For a few years in the late 2000s, Pandora was the on-demand DJ for tens of millions of people, creating the soundtrack to college dorm room parties, quiet coffee shops, busy kitchens, and family get-togethers. The days of building massive MP3 music collections through file-sharing was receding quickly into the past, and instead the shared experience of radio was making a comeback via the clever algorithmic matchmaking of Pandora's endlessly customizable stations based on individual taste. Today it's a feature we take for granted across every music service, even if Pandora's implementation still seems like it was the best. Pandora itself, however, can feel like an afterthought. Betamax to Spotify's VHS, or maybe more accurately, MySpace to on-demand streaming's Facebook.
It's not that Pandora was oblivious to its competition, or complacent about its place in the industry. It was, however, an innovator in digital music at a time when the major labels were hostile to the entire concept and would fight on every front to preserve the lucrative of the compact disc era. To take on this legal and lobbying juggernaut, Pandora needed a clever strategy to avoid the kind of head-on fights that had sunk Napster. The solution was the radio model of music licensing, a brilliant strategy at the time, but one which would be the subject of a long fight between Pandora and the recording industry. Pandora would win that battle, but in doing so, it also found itself stuck with a business model that could not evolve alongside the streaming space.
Pandora’s personalized radio service took the FM listening experience, put it online, and exploded the typically narrow path to music discovery for millions of people. It was the first real introduction to digital streaming music for a lot of America’s population in the mid-2000s. People could type in a song or artist they liked and get a never-ending stream of related music.
On the surface, at the listener level, the magic was in how all the songs were linked and connected. Behind the scenes, the magic was in how Pandora was able to provide access to all these songs, without asking the major labels for permission.
Radio broadcasters don’t have to spend exorbitant amounts of time and money trying to license every song from major record labels: they only need to pay a small fee each time songs are played. So after developing its music discovery technology, the Music Genome Project, Pandora went into radio to take advantage of the same licensing agreements already in place.
For Joe Kennedy, Tim Westergren, Jessica Steel, and Tom Conrad, the four Pandora executives leading the company's rebirth as a digital broadcaster in 2004, this was the only path forward. Radio was the path to profitability in the music industry.
“It was already clear to us that most of digital music was a disaster from a business standpoint,” said Former Pandora CEO Joe Kennedy. “At that time the four labels didn't want digital music to really exist, but we found this one piece in internet radio where there was the statutory license that, I felt, gave the potential to create a business of value and avoid this phenomenon one board member referred to as an organ donor business — where you just exist to funnel money to the major record labels.”
As a quick background, Savage Beast Technologies was founded in 1999 by Tim Westergren and would later become Pandora. The company spent years developing its Music Genome Project music recommendation engine to help people find new songs and artists similar to their existing tastes in music. It powered a few in-store kiosks at Best Buy and Tower Records, but the Music Genome Project itself never found a viable market fit. In 2004 Larry Marcus and Walden Venture Capital’s lead investment saved the company, starting a new chapter. Jessica Steel joined as VP of Business and Corporate Development. Tom Conrad came into engineering, but quickly rose to Chief Technology Officer. Founder Tim Westergren moved from CEO to Chief Strategy Officer and Joe Kennedy was hired to be the company’s new CEO.
Kennedy got to work on a new business plan and put the pieces together to pursue radio. Looking at LaunchCast’s success (a similar service to Pandora, which would later be acquired by Yahoo), along with potential market size and fixed costs through the statutory license it was their path towards profitability for the startup.
Pandora launched its personalized radio service in the second half of 2005 paying a per performance rate of $0.000762 each time a song was played.
The Battle Over Rates
Pandora and other webcasters were paying song rates from the 1998-2005 timeframe. By the fall of 2005, Pandora was already too late to officially participate in the hearings happening to assess rate adjustments for the next period of time. Kennedy did, however, stay connected to how testimonies to the Copyright Royalty Board were progressing. The CRB was created under the Copyright Royalty and Distribution Reform Act of 2004 with three permanent copyright royalty judges.
There were several different groups and sides presenting their case for what they wanted, one of which was the Digital Media Association (DiMA) representing 42 companies including AOL, Live 365, Microsoft, Yahoo!, and AccuRadio LLC. Another, representing the entrenched music industry, was SoundExchange and executives from Atlantic Records, Sony BMG, Universal Music Group, and other labels. Mixed in were groups identifying as terrestrial radio, small webcasters, large webcasters, commercial, non-commercial, and so forth. There was plenty of nuance to the companies and groups wanting special considerations to the rates they would pay.
