By Cherie Hu
Twitter’s recent housecleaning of some 70 million fake and automated accounts illuminates just how pervasive audience manipulation has become in the digital era. For Twitter, the fake accounts can create a shadow army of followers that has comparatively little monetary effect. But perform the same manipulation with music streams, and it constitutes fraud.
Tidal has found itself awash in accusations of data manipulation. As recently as May, Norwegian newspaper Dagens Næringsliv and the Norwegian University of Science and Technology (NTNU) accused the Jay-Z-owned service of falsifying tens of millions of streams for Beyoncé’s “Lemonade” and Kanye West’s “The Life of Pablo” albums. While Tidal has denied the claims on multiple fronts, a company rep tells Variety that several investigators are currently on the ground at the company’s offices looking into a potential data breach.
Fraud is applicable because there’s a tangible price tag involved in the consumption of a song: Labels and other rights owners are paid on a pro-rata basis, according to proportional volumes of on-demand streams. The average per-stream payout may not look like much — $0.004 for Spotify, slightly more for services like Apple Music and Tidal ($0.008 and $0.012, respectively), although exact rates depend on the type of artist or song.
But they can add up. A top hit like Ed Sheeran’s 2017 monster “Shape of You” would distribute millions of dollars in performance royalties to its songwriters and even more to the master-rights owner. Using Goldman Sachs’ projection that the streaming sector will hit $34 billion by 2030, millions of dollars in fraudulently acquired funds could be making their way through the royalty chain. Though unlike Twitter, which wiped out 6% of its users, the number of fake music streamers has not been determined. Says one major label head: “It’s not something we’re currently concerned about, but that’s not to say we won’t be in the future.”
“Music streaming payouts are a zero-sum game,” says another industry insider. “It is imperative that services are vigilant and sophisticated in their controls to ensure that streaming fraud doesn’t dilute payments to the artists who have rightfully earned those payments.”
Here’s how “playola” works at playlist-promotion companies like Spotlister: A customer pays the company to secure prominent placement of a song on key playlists, such as those on Spotify. When a track is uploaded, it is analyzed and its metadata is used to send it to the most appropriate playlists.
But that’s not the only way to game the system. In 2017, Post Malone’s label, Republic Records, found itself the center of controversy for a seemingly sanctioned loop of the hook to his song “Rockstar” that was posted on YouTube. Although it contained only a snippet of the tune, it played continuously for three minutes and 28 seconds and quickly racked up more than 40 million views.
“It’s pretty easy to buy guides online on how to stream your own content repeatedly.”
Christine Barnum, CD Baby director of finance
Two months later, YouTube acknowledged, “Loop videos that feature misleading and inaccurate metadata violate YouTube policies, and we are actively working to have them removed.”
But earlier this week, the New York Times published a piece titled “The Flourishing Business of Fake YouTube Views” in which it contended that “tens of millions of fake views” make it through the platform’s anti-fraud detection triggers daily.
Spotify has kicked outlets like Spotlister off its platform on a case-by-case basis — stand-alone sites like Social Media Experts, Streamify and StreamKO offer “Spotify promotion” with prices ranging from $5 to $995 — but concerns remain about whether that enforcement is being replicated meaningfully across entire platforms. And if not, are the labels or the streamers themselves complicit in royalty fraud by transferring payouts? The legal consequences for fraud convictions, and even of conspiracy to defraud, include hefty fines and up to five years in prison.
“We’re seeing an uptick in the type of fraud where people are distributing their content through us and signing up with a legitimate credit card, but then their intent is to manipulate streams and rip off a [digital service provider],” Christine Barnum, director of finance at CD Baby, tells Variety. “It’s pretty easy to buy guides online on how to stream your own content repeatedly. There are also more instances of well-meaning artists accidentally signing up for some sort of ‘marketing service’ that’s actually committing fraud.”
Moreover, on Spotify’s free tier, ad fraud is also becoming more prevalent. Spotify had to decrease its total reported content hours streamed in 2017 by 500 million, due to 2 million listeners who used unauthorized apps that blocked ads on the service. With freemium, there is “not only more susceptibility to fraud but also less incentive for Spotify to do something about it,” argues Rami Essaid, CEO of bot-defense start-up Distil Networks. “With the subscription model, there’s a finite number of dollars to split up, and if part of that goes to fraud, no one else in the ecosystem is happy.”
Interestingly, the Q2’18 report still included these aforementioned, suspicious ad-blocking users — which account for just under five percent of the company’s total ad-supported base — among the service’s 180 million total monthly active user count.
“In general, we should not think of fraud as the polar opposite of what Spotify and other platforms do,” says Patrick Vonderau, a professor at Stockholm University who successfully forged fake Facebook and Spotify accounts for his research on streaming. “I do not believe that bots are a total breach with Spotify’s own informational norms. As reports have shown, the system can be easily gamed. A word like ‘fraud’ does not acknowledge these fleeting boundaries between what is legal and illegal, acceptable or not.”
Some startups like Rebeat are building independent solutions for labels to monitor and detect irregular streaming activity on their content. Others have suggested a “user-centric” royalty distribution model for premium tiers, whereby payouts are determined on the individual subscriber level rather than on the aggregate company level. For instance, if a subscriber paying $9.99 a month for Spotify listens exclusively to Rihanna, Rihanna will get all of that user’s $9.99, rather than having that cost distributed pro-rata among the platform’s millions of artists. That way, a bot generating artificial streams would simply be capped by the subscription fee.
But that still doesn’t account for streaming services’ increasing reliance on advertising, especially as they expand into under-banked markets across Asia and Africa. As the potential scale and business models of music streaming increasingly mirror those of Facebook, Twitter and other social media platforms, they are inevitably facing the same rogue activity, not to mention the same incentives — or lack thereof — to take action.
For its part, Spotify stated in a just-released second-quarter earnings report, “We continue to work to identify and remove users from our reported metrics that we consider to be ‘fake’ users based on various criteria,” with the caveat that “some such users may remain in our reported metrics because of the limitations of our ability to identify their accounts.”