Discovering new music talent like James Bay, this year’s tipped breakthrough act whose debut album duly topped the charts, is an expensive business for record companies.
Yet investment in A&R (artists and repertoire), the talent-spotting department that is the lifeblood of any music company, has been reined in as a result of the calamitous decline in recorded music sales over the past decade. The days of signing 50 acts a year and seeing what sticks are over.
Identifying and developing talent is becoming a science – much to the chagrin of the traditional A&R man relying only on instinct and “good ears” – and the tool that is increasingly being used to define a hit or a miss is “big data”.
Instead of trawling the back rooms of the country’s pubs for the next Ed Sheeran, what if “moneyball” – the analytical, statistics-based approach successfully used to assemble a competitive sports team – could be adapted to pop music?
Moneyball was immortalised in the eponymous 2011 film that showed how manager Billy Beane used the strategy to transform the Oakland Athletics baseball team, in his case by identifying undervalued players who could be turned into assets.
Now record companies have seized on a University of Westminster MA thesis by Prithwijit Mukerji, titled Moneyballing Music: Using Big Data to Give Consumers What They Really Want and Enhance A&R Practices at Major Record Labels.
“Moneyball is now common among baseball teams throughout the US. I thought that if it works there, why not apply it to music?” Mukerji told a debate held by Westminster University’s Music Tank “knowledge hub”.
The process began when researchers from the School of Electrical Engineering at Tel Aviv University created an algorithm that predicted the rise of artists Soulja Boy and Sean Kingston in 2007, two months before they reached the top of the Billboard charts.
The algorithm used data from Gnutella, a peer-to-peer file-sharing network, and found that the hit potential of a song depended on the artist’s level of success clustered in a geographical area, as well as the speed at which that success grew in that area.
Sam Lowe, A&R at the independent dance music label Ministry of Sound, told Mukerji that Twitter traffic is now a hit-spotting tool. Twitter confirmed the popularity of a “type of organ bass-line sound originally used in drum ’n’ bass and garage music”, which Ministry of Sound was able to capitalise on as it became widely sampled on house music tracks.
The technology is becoming more sophisticated. Mukerji cited Spotify’s purchase of The Echo Nest, a company that collected 1 trillion points of data last year. It used the information to create personalised music-recommendation algorithms suggesting which songs listeners would enjoy depending on their mood and the time they were listening.
Warner Music Group is partnering with Shazam, which has 100m monthly active users, who together “tag” 17m songs, TV shows and ads daily on their cellphones. Shazam tags alerted Polydor Records that the song Video Games, first posted online by Lana Del Rey in 2011, was gathering a huge response. Jamie Spinks, A&R scout at Polydor, told the Money-balling Music report that the Del Rey success “happened without anyone really knowing. That’s the sort of thing we’re looking for”.
The Polydor A&R team now holds a weekly “Shazam meeting” to spot emerging break-throughs before the competition.
Analytics company Next Big Sound publishes a “dashboard, enabling users to analyse artist popularity by tracking weekly changes in Facebook page and Instagram likes, Twitter followers and mentions, YouTube and Vevo video views, SoundCloud plays and even Wikipedia page views”, Mukerji wrote.
Next Big Sound claims to predict album sales to within 20 percent of the actual figure for 85 percent of artists.
Shazam tags, Instagram likes and Facebook popularity are now used by the Radio 1 playlist committee to help decide whether a song is working for its teen-focused audience. The station insists that data does not override the passions of producers and DJs, however.
Mukerji admits that data “may not be able to predict the next Sex Pistols or Adele, but it can help A&R scouts and marketers notice trends earlier than others, giving them a commercial advantage over competitors. Music is yet to moneyball, but it is less a tipping point and more a continuous process.”
He added: “Forbes recently valued A&R as a $4.5 billion industry. We have to make money and we do bear risk. What big data can do is help to minimise that risk.”
However, Infectious Music founder Korda Marshall, who has signed Take That and alt-J during his career, warned that “50 to 75 percent of music’s greatest moments wouldn’t have occurred” if they were based on data. “I’ve made £1m decisions on whether the hairs on my arm stand on end,” he said.
Following data, not gut instinct “leads to boring, not innovative, music…”, he warned.
If a record company were to feed the “Sound of 2015” into a big-data cruncher – a pinch of Ed Sheeran, some George Ezra-style strumming, Sam Smith’s soulfulness and a splash of Johnny Depp’s looks – the result would probably look and sound a lot like James Bay. – The Independent