Last.fm
I worked as a designer and led design at Last.fm for six years. During that time, the projects I worked on changed rapidly, we scaled, I product managed many featured to market. This case study is about Mix Radio.
What's "new" music?
The challenge & opportunity:
Create a radio station of “new” music to listen to, based on your taste
My role:
Design, product
Curiosity & hypothesis
We kept hearing from users that they needed new music to listen to, and that discovery is difficult. This is true, and especially before recommendations went mainstream. But, as I did UXR, I discovered they don't want a barrage of all new music, and that the definition of “new” is nuanced, sometimes it means old things you haven’t heard in a while.
The radio was one of the more difficult and brittle parts of the platform to work on, and it was difficult to sell in this idea when the project was already out of the gate, but we though that a more nuanced mixed station would better meet the goal of total minutes per session.
Exploration & Plan
Research: We observed users listening to different kinds of radio stations: all music they knew, all new music, and a mixture of the two. When we learned that the mix was more satisfying, then we had to figure out what the algorithm was going to be.
Creating the station: recency was a big factor for creating this. Did you listen to that song last week or last year? The last year song might be “new” again. Mixing in the right about of never heard before content was less than we originally expected, in fact – the mix ended up resembling more similar to the ratio of known to new that terrestrial radio stations play which didn’t align with the company’s disruptive ethos.
The data: This one was impossible to argue with the data. Renaming radio stations was hardcoded into the system (which needed to be translated into 8 languages), so it wasn’t straightforward to make or name a new station (when the name changed due to our findings). I saw it through that the station went live, and later we integrated it into our new iPhone app, which I worked on the second iteration of, as well as Last.fm’s XBox Kinect app, which I also designed.
XBox Kinect: This echoed the problems I’d just been through with the new mix radio station. When we had the hardware and prototype game delivered to play with and QA after working with XBox on the designs, I was positive nobody was going to hold their hands up to play music. It was too hard! I got tired just testing it. The Kinect had early voice commands though, which was really exciting. This seemed like the much bigger opportunity, and I worked with XBox to change the plan to have voice commands for playing, skipping tracks and pausing radio
Outcome
It worked! Mix Radio became one of our most popular radio types.
The iPhone app felt magical (at the time) to listen to music from your phone in the cloud. The Kinect app was really cool, so much so there was even a Diesel Sweeties comic.
