"We're like, 'Oh,' " Westergren recalls. " 'This thing doesn't work.' "
But it did work. The more they tinkered, the more they believed. Even the Beatles-Bee Gees mashup that made so little sense at first seemed to compute because this wasn't the later, saccharine Bee Gees, but the earlier, bluesier Bee Gees.
"Our big challenge was the lexicon; literally 'the Word,' " says Westergren, who rattles off phrases such as "compositional dominance." "You felt a breakthrough moment when you found the word. Then you forgot the time before you had the words."
They settled on five genomes — pop/rock, hip-hop, jazz, world and classical. Their office in San Francisco, and later in Oakland, swelled with hip-hop rhymers, classical music performers and jazz guys who spent days listening to music and analyzing each song based on more than 400 characteristics.
The project was thriving, the business wasn't. Pandora's leadership "originally thought raising money was going to be much easier," Gasser recalls. A possible investor sneered, "So you've basically got the 'no-model' business model," Westergren recalls.
Westergren maxed out 11 credit cards, taking on more than $300,000 in personal debt. Stress sent him to the hospital with heart palpitations.
The company's finances were so dire that Westergren had to stop paying staff, asking them to defer salaries, which he didn't realize was against California labor law. "When you aren't paying everybody on a monthly basis, yeah, it can cause some tension," Gasser recalls.
"Pretty much everybody took off," says Steve Hogan, who now oversees 25 music analysts as head of music operations. At the nadir, Hogan was the only staff analyst left.
In 2004, an infusion of venture capital saved the company — the result of what Westergren estimates was at least his 349th pitch. But to keep his company alive, Westergren has had to sign most of it away to financiers. He now owns only about 3 percent of the stock.