- theguardian.com, Wednesday 9 April 2014
The internet is taking power away from record labels but it’s also giving them the ability to predict future hits
Streaming sites and social media platforms have given record labels and other music industry stakeholders access to vast amounts of data on listening habits and trends. Photograph: Christian Hartmann/Reuters
The world’s listening habits were once relatively private.
Music lovers declared their allegiances with posters, festival wristbands and physical copies of the music they loved. But when listeners retreated to the safety of their bedrooms, they could indulge in guilty pleasures without Last.fm or Spotify telling the rest of the planet.
Record companies were aware which radio station played their songs and where their CDs were popular, but that information painted an incomplete picture at best. Who knew what music people were sharing on tapes and CDs burnt in the privacy of their own bedrooms?
“The traditional metrics like sales told us a record or CD was sold, but nothing about what happened after that,” says Paul Smernicki, director of digital at Universal Music UK.
That’s all changed. The explosion of data from sources like torrenting, music streaming sites and social media platforms has offered the music industry a huge opportunity to understand their fans and spot upcoming artists like never before. Music analytics is now worth an estimated £1.8 billion per year.
At the same time that the internet is taking power away from record labels, it is also giving them the ability to predict future hits.
The artist as a machining tool in a factory
Music lore is full of talent scouts whose refined ear led them to find ‘the next big thing’ before anyone had ever heard of them.
How much of this is myth-making is debatable – inevitably there are fewer stories about the scores of artists who flopped – but music analytics is turning the art of finding the right artist into more of a science.
After tracking thousands of artists from nothing to global success over the past five years, music analytics firm MusicMetric is beginning to be able to spot the signs that distinguish an also-ran from a gold medalist.
Today, almost any artist that comes to the attention of a music talent scout is going to be generating some buzz. They’ll have a followers on Twitter, likes on Facebook and maybe sales via BandCamp. MusicMetric say they’re increasingly able to tell who is buzzing in the ‘right way’.
“It’s no different to a sensor in a factory that’s detecting vibrations on a machining piece and when the vibrations start vibrating in a particular way they can detect that it’s going to fail,” says Gregory Mead, chief executive of Semetric, the company behind MusicMetric.
Translated to the music industry, when an artist starts buzzing in a certain way, they can detect that it’s going to go big. In one study for an unnamed client, Mead says they were able to predict with 90% accuracy who would be number one three months in the future.
Making these predictions, however, requires coming to terms with a flood of information across scores of different websites, platforms and services.
Multiple floods of information
Any one artist today might have several albums with dozens of singles and music videos spread across multiple platforms.
To understand the popularity of just a single song, you need to track official and unofficial uploads on YouTube, plays on the ever expanding list of music streaming sites, physical and digital purchases on Google Play, iTunes and Amazon, and then of course there’s the controversial issue of music piracy.
“For a big artist like Katy Perry, there’s about 19,000 different signals we have just for that artist. Each video, in each territory, on each platform, each release and each song and so on. No-one’s going to be able to look at that [by hand],” says Mead.
Heavy lifting aside, the challenge is to turn those signals into useful information a manager can use when deciding where to take their artist on tour, or that a talent scout can use when deciding which recommendations are worth following up.
MusicMetric isn’t the only company crunching the data. In March, Spotify announced it was acquiring The Echo Nest, a music analytics company in the US, for a rumoured $100 million (£59,718,616) in order to beef up its music discovery services while competitor Next Big Sound was dubbed “Moneyball for Music” by Forbes. In just a short few years, the music industry has woken up to the need to examine social data more rigorously.
“When we started there were basically no companies doing this kind of analytics in the industry,” notes Mead.
Indeed when Mead first began working on MusicMetric in 2007 while at Imperial College London, he was prototyping the technology on a single server tucked away in the back office of the student newspaper, where he had previously been a music editor. Now they use an ever-expanding farm of servers, have an office in Los Angeles as well as in Old Street and count the university’s technology spin-out company, Imperial Innovations, as an investor.
But now that it’s here and being embraced by artists and record labels alike, does the ability to predict who is going to be a superstar based simply on a graph, the Football Manager-ificiation of music, on a computer screen fundamentally change the artistic process?
“It doesn’t tell you what music to make,” replies Mead. “It doesn’t say this is the song you should make. It just tells you a little bit in advance that it’s going to be popular.”
In fact he suggests that it allows for greater diversity, by making it easier for record labels to recognise the existence of niche audiences and fan bases: “It opens up opportunities for artists to help promote themselves, rather than restrict everyone to conforming.”
Universal Music UK’s director of digital, Paul Smernicki too thinks that ultimately the music industry will always be focused on content, no matter what analytical tools are available: “I think the mystique [of the music industry] is around artists, and that’s under their control.”
He’s sober about the impact that music analytics tools are having on the industry: “It’s important to remember that it’s just a set of tools to help inform us. The data doesn’t make the decisions, that’s an un-replicable part of what we do.”
But with robots replacing humans in every sphere of work, who knows how long it will be before the most successful talent scout in the industry is an algorithm.