The week of March 15, 2015

Big data will find your next favorite band

By Dylan Love

Your next favorite band won’t be signed for their explosive live show but for their YouTube views. Their big break won’t come from a copy of their cassette debut but from the tracking of their weekly SoundCloud spins. It’s not the underground buzz that will matter but the quantifiable chatter taking place on Twitter and Tumblr.

This isn’t a scene from the future. It’s happening right now. Big data is radically transforming the music industry, replacing the traditional role of A&R with complicated algorithms that will help labels not only decide which artists are worth signing but what songs should be released as singles and what cities those bands should hit on tour.

That’s not to say that live shows don’t matter. There’s just been a changed sense of what is “significant” in the business of music. A concert on a weekday evening might reach a couple hundred people, but when a well-followed Twitter personality tweets about a band, word could reach tens of thousands. An album release party might win fans in the double or even triple digits, but as soon as that album hits Spotify, it has a potential audience of millions to win over.

Our online interactions with music are interesting to the music industry because each interaction yields valuable data about who we are and what we like. On an individual basis, these bits are pretty worthless and disparate, but in the aggregate, they converge to tell a rich story about changes in taste and musical tide. The only problem is how to best organize and make sense of that data, so that artists, A&R companies, and all order of music industry professionals can make actionable (and hopefully lucrative) plans for the future.

Big data called the radio star

The radio used to matter. A dominant medium for numerous generations following its invention, the radio once represented the convenient magic of everything. It served as a means of personal communication, it publicly spread word of current events around the world, and, once upon a time, it was even the arbiter of musical cool.

But truth of truths: The up-and-coming musical artists of today who see airplay rotations as the route to upstart musical success don’t have a clue. For every radio play tallied by a conventional mainstream star, there are dozens of bands trying to get out of the garage and in front of people in order to carve out their own idea of commercially viable artistic triumph. There’s not enough room on the radio for everyone—literally only so much electromagnetic spectrum in the air—but this is a problem that the Internet solves, democratizing distribution so effectively as to embarrass the radio devotees. It’s so easy to make your music instantly available to the entire Internet-connected world; all you really need is the desire to do so and a SoundCloud account.

An artist’s radio play statistics have gone from being a key set of numbers for gauging commercial success to being just another bunch of numbers in a rich variety of data sets. (Better to have the most popular song at one Milwaukee radio station or to be trending worldwide on Spotify? Exactly.) Getting an individual’s email address alone can spell out big monetary dividends over time: An Arcade Fire fan email address is worth an average of $6.26 over the lifetime of the band, and Sigur Ros’s fans are worth $10.91 per email address.

Big data is radically transforming the music industry, replacing the traditional role of A&R with complicated algorithms.

Software-as-DJ streaming service Pandora has tried to use its trove of user data as a means of luring more and more artists to its platform: It means more content for Pandora to sell ads against (or subscriptions for), and bands get to learn a little bit more about who likes them and what songs people are responding to.

Iron Maiden reportedly planned a tour based on where its albums were most heavily pirated. The story turned out to be false, but the reasoning is sound as ever. Bands have always used to data to plot their tours: Traditionally, artists looked at Soundscan numbers to see where their albums were being sold and at what volume—data that was then used to line up shows across the country. As the number of albums has steadily declined, that data is being seriously upgraded. Bands can mine their Facebook data to see where they have hubs of fans—and potentially geotarget concert listings—and use apps like BandsInTown to alert them of upcoming gigs.

Even Jay Z knows the value of having data on your fans. (After all, he’s not a businessman, he’s a business, man.) The rapper came under criticism in 2013 after a partnership with Samsung. Fans who installed Samsung’s Magna Carta app gladly volunteered all kinds of information about themselves to get the album for free and “unlock” lyrics to the songs. It also came to light that the app could log calls and track GPS coordinates.

Plenty of small-time musical acts acts are carving out worthwhile full-time livings with few or no middlemen separating them from their paying fans, and they’re doing it online. The Internet has slain the radio many times over as a meaningful source of new music—and especially as a source of data about that new music. The tides of musical taste are constantly turning, not only in terms of album sales but in the quantifiable chatter taking place on Twitter, Tumblr, and beyond.

And where there is novel information to be studied, there are entities to monetize it.

The next big thing

A New York City company called Next Big Sound tracks all order of musician-fan interactions online, measuring the digital footprints of music lovers everywhere to find the musical acts of financial consequence of the future.

The company’s data index is effusively rich. Since its formation in 2009, Next Big Sound has tracked Spotify, YouTube, and SoundCloud plays; every musician interaction on Facebook, Twitter, Tumblr, Vine, and Instagram; and even keeps tabs on Wikipedia page views as well. Consider that it has consistently measured all of these for every band for the past six years, and it’s clear that Next Big Sound indisputably falls into the category of being a “big data” company, sorting the signal from the noise in an unsightly heap of information.


