Another sign that recommenders are perfectly positioned to be the next “big thing.” AT&T has tied up with Pandora’s Internet radio, allowing users to create up to 100 customised radio stations. The service – which will cost AT&T subscribers $8.99 a month – will also let users buy tracks they like (thanks to the recommendations, no doubt) through the AT&T music store.Pandora’s Music Genome Project recommends tracks and artists to users by linking their current listening tastes to new songs based on melody, harmony, rhythm, instrumentation, arrangement, lyrics and vocals similarities. This recommendation functionality – based on the Amazon model – has the potential to revolutionise the mobile music market, delivering new content to specific user profiles quickly as well as avoiding cumbersome search techniques.

The practicalities of a pricing strategy where the user has to pay to listen and then pay to own individual tracks isn’t clear, but the fit between music and recommendation certainly is. (Consider the recent tie-up between Orange and Xiam to deliver music recommendations and you’ll see where this is going.)

Peggy adds: Mobile search has a number of shortcomings (all well documented in posts and analysis on this site). Most of the problem has to do with poor usability (limited screen real estate, no room of patience with a long list of links…the list goes on.) Take a closer look and it’s plausible that the real problem is the paradigm. 

Mobile search is built on the premise that users know what they want and are prepared to go look for it. That’s quite an assumption when it comes to fast-paced mobile such as entertainment and multimedia which changes faster than users can keep up. And it gets really tough if you imagine a user looking for some down-tempo tracks, for example. Mobile search only works if users know the artist/track, so searching for down-tempo is futile.

AT&T joins what will surely be a slew of mobile operators harnessing recommenders to cross-sell and up-sell individual users to more content based on their purchases, passions and past click-behavior. It’s bound to be even more compelling if the technology can learn users’ likes and dislikes over time to dynamically and consistently deliver the right content mix.

This entry was posted on Monday, November 12th, 2007 at 10:01 am and is filed under Content Discovery, Mobile Music, Mobile Search, Recommendation. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.