Is it a wrap for Pandora? Not so fast. Let’s compare the two.
Spotify, the 800-pound gorilla in the music subscription market, recently launched a custom radio service. Even before the e-ink dried on the press release, the immediate reaction was, “what does this mean for Pandora?”
If you’re like me, an introverted, borderline sociopathic geek with lots of free time, then you have a couple of Pandora stations that have been carefully curated over the last few years. Is big giant Spotify gonna swoop in and steal all of Pandora’s customers? I performed a brief, very un-scientific comparison of the two.
On the front end, the two look very similar. Basic playback controls, the standard thumb-up/down buttons. You can build radio stations by selecting an artist or song. Because Pandora is backed by the Music Genome Project (more on that later), a lot more detail is presented on the artists and the songs. Both are ad-supported, though at this time, there appears to be many more ads on Pandora. Pandora has functions to “bookmark” songs to check out later and also links to purchase songs on iTunes. A nice advantage of Spotify Radio is the connection to the paid(or limited free) Spotify subscription service. “Thumbed-up” songs are automatically added to a special playlist within Spotify. You can also link to all available content by that artist inside Spotify.
The real test of these two services is how well they provide a “radio” experience. How well does it provide a reasonable variety of music, based on your choices? How does it aid in the discovery of new music? Here is where Spotify does not (yet) quite measure up in some ways. In my testing, I created a station on each service based on one awesome song: “Award Tour” by A Tribe Called Quest. Despite having a much larger library of songs, Spotify Radio did not offer much variety in the music offering. Within the first 2 hours, it played 3 songs by A Tribe Called Quest (the source artist for the station) and 3 by Outkast. It also played some Kweli and Mos Def, but repeating artists is not exactly what I would call a “radio” experience. Another thing I noticed throughout the first few days was that the music selection leaned toward more released singles and more popular tracks. It did not play any deep catalog cuts or introduce me to any new songs or songs I may not have heard in a while.
Pandora, on the other hand, provided a vastly different experience. From the start, it played a great variety of music that it “thought” I would like, based on “Award Tour.” Within the first few hours, I heard some Lupe Fiasco, Kid Cudi, Slum Village, and some Kweli like the Spotify station. And I was also (re)introduced to some awesome but less popular artists MF Doom, Black Milk, and Zion I. I loved the fact that I did not hear another A Tribe Called Quest song for the first day of the test. There definitely was less track-skipping on Pandora.
Here is the key difference between the two: music discovery. Spotify relies on the “social graph” for discovery. That means that it finds songs that are liked by people who like “Award Tour.” This type of discovery can be great if you have a very large social graph, but it doesn’t seem to be effective in this example. Pandora uses the Music Genome Project, which uses a committee of music pros to create an a kind of DNA map for each song in its 800,000+ library. It contains descriptors like, “east coast rap roots”, “danceable drum patterns”, and “rhythmically complex rapping.” Pandora uses these to connect similar songs, i.e., you will probably like other songs with “east coast rap roots” and “rhythmically complex rapping.” Advantage: Pandora.
Granted, Spotify Radio is very new, so they should be given time to catch up. They still have a larger library of songs (15 million vs Pandora’s 800k), they just need to find a better system utilize it. For now, I’m sticking with Pandora.
charliedigital is guest contributor to MusicandModeling.com. Find his random tweets and re-tweets on hip-hop, black culture, and technology @charliedigital