Employees crunch numbers to find the hits at the Next Big Sound office.

“My favorite definition of ‘big data’ is that it crashes Excel when you try to open it,” Next Big Sound CEO and cofounder Alex White told me at the company’s Chelsea office, a homey workspace with vinyl records everywhere and a makeshift recording studio for employees.

Where there is novel information to be studied, there are entities to monetize it.

The value of data touches every industry. Consider Target, which famously predicted that a woman was pregnant before she knew it herself and began marketing diapers and baby food to her accordingly. By simply tracking her purchases on a long enough timeline, Target’s algorithms pegged her as a woman soon to be with child. Now imagine what someone could do with the aggregated data surrounding the music industry online.

“What we’re talking about is every social interaction with an artist being timestamped and geolocated globally,” White added. “How do you process that quickly and in a way to help a band, agent, or promoter make sense of it? Our company’s mission is to take all this data and make it useful, instead of overwhelming, confusing, and annoying.”

Next Big Sound was born after White met cofounders David Hoffman and Samir Rayani during undergraduate courses at Northwestern University. The trio was fascinated with solving the problem of how bands go from playing their garages to headlining nationwide tours. White seemed predisposed to work in the music industry—he’s the son of a professional cellist and fondly recalled an internship that saw him stapling CD sales reports together for his superiors—but he knew there had to be a better way to keep tabs on popular music. He and his cofounders sought to “unlock the black box that is the music industry,” as White put it—to mathematically find the point where bands catch enough buzz to take off on their own.

“The thesis of this company is that attention precedes monetization,” said White. “Before you go to a concert, before you buy an album, you do a Google search or look up an artist on YouTube or Spotify. All these little signals are indications of intent or likely future purchase activity.”

Next Big Sound harnesses the power of these data points to establish a sense of new artists’ trajectories. Are they gaining momentum or are they stagnant, like so many artists that find a way to crack Hype Machine charts only to never be heard from again?

It bears mentioning that there is a band called Big Data, which White says has “a very high likelihood of hitting the Billboard charts in the next 12 months.” And he’s got the clout to make that kind of call. Next Big Sound’s big data algorithms called out Sam Smith as a star with commercially viable potential in early 2013. He performed on Saturday Night Live in March the following year, cementing himself as a participant in the contemporary musical zeitgeist, and the company boasts “tons of examples like that,” White said. To put it another way, Next Big Sound is a fortune-teller of the music industry, backing up its A-lister predictions with math and statistics. (For the record, Next Big Sound thinks you should be keeping tabs on Elliphant, Sheppard, James Bay, Fetty Wap, and Kygo.)

Next Big Sound is a fortune-teller of the music industry, backing up its A-lister predictions with math and statistics.

Liv Buli is Next Big Sound’s data journalist; it’s her job to write about bands that the company’s software has pegged as on the move or noteworthy. She’s a music lover who comes from Norway, where streaming was the norm long before it was in the U.S. “It’s easy to forget that for most of [the U.S.], AM/FM radio is how a lot of people in cars are listening to ‘new’ music.” She recalled that when she joined the company three years ago, it was “an uphill battle” to convince others that music data mattered, that it was worthwhile source material to distill for marketers and music industry professionals. “Now it’s much more of a given to anyone I talk to about it. They understand that there’s value in leveraging these numbers,” she said.

This delayed realization that the numbers tell the truth ahead of time echoes White’s experience as well: “The drum we’ve been beating for six years around data in the music industry is finally starting to beat itself,” he said. “I’m probably most proud and happy when I see conferences that have a data track, a panel or three. This wasn’t the case even three years ago. We had to fight for attention. Now it feels like everyone knows it’s important. Of course you’re collecting and analyzing the data you have.”

For many years, Next Big Sound’s charts were based on “derivative acceleration,” a measure of how quickly an artist’s fan base grows at a given moment compared to its usual rate of growth. Bands suddenly getting noticed much more often than their baselines would be flagged as acts of interest. Buli explained that the company makes use of several social metrics to arrive at these figures, pulling data from the aforementioned Facebook, Twitter, YouTube, and beyond.

“Last year we updated the chart to reflect our new predictive success algorithm,” she said. “The model is built based on artists that have [previously] reached a certain success criteria, and what their social and streaming metrics look like.”

With a literal mathematical model of commercial success, Next Big Sound can apply it to every artist in its database and assign each one a likelihood of success depending on their social numbers. It’s just running calculations on lots of data, digging through the dirt to find the gold. (In further testament to the value in collecting and analyzing the public’s digital echoes, the company has already kicked off Next Big Book, which plies the same big data methodology for the publishing industry, tracking new authors and newly released books on their way to the bestseller lists.)

