When Pandora launched its music service in late summer 2005, it came with a straightforward pitch: ten free hours of music a month before users had to subscribe. “We didn’t have much money in the bank,” recalls former CEO Joe Kennedy. The paywall was meant to keep costs in check while laying the groundwork for a subscription business.
Except, people didn’t want to pay. The free trial drew huge interest, while conversions to paid were abysmal. Even worse, the most devoted fans found ways to bypass restrictions. “People were getting to the end of their trial, and hacking around [to keep] listening,” says an early employee who spoke to Fast Company on the condition of anonymity. “It was very clear that we were onto something.”
Within months, Pandora pivoted to a free, ad-supported model, and growth exploded. Over the next decade, Pandora became North America’s most popular music streaming service, peaking at more than 81 million monthly listeners.
But the past 10 years have been defined by steady decline. Last year alone, Pandora lost close to three million monthly users. This spring, its audience averaged around 43 million.
In hindsight, the culprit seems obvious: Spotify. But Pandora’s two-decade journey is not simply a story of losing to a superior rival. It’s a case study in the innovator’s dilemma—riddled with missed opportunities, flawed pivots, and a toxic relationship with major labels that undercut the company long before Spotify entered the market.
“They were never going to give us deals that were going to allow us to succeed,” says the early employee about the major labels. “They just hated us.”
This two-part history of Pandora draws on numerous interviews with former employees and executives, many of whom were granted anonymity to speak candidly—either out of fear of retaliation or because their current employers bar them from speaking to the media. (Part two of this story will be published later this week.)
Finding the core of every song
Pandora’s paywalled launch in 2005 wasn’t the only sign it was running on a shoestring budget. Another was its headquarters: instead of leasing space among San Francisco’s trendy startups, the company moved into a nondescript office building in downtown Oakland. “It had previously been a parking garage,” recalls Kennedy. “Very unglamorous, not particularly a great space.”
If you visited Pandora’s office in those days, you’d find a bunch of people hunched over their computers, wearing bulky headphones, cardboard boxes full of CDs within reach. Many of them were musicians or musicologists, hired by the company for cheap on a part-time basis. Their job was to meticulously analyze every song the company could get its hands on not just by style and genre, but also mood, instrumentation, vocal timbre, and more. That painstaking work fed into the Music Genome Project, a vast database that powered Pandora’s playlist algorithms.
Over the years, human experts hand-categorized 2.2 million tracks. That information not only shaped recommendations but also trained machine-learning models that expanded the catalog further. Even today, the Music Genome Project influences what Pandora streams next. “You probably couldn’t get funding to do that kind of thing anywhere now, but it’s a hell of an incredible data source to have,” says a former employee who worked on the recommendation engine.
The Music Genome Project was the brainchild of cofounder Tim Westergren, who started Pandora’s predecessor, Savage Beast Technologies, in 2000 as a way to boost CD sales for B2B partners. Monetization proved elusive, forcing the company to reinvent itself with a direct-to-consumer model focused on streaming, and seek new capital. Westergren had to pitch Pandora hundreds of times until he was finally able to raise $9 million from Walden Venture Capital, $2 million of which immediately went to paying back employees who had worked for months without a steady paycheck. “It had been a rough road,” Kennedy says. “Picture a group that had spent the last five years crossing the desert, running out of water.” (Westergren did not respond to Fast Company‘s request for comment.)
What Pandora lacked in money, it made up for in grit and determination. “It was a super resilient group,” Kennedy says. That resilience showed when staff hacked an early iPhone to start building an app before Apple had even opened the device to third-party developers. The gamble paid off: Pandora became a prominent App Store launch partner in 2008, and usage skyrocketed.
“That was a serious inflection point,” Kennedy says. “We would have a hundred thousand new listeners every day.”
A clash with the major labels
Listeners flocked to Pandora in part because of its simplicity. Type in the name of a favorite band, and Pandora would generate an endless stream of similar music. Users could refine those streams with thumbs up or thumbs down ratings, until eventually they could just let Pandora run for hours, interrupted only by the occasional ad break.