SoundExchange and its side argued for 30 percent of gross revenues from webcasters or a performance rate beginning at $.0008 per performance in 2006 and increasing annually to $.0019 by 2010, whichever was greater. It wanted the rates to be as high as possible.
The DiMA group wanted a fee structure of either $.00025 per performance or 5.5% of revenue directly associated with the streaming service. For a lot of these internet companies, music was not their main business, but an ancillary one.
Pandora may have entered the music industry a little naively, planning for a world where the statutory rates remained low and they had time to bloom. Even before its plans for radio, it had run out of cash multiple times. In 2000, venture capital dried up fast from the first dot com bubble bust and left the company scrambling for new funds early in its life. Tim Westergren, always the easygoing and formidable guy, never laid anyone off, he simply asked those who could continue working without pay to stay with the eventual promise of reimbursement once new funding came in.
Even after receiving around $7 million in funding in 2004, Pandora didn’t have the capital to either negotiate direct licensing with record labels or the runway to offer free, ad-supported radio. It had to launch as a subscription service to cover some of the costs. But, shortly after launch it secured a large investment by James Feuille and CrossLink to make the transition to free ad-supported radio.
“Fundamentally, I did not believe people would pay for radio,” said James Feuille. “The idea was $3 a month, $36 a year, with no advertising and I just didn't believe you could build a big business like that.”
The numbers from the 2005 launch reflected most people’s reluctance to pay. Pandora radio launched with 10 free listening hours before it presented a paywall and cut off listening to non-subscribers.
“They generated 500,000 users, zero to 500,000 users in six weeks,” said Feuille. “At the end of six weeks, 40,000 people had subscribed, it was .08, not even 1 percent.”
Despite its growth and the excitement it was generating, Pandora spent its first several years struggling to keep the lights on. It had publicly warned of pulling the plug. It was paying artists according to the law — which wasn’t a given in the early days of digital music — but it didn’t think it could afford to have the rates double or triple.
The Night The Lights (Almost) Went Out in Oakland
The Copyright Royalty Board presented its decision in March 2007. Included was a summary each side had been making against the other: “SoundExchange accuses the Services of seeking a marketplace characterized by perfect competition. DiMA and the Radio Broadcasters claim that SoundExchange is championing a marketplace characterized by monopoly power on the seller’s side.”
The three-judge CRB sided with the proposed rates from SoundExchange and the major labels. The new per play rates would be $.0008 for 2006, $.0011 for 2007, $.0014 for 2008, $.0018 for 2009, and a per play rate of $.0019 for 2010.
“It was March 2007, I'll literally never forget the moment,” said Joe Kennedy, Pandora CEO at the time. “I had just come through the tunnels on my way to work in Oakland and Tim [Westergren] gives me a call and says there's a decision out of Washington. He read me the rates and it was like, oh my god, the rates are almost triple. I immediately knew that’s game over, there's no path forward.”
Westergren knew Pandora was in trouble so he immediately spearheaded a grassroots campaign to persuade its listeners to contact their congressional representatives. This effort resulted in the Webcaster Settlement Act of 2008 that was able to bring those involved parties back to the table and try to negotiate a different rate settlement. It allowed for a simplified process to amend any deal that could be renegotiated between the parties. Webcasters, including executives at Pandora, were anxious to find a number they didn’t think would put them out of business.
Although the bargaining table had been set again, this time joined by Pandora, there was a significant amount of division among the companies and groups. RealNetworks with its Rhapsody streaming service saw Pandora as a competitor and potential threat. The National Association of Broadcasters saw internet radio as less desirable and profitable and was happy to see it go away or remain small. They were primarily interested in addressing their problem of simulcasting over the internet. Yahoo was an advertising company and so it didn’t want revenue sharing from non-core products to eat at its bottom line.
The labels had the upper hand with the increased rates they had asked for already in their pockets. Plus, negotiators for SoundExchange and the Recording Industry Association of America understood these divides and played them against each other.
As CEO of Pandora, Joe Kennedy was representing it in these negotiations, trying to keep the company on a path to profitability. Pulling out all the stops, Pandora even went as far as to offer private stock in exchange for a compromise agreement. The labels weren’t interested.
“The major labels came back and said, Well, you know, we shared this with our business people, and they took a look at it, and their analysis says, even at the lower rates, there's no chance you guys are going to make it,” said Kennedy.
In its first years as a digital radio service, Pandora and its millions of listeners had become an annoyance to the major labels. Historically, major record labels exhibited all the power in the relationship with terrestrial radio. FM radio was a promotional vehicle for labels to sell their artists. Digital services like Pandora were shifting that promotional power away from the labels with algorithms and user input.