The music industry’s prior efforts to identify the moneymakers have more in common with gambling than they do with thoughtful business. “In the leadup to an album release, before you have any real revenue data, which of the millions of metrics should you look at to know if you’re on pace to hit your first big number or not?” White asked. “Traffic to an artist’s Wikipedia page is something we’ve identified as being important, and we are the sole provider of that data to the music industry.”

“The thesis of this company is that attention precedes monetization.”

It’s a sensible assertion; reading a Wikipedia page requires attention, and in a streaming-based music economy with limitless shelf space, attention is the scarce resource that spells out dollar signs. It’s an approach that had never been really been adopted before, with terms like “data science” and “big data” only really coming around in 2011 or 2012.

“It’s really funny to be categorized as a big data company,” White said. “We didn’t even know that was a thing when we started in 2009.”

Leveling the playing field

I’ve recently become taken with a raucous three-piece band from Beaumont, Texas, called Purple. They’re still largely under the radar, playing small shows that only attract their more committed fans, but a recent plug on NPR’s All Songs Considered dropkicked them to the center of my attention, and Next Big Sound has gobs of information on this young group of merry noisemakers.

Purple is currently tagged as “undiscovered.” It’s the first of five stages the company has designated for all musicians; from there the band will eventually move to Promising, Established, Mainstream, and Epic. They boast an “engaged” audience that interacts with the band online often, and they are generally growing in notoriety.

Given the undeniable charisma of Purple’s singing-and-drumming frontwoman, Hanna Brewer, the band’s listenership skews a bit female. Most people interacting with them on Twitter are of the double-X-chromosome persuasion, according to the numbers: 



When it comes to how musicians are actually using Next Big Sound data, it varies. Commonly it’s to “assess releases, tours, and other promotional efforts and what impact these efforts have. Beyond this, users have used Next Big Sound for anything from determining their next single based off of online traction to determining whether an artist is prime to break in any particular market,” Buli said. 

Familiarity with one’s data, fan base, and how those fans interact with the band is an essential asset in a fast-changing market. It’s a similar story for non-musical brands as well.“We are seeing major brands looking to smaller artists early on in their career to work with. They want to establish a long-lasting, deeper relationship with these artists and their fans. Artists need to know exactly who these fans are, and how they align with a brand’s target market, to help make an argument for partnership,” said Buli.

The Internet is the hammer and the nail all at once—both the source of valuable data and the best place to go to squeeze money out of it quickly and efficiently.

Next Big Sound has collaborated with Pepsi to help the company understand which artists it should work with next and to understand the impact when it does get involved. In practical terms, this means finding the best songs for Pepsi commercials or finding the best musicians for the company to sponsor. It identifies which artists should the company be collaborating with to the benefit of both parties. 

Bands aiming to actually make a living need to do whatever they can to both cut costs and maximize potential income. The playbooks for tech startups and technologically savvy independent music acts have a lot in common, and big data technology means that even the little guy can have the same access to valuable statistics that a Photoshopped, focus-grouped boy band would. In this case, the Internet is the hammer and the nail all at once—both the source of valuable data and the best place to go to squeeze money out of it quickly and efficiently.

So much of our behavior takes place online nowadays that it leaves behind quantifiable information, captured somewhere—counted, preserved. “Big data” doesn’t simply refer to having lots of data; it really means having too much.

“There’s so much unstructured data out there that it’s beyond human control,” said Nick Goggans of audience analytics company Umbel. While big data technology certainly requires some right-brained engineers to keep the machines and algorithms properly tuned, Goggans says that the liberal arts play an equally important role in turning unstructured data into useful information. “Big data needs people who understand how to ask questions of things. When you ask questions, you simplify the equation and cut to the chase.”

Next Big Sound’s perpetual question to its dataset is “What should we be listening to next?” The company faces very little formal competition in its space, being the only entity to tie together social, sales, streaming, and tour/event data together every single day. Nielsen notably provides weekly music sales reports to its clients, but for now Next Big Sound can lay claim to being “the number one provider of artist recommendations to brands,” Buli said. 

Even though the simple joy of listening to music has become a professional obligation, White still enjoys it. “My favorite day of the week is still ‘new music Tuesday,’ when all the new albums come out. I throw everything in a playlist and add the stuff I like to my collection.” Do his friends constantly ask him what’s new and what they need to be listening to? “Yes,” he said. “And that’s OK.” 

White surely has plenty of names to throw at listeners eager for a new sound at a moment’s notice. And he can give those recommendations with the confidence of a guy who’s quite literally been keeping tabs on what the Internet-connected world has been listening to for the past five years.

Illustration by J. Longo