That stood in contrast to other services of the time. Apple’s iTunes store, launched two years earlier, charged $0.99 per download. And in 2001, Listen.com introduced Rhapsody, a $10-a-month all-you-can-stream subscription, pioneering a model that was later perfected by Spotify.
“Nice idea, but there’s just not a business there,” Kennedy remembers thinking when he first heard of Rhapsody. Pandora chose a proven model instead. “Eighty percent of music listening historically, pre-internet, was to radio,” Kennedy says. Pandora wanted to evolve that radio model, and supercharge it with the capabilities of the internet.
There was also a practical reason behind Pandora’s radio-like approach. Services such as Rhapsody had to secure expensive licensing agreements for every song in their catalogs. Pandora instead relied on a copyright law provision that let it bypass individual deals and pay into collective licensing pools, just as traditional radio did.
That strategy infuriated many in the music industry, still reeling from the MP3 piracy wave. Unlike Napster, Pandora did pay royalties for every song it streamed, but major labels believed those payments were far too low. “It got pretty tenuous pretty quickly,” recalls an early employee.
A year after Pandora’s founding, the major labels pushed to triple its royalty rates. “[Those rates] would have put us out of business,” Kennedy says. Pandora started lobbying lawmakers, and even got listeners to contact their representatives en masse. “It took two bills being passed by Congress, signed by the president, that enabled us to survive,” Kennedy says, referring to the 2008 Webcaster Settlement Act and a subsequent 2009 settlement agreement that set lower royalty rates..
The resulting more moderate rate increase was welcome news not only to Pandora and its users, but also to a growing number of independent musicians. Smaller bands discovered that Pandora’s algorithms offered visibility they could never get on mainstream radio. “Pandora ended up playing over 30% independent music,” Kennedy says. “It wasn’t because we tried to. It’s because we actually were so good at the needle in the haystack problem.”
Still, the fight cemented the major labels’ hostility toward Pandora. “They were just so adversarial,” the early employee says. “They really thought we were pure evil.”
A competitor with some powerful backers
With royalty rates under control and smartphone ownership accelerating, Pandora’s business was booming. When the company filed to go public in early 2011, it reported $55 million in revenue from the prior fiscal year. By the end of 2015, annual revenue had soared past $1.2 billion, with 81 million people tuning in every month.
But a new rival was gaining ground. Spotify had launched in parts of Europe in 2008 and expanded to the U.S. in 2011. Pandora’s leadership dismissed the threat. “Here comes another try,” Kennedy recalls thinking. “This takes billions of dollars to do. Good luck!”
“[We] believed that the majority of people weren’t going to pay for a monthly subscription, or make playlists, and actively interact with their music,” says the early employee, adding: “That was obviously wrong.”
Spotify had been able to raise hundreds of millions of dollars in venture funding, and was putting that money to good use. “Spotify offered student programs, family programs, and free listening for six months,” says a former Pandora employee.
In addition to deeply discounted rates and generous free trials, Spotify had another ace up its sleeve: Free on-demand listening, which allowed desktop and tablet users to pick and choose individual songs without having to sign up for a paid subscription. By comparison, Pandora’s model—personalized streams with a cap on skips—suddenly felt antiquated.
“Because of our adversarial history, the labels refused to give us free on demand,” says the early employee. “But they were very willing to give that right to Spotify.”
These discrepancies weren’t just based on past animosities: Spotify had given Warner Music, Sony Music, and the Universal Music Group significant equity stakes in exchange for access to their catalogs. The labels had chosen their horse, and Spotify sprinted ahead. By the end of 2015, Spotify counted 91 million monthly users—10 million more than Pandora. A year later, Spotify reached 123 million users per month, while Pandora’s growth flatlined.
Spotify’s rapid ascent quickly became too hard to ignore for Pandora. “A board meeting couldn’t go by without some discussion of Spotify,” recalls Kennedy. But with more than 80 million users who loved Pandora’s existing service, executives were wary of drastic changes. And without Spotify’s labels connections, they felt they were at a significant disadvantage.
“It was a little bit like fighting Spotify with one arm tie behind your back,” says the former employee. “We had no response.”
Part two of this story will publish next week.
No comments