If Pandora continued growing it would become a threat to power and control within the music industry. Circumventing directly licensing music through its use of the DMCA was seen as combative since digital music was already marked as undermining established business. Pandora was scrappy, having already moved Congress into action once, and that was worrisome.
The 2008 Settlement Act didn’t do much when it was signed by Congress. What it did do though was allow for a simplified process for new settlement approvals. The stage was set, if Pandora and its other webcasters could reach a deal with SoundExchange and the labels by the stipulated date of February 15, 2009 then it would go into effect simply by submitting it to the Copyright Royalty Board. There wouldn’t have to be any pre-hearings, public comments, or any of the previously required steps in the tedious process.
Despite Joe Kennedy’s best efforts, the deadline passed and Pandora was momentarily left with a decision. Would it accept a mountain of debt that rose higher as it gained popularity and more songs were played or would it pack it all in and call it quits?
“I wrote an email for the board that night and more or less said, I tried, failed, it's over,” said Kennedy.
Kennedy called Tim Westergren in the middle of the night to let him know. Westergren answered from a bus in Jordan. He was in the Middle East on a congressional delegation with his wife. Kennedy explained that the deadline had passed and negotiations had failed. The previously tripled rates set by the Copyright Royalty Board in 2007 would take effect.
Westergren replied that Congressman Howard Berman was currently on his bus. He would talk to him about the situation and see if there was anything he could do.
Congressman Berman was representing Los Angeles’ 26th district in 2008 and was on the Intellectual Property subcommittee, part of the Judiciary Committee. Congressman Berman had the right connections with people from most parties involved in the entertainment industry.
“I fully believe we had Right on our side back then,” said Westergren. “And I had the perfect timing to just explain the whole story, not the caricature that had been told to him. I think he heard it and said, what you're asking for is fair. Ultimately that’s why I think he intervened.”
“[Berman] called from Amman back to Washington and got some kind of procedure where a page on the floor can open the chamber and make some change,” recalled Westergren. “And because he was Chairman of the Judiciary he had some abilities or power to extend the deadline.”
Westergren called Kennedy back and told him that Congressman Berman said to keep going. This was happening in the middle of the night back in the U.S., a few hours after the deadline had passed. Because of the speed, negotiations didn’t have a chance to unwind from where they were left the day before.
“I immediately got on the phone with two more people,” said Kennedy. “First the point person in the negotiations and then our biggest ally who was with the Independent Music Association. I said, I just talked to Tim, Tim just talked to Berman, and Berman said we should keep going.”
In light of the immediate progress, Kennedy quickly amended his initial email to Pandora’s board admitting defeat, saying everything was over — it wasn’t yet.
“I think by the time the board got the first memo, I'd already sent a second memo that told the whole story of Tim talking to Berman and to hold on because maybe there's still some hope,” recalled Kennedy chuckling.
On July 7, 2009, more than two years after the rates had been set to increase, a settlement deal was formally announced. In the press release, John Simson, Executive Director of SoundExchange, said, “It’s a creative, groundbreaking approach that we wanted to try, and we hope it will work well for everyone involved—the artists, labels and eligible webcasters.”
The new agreement allowed for webcasters to pay per performance rates ($0.0008 retroactive to 2006 and increasing to $0.0014 by 2015) or 25% of revenue — whichever was greater. The per performance rate was a little bit lower, but the introduction of the revenue sharing would help webcasters keep costs in check and plan for the future.
The Power of Tens of Millions of Listeners
Pandora might have been on the ropes when it came to money, but it was a growing powerhouse of influence. In 2009 Pandora had seven million monthly listeners. Five years later in 2014, it had over 81 million monthly listeners. Westergren even attributed its active listeners calling members of Congress in the early days to its new settlement and Pandora’s ultimate survival.
Pandora’s radio service might have just hit at the right time to fulfill consumer demand and grow in popularity. It could have also had the right timing to land in the middle of Apple’s mobile revolution fueling its growth, but the truth is that it spent years working on its streaming backend and engineering so that it could hit it big overnight with the iPhone’s App Store.
Through Jessica Steel’s leadership, the team identified mobile early on and started by putting Pandora on flip phones with network carriers like Sprint and AT&T. At the time, this only resulted in a few hundred listeners, despite the major engineering effort involved. After Steve Jobs announced the iPhone in 2007 it became apparent that this new internet connected, “music player,” device in people’s pockets needed to be the future of its mobile efforts.
After pushback on only allowing web apps for the iPhone, Steve Jobs announced that native apps would be coming to the iPhone. In the interim, Apple Senior Vice President Scott Forstall invited Tim Westergren and his CTO, Tom Conrad, over to a local Cupertino lunch spot. The trio talked for hours about what Pandora had learned about streaming audio from putting apps on flip phones, like Motorola’s RAZR, for wireless carriers. The meeting ended with a question for Forstall.
“What, if anything, can we do at Pandora to get ready for the next generation of iPhone that includes an app store and native APIs?” asked Conrad. “Forstall said, it wouldn't be a waste of your time to jailbreak some iPhones and use the kind of back door toolkits that were being distributed by other people to build a native Pandora app while we get our act together at Apple on something more formal.”
So, Conrad, designer Dan Lythcott-Haines, and many others on the team got to work jailbreaking iPhones and working on a Pandora iPhone app ahead of the official APK release. Then, on day one of the App Store launch, Pandora was the first internet radio app available. Nine months later the Pandora app was installed on 21 percent of iPhones.
Five years after that first iPhone app, nearly 80 percent of Pandora’s radio listening was on a mobile device. But even with a great mobile experience and ubiquitous access across all kinds of devices into the future, its linear, radio style appeal was encountering pushback from listeners.
“We always knew there was a portion of how people listened to music that we were not addressing,” said Joe Kennedy. “I think what changed over time was obviously the emergence of Spotify, but I think alongside that is not just in music but across the board, people's willingness to subscribe to things on the internet grew quite a bit over time.”
The Spotify Effect
By 2010, after it went through these settlement negotiations, Pandora was fully entrenched in radio. It had gone to court for it. It had become the face of internet radio and was becoming the torch bearer for the idea of radio in a lot of people’s minds. There was a consumer demand for this style of music service but Pandora and its executives were getting locked into the protection and cover they thought the statutory license afforded it.
Pandora was feeling heat from listeners who wanted to be able to listen to any song they wanted to at any time. They wanted more skips and other features that were outside of the statutory license’s scope. This pressure was one piece, of many, that led to Joe Kennedy announcing he would be departing as CEO in 2013.
“There’s many pieces to this puzzle of why I decided to leave,” said Kennedy. “Among them was the natural strain of having worked with the board for a long time, the strain of being a public company, and this very significant pebble in our shoe of what to do about Spotify and the on-demand business.”
Pandora finally launched an on-demand offering in 2017. It took years to repair relationships and get to a place it could work with the major labels. The work started secretly in 2014 which eventually led to Pandora acquiring Rdio’s on-demand licensing deals and other assets in 2015. In exchange for these on-demand licensing deals, Pandora agreed to renegotiating new, direct, radio licenses outside of the DMCA shelter.
Pandora spent the first half of its life chained to and defined by the DMCA’s statutory royalty rates. It then spent the next years not quite sure how to move forward. It saw the terrible financial deals that Spotify was making in order to gain on-demand streaming access to the major label’s music catalogs and didn’t want to be in that same boat. It tried getting into ticketing and events in order to control its own financial destiny, but it didn’t pay off.
“The reason we really did jump into on-demand was more about the porousness of publishing rights than anything else,” said Tim Westergren. “The industry had us in their crosshairs and after a while it's hard to be at war with your suppliers. There was too much potential for publishers to do monkey business and we had less and less confidence in the security of some of these statutory structures. We were vulnerable.”
“I think that we still squandered an enormous opportunity having survived all [those settlement negotiations] by not pivoting to on-demand fast enough,” said Westergren. “I feel incredibly proud and sort of marvel at what we got through, but I also have a lot of frustration about how we let it slip away after we established such a lead.”
The way Westergren describes it, Pandora should have become Spotify, before Spotify had the chance to.
“We should have done what Spotify did and ate a pound of flesh to get the industry on our side, then expanded the scope of the product and then really gone global and become an all-you-can-eat service,” said Westergren.
Today, in 2021, it’s nearly impossible to determine what Spotify pays in royalties per play. How much a song is worth varies among labels, artists, and other complex contractual details — including whether plays come from Premium subscribers or free listeners. The deals remain secret until frustrated artists eventually spill the beans on their own terms and provide a peek behind the curtain. In the end, musical artists continue to be the ones dealt the bad hand. Spotify pays major labels huge financial sums and upfront guarantees while it’s estimated that most artists will see payments of between $0.003 and $0.006 per song play.
Pandora still remains an active player in the music streaming space and counted 58.5 million monthly active listeners at the end of 2020. Depending on which music listener you ask, Pandora is a fond memory of digital music coming into its own, or its radio service may still be a vital part of their passive listening experience. Attention wise, however, Pandora is being left behind as Spotify and Apple Music push faster and further into on-demand streaming, sucking all the oxygen out of the room.