How would you create a playlist for a user after they listen to a single song? (in an app like Spotify)
Google- Eva Richardson
Clarification
Going to assume we are Spotify and that this is an individual consumer account and not an account used by a business like a restaurant or a bar. The user could be listening from their computer, our mobile app, or their car which may be helpful information but our task at hand is the same.
Goal
Before we dive into the problem, let’s first take a step back and think about why are we doing this and what are we hoping to accomplish. Is there some sort of strategic initiative we need to align with or was there a specific goal in mind?
If the interviewer doesn’t have anything planned we can devise this ourselves by examining Spotify’s mission statement which is to unlock human creativity and give artists the ability live off of their art.
By making recommendations we help users listen to more music and get the creative juices flowing while simultaneously paying the artists for the ability to play their music.
If we can increase the amount of time users spend listening to music we can further our mission statement so let