SXSW Panel Report: Man Vs. Machine -- Data Science and the Future of A&R

As the music industry wrestles with how to find the next big hit, research A&R -- which uses data to identify artists' movement in the marketplace -- is rapidly gaining importance compared to traditional on-the-ground A&R. In this panel moderated by data scientist Zanab Hussain, Sony ATV Director of A&R Jacob Fain and Next Big Sound CEO Alex White sought to answer the question of which will ultimately prevail, the golden ear or the algorithm?

Fain explained that early research A&R, which he helped pioneer at Universal, meant hearing a song on the radio and calling record stores to find out if people were buying the record. "That's how we found Colbie Caillat -- she was a MySpace artist but we'd call Best Buy and they'd say all these 16 year-old girls were coming in asking for 'Bubble Girl'." Today, services like Next Big Sound, which feeds Billboard's Social 50 and Next Big Sound charts, gather artist data from many sources including Facebook, Twitter, streaming services and more to create a comprehensive picture of how many people are actually listening to and engaging with a band.

"[With traditional A&R], you would go to a show and half watch the band, half watch the audience to see how they were reacting, what the gender balance was, etc.," said White, whose company is in the process of launching a new product for A&R professionals. "What I'm interested in is the digital version of that. You can't go to 10,000 shows, you in Austin know how hard it is to get to 15. What is the trackable digital equivalent of, say, an in-person crowd surfer. The volume of data is just exploding and humans need help to make sense of all this."

White also emphasized that while qualitative methods like focus groups can be useful, there is often a big difference between what people say they're going to do and what they actually do. He described tagging along as an intern in 2005 when L.A. Reid invited 25 kids to the Island Def Jam offices and played tracks by pop artist Fefe Dobson. The excited visitors expressed huge enthusiasm for the songs, said they loved her goth look (which didn't match the music at all), and would definitely recommend the music to their friends -- but Dobson never took off. "We're interested in who actually watches the video, streams the song," said White.

When asked if data ever is at fundamental odds with gut feeling, Fain cited Owl City, whom "everyone [in A&R] hated."

"We brought it in and people didn't get it, but it was exploding off of the charts. Finally he was signed, and four months later it was a No. 1 worldwide smash," said Fain. "What A&R guy doesn't want his name on that?" On the other side, said Fain, one of his favorite signings is Nashville songwriter Mark Sibilia. "It's very rare, but he was Springsteen meets Dylan, we loved him -- there was no data, but we signed him anyway."

The panel discussed the role of social sharing in an artist's dataset, where for example users connect their Spotify listening with their Facebook feeds. "Think about 1998, or even 2006, when you'd hear about a band and you'd buy the album in the store," said White. "You'd have to play it in your car for a friend, and they'd have to remember it and then think of it the next time they were in the record store. Now you see it in your friend's feed, and it's right there in your Spotify. The friction that's reduced is orders of magnitude."

When asked if the human art of A&R would ever die as some predict, Fain said that it's important to remember that A&R isn't just finding the artists, it's managing the entire process of creating the record. But even on the artist identification front, Fain argued that "there's a human element you can't take out. I am a tremendous advocate of [Next Big Sound's A&R data product], but you can't just program this in and take the top 10 and sign them. It takes you to the 15 yard line but you really need a person and relationships to take it into the end zone."

Several audience questions focused on the degree to which A&R software levels the playing field, for good or ill. Domino Records North American Sales Director asked if all the labels invest in the same software, wouldn't there just be more math in the same footrace, where everyone was still chasing the same bands? White explained that many clients are concerned about that and ask his company not to sell to the other guy -- but the product, instead of creating one huge ranked list, allows A&R professionals to enter a wide variety of criteria that they're interested in -- so different criteria could enerate very different lists. He added that "this tool doesn't sign the band, negotiate the contract, or do any of the things that need to be done after the artists are identified."

White also highlighted that artists are starting to use the data, which levels the playing field in another way - artists can be armed with their own data when they go into a label negotiation.

Ultimately, said White, even as A&R's use of data becomes much more sophisticated and everyone gains the same tools, "we're just at the beginning of all of this. Everyone thinks that everyone else has figured it out, but remember, if you're the most experienced Facebook marketer there is, you still can't have been doing it for more than two and a half years." And, as Fain emphasized, "the human element is still very important."