HOW A&R'S HAVE CHANGED THEIR APPROACH IN SCOUTING FOR NEW TALENT

How A&R's are scouting for new talent... So what has changed in their approach?

Most of the labels have some sort of an Artist & Repertoire department. The spectrum of activity may vary - it can be just scouting for talent to sign or working on everything from branding to building a creative team around an artist. 

The A&R work has changed substantially with the emergence of digital technologies. The increasing use of data in business is changing the role and workflow of A&R in today's music industry.

A data-driven A&R process has long been a buzzword in the music industry. At the same time, the details of how each company implements the use of data and the benefits of a data-driven approach have never become public.

Compared to the classic "intuitive" expertise of an A&R specialist, which was the last filter of artists who get signed, data analysis allows today's A&R to test their intuition and justify the attraction of certain artists using predictive modeling. According to the 2019 IFPI Global Music Report, record labels invest more than one-third of their total revenue ($5.8 billion) in A&R and marketing each year. The bottom line is that with such budgets invested in artists, their main goal is to minimize revenue loss, and data can help the labels take the right steps in that direction.

The second reason the industry is shifting to full data use is the scale of music production: there are thousands of tracks uploaded to digital music platforms every day. With such an impressive number of records, it is impossible to use only intuition. Music production became affordable - most of the first records are born in bedrooms and garages without the involvement of label’s investments.

Because of these factors, record labels are extremely careful about what they publicize about the technologies and methods they use. To say too much is to lose a competitive edge over another label and possibly miss out on some important deal.

New advances in technology have already changed the role of A&R in the music industry, changing the relationship between labels, artists, and managers. However, the A&R process based on the use of data, as well as the specific methods and technologies used, are kept secret, although everyone knows that such a thing exists. The bottom line is that very few professionals have a clear understanding of what A&R is and how it works in the music industry today.

One of the interesting shifts in the recording industry is that licensing deals emerged as an alternative to classical artists deals. Labels can just license the existing catalog of some artist, promote it, and profit from it. Such a phenomenon is one of the reason of the increasing importance of data in the work of A&R.

The use of data does not solve all the existing A&R problems. While we are good at designing algorithms that can outperform humans on some specific tasks, the work of A&R is so varied that a machine is extremely unlikely to replace A&R. First, the numbers can lie and do not necessarily indicate the quality of the repertoire. Any artist might just get a lucky placement on a playlist or accidentally go viral on TikTok. As a result, successful record deals based solely on data are extremely rare.
Besides, there are statistical emissions (outliers). Just because some artists aren't doing well early in their careers doesn't necessarily mean they don't have potential. They can do something completely different from other musicians, and, in most cases, the artists who don't follow the trends succeed.

For better or worse, modern music analytical tools are heavily focused on audience analysis. Each like, repost, or comment can indicate that the musician is resonating with a potential audience. However, it doesn't give any idea of how artistic the musician is, for example.

Three key factors underlie either the success or failure of a new musician. The song is probably the most complex factor due to the large number of variables. However, the algorithms responsible for classification are getting better and better at identifying characteristics such as the general mood of a song and its genre. Tools like Music Xray can analyze the structure of songs and offer solutions for its improvement, while platforms like Hit Song Deconstructed provide a thorough level of songwriting analysis.

Theoretically, a computer can assess the potential of a song. However, lyrics and personal aspects of performance are still quite difficult to digitize, especially when elements such as metaphor and figurative expression are taken into account.

There is still the problem of defining the "success factor" of a popular artist. Elements such as charisma, stage presence, and artistry cannot be interpreted by a machine. The way an artist moves across the stage, the way an artist interacts with the audience, even the timbre of the voice - all of these are characteristics inherent in the individual and cannot be defined as "good" or "bad".

Even if we find individual elements that can technically be assessed using AI, the challenge is to assess the artist with the whole complex of these elements in a holistic way, as well as to include an element of subjectivity. Until we have even more advanced AI, humans will have a distinct advantage over machines, and the field of A&R will likely be based on a combination of data-driven decision making and traditional scouting.