful.
Another big problem of only utilizing data is that it gives no competitive advantage. Using the words of Jerry Zhang, Co-Founder of Sodatone, major labels “are looking for the 10 artists in a million that have abnormally good performances on different platforms.”
Unfortunately, abnormally good performances on different platforms are all the more visible. Once you see those unambiguous signals of success, other labels that have access to the same dataset have likely already spotted those signals and leveraged their competitive advantage.
So, if a record label’s objective is to arrive before everyone else, how is that even possible when everyone is looking at the same signals? What are the other factors that make a record or an artist stand out, and which of these factors can be assessed in a predictive capacity?