"Recommendation engines" lead fans to music

Andy Sullivan

WASHINGTON: High-school rocker Chris Looney isn't stuck playing music in his parents' garage. The Internet has helped his band has win fans across the United States and a booking agent who's found gigs around Northern Virginia.

"It's really helped out, because if you're ready to have a CD to put out people in California can buy it," said Looney, who plays guitar and drums.

The Internet has for years been touted as a music lovers' paradise, a "celestial jukebox" where any song ever recorded will eventually be available instantly.

While download services like Apple Computer Inc.'s iTunes now offer millions of songs, the central challenge of the music industry -- connecting listeners with music they might enjoy -- remains as daunting as ever.

That could change as a new breed of "recommendation engines" aim to duplicate the experience of a trusted friend saying, "Hey, check these guys out."

Some like MySpace and Soundflavor serve as a sort of Friendster with a beat, allowing musicians to build an audience through relentless online socializing.

Moodlogic and Musicbrainz create detailed "tags," or short descriptions, of songs that may unearth hidden connections between material that may not otherwise appear to have much in common.

Mercora allows users to listen to playlists set up by other users, a sort of cross between Webcasting and "peer to peer" networks like Kazaa. Grouper works like a limited peer to peer network, allowing groups of up to 30 people to share music, photos and other media stored on their hard drives.

Online music stores like RealNetworks Inc.'s Rhapsody and iTunes, meanwhile, encourage their customers to create "playlists" of their favorite music that might interest those who have similar taste.

These technologies could boost music sales across the board by stimulating interest in lesser-known songs that don't get played on TV or the radio, said Sandy Pearlman, a former record producer who helped develop Moodlogic.

"If a recommendation engine can take you from something you love to something you would love if only you knew about it, that would explode the market for music," Pearlman said at a recent music conference in Washington.

There are signs that the approach works.

In a Gartner survey of 1,500 Internet "early adopters" conducted this spring, one in ten said they often buy music based upon the recommendations of others, a figure Gartner analyst Mike McGuire expects to rise.

One in three said they would be interested in recommendation technologies that are powered by consumer tastes, McGuire said.

"The magic question is, how much is this driving incremental sales? We don't know yet," he said.

Customers of the online store eMusic who use its recommendation engines are twice as likely to upgrade to a more expensive subscription plan that allows them to download more music, according to ChoiceStream, the company that developed the engine.

They also stay with the service three times as long as those who don't use the recommendation engines, ChoiceStream said.

eMusic can't rely on a handful of well-known hits to drive sales because it only carries music from independent labels, so it must provide visitors with extra guidance, eMusic CEO David Pakman said in an interview.

While music fans might visit iTunes or Napster to look for one song in particular, eMusic users are more likely to browse for material they haven't heard before, he said.

"If you really want to stretch your tastes ... these kinds of users come to eMusic," he said.

While eMusic relies on sophisticated technology to lead its users to new music, MySpace lets users do the work themselves. Originally catering to individuals who set up their own personal pages, the site now hosts more than 350,000 bands, including many major-label acts, that showcase their music through a built-in media player.

MySpace ranked as the 22nd most popular Web site in July, according to comScore Media Metrix, and that month its parent company, Intermix, was bought by News Corp. for $580 million.