Evaluating iTunes Radio and Pandora: Intelligence and inventory will shape the outcome

Media reviews tend to binary considerations — either/or judgments of new products. Accordingly, reviews of iTunes Radio are commonly framed in opposition to Pandora, as in SFGate’s review premise: “Now that iTunes Radio has shipped to millions of iPhone and iPad owners, does that mean you should delete Pandora?”

The single-service presumption is wrong for people who carry a few different listening platforms on their mobile devices, but has an underlying reality. First, it is a natural consumer behavior to settle into one solution and build a habitual connection with it. Second, any streaming service grows in personal value the more it is used, as the user invests time listening, evaluating, creating playlists, sharing socially, and generally building equity in the platform. Loyal use pays off in better experience, especially when a recommendation engine is guiding the music. All such engines become smarter over time, as you thumbs-up and thumbs-down songs and skip tracks.

Platform intelligence is the first of the three i’s which frame the user experience of iTunes Radio, Pandora, Google All Access, and other internet radio providers that algorithmically personalize audio to the listener’s preferences. The Echo Nest is a large solution provider for many platforms. Pandora, of course, has built a proprietary recommendation brain (its Music Genome Project) over many years.

The second of the three i’s is the service interface — the look-and-feel of the desktop and mobile clients in which the user builds a listening home and identity.

Finally there is inventory — the underlying catalog from which programming emanates. Some listeners want Top-40; others plumb an immensely long tail of 30-million tracks.

Accepting for now the premise that iTunes Radio is facing off primarily against Pandora for the lean-back listening market, the intelligence and inventory aspects of each will tell the story. Pandora’s genomic brain has proven out to millions of users as an exceptional listening experience. In a pre-RAIN Summit conversation, Geoff Snyder of Pandora described how algorithmic calculations enhance the intensive human evaluation of song characteristics. The gigantic collection of share, thumbs up/down, and skip metrics amassed by Pandora creates an amorphous cloud of intelligence that envelopes genome considerations and further personalizes an already smart engine.

What does iTunes Radio bring to the table? A recommendation intellect enlightened by the user’s personal music collection, in some cases garnered over ten years of iTunes purchases and imports. For a new service lacking any history of thumb-and-skip actions, that insight represents a meaningful and instant snapshot of user taste.

The inventory issue plays out sharply in comparing iTunes Radio with Pandora. While the exact scope of Apple’s catalog is not disclosed, the company’s direct licensing extends its existing comprehensive label relationships. Pandora, which pays statutory royalties, owns a slow-growing catalog whose every track is hand-qualified. Pandora’s library is fractional compared to Apple’s, and a devoted listener can hear some staleness over a period of time.

Our intensive listening test of custom artist-centered stations running in parallel on iTunes Radio and Pandora revealed more adventurousness in iTunes Radio, as it explored the wider boundaries of its catalog, but also more mistakes. Song skips seemed to throw the stream off-track from personal preferences, as if they were over-weighted. (The testing was located in an iTunes account with zero purchasing history and no local collection.) Pandora provided the usual strong customization, but with selections previously heard in a long-standing station.

Apple will benefit from its ecosystem advantages, as commonly predicted. Perhaps a surprise that iTunes Radio was not surfaced as a distinct icon on the mobile desktop, it nonetheless will probably attract first-time internet radio users. Placing the iTunes download store links against full-track listening could extend iTunes’ legacy business, which might be facing a downtrend as access replaces ownership.

But there is probably not enough intelligence in the Apple system to lure Pandora users who have built years of customization equity in their platform. What Pandora needs to keep the scales balanced is a much fatter catalog — and better sooner than later.

Brad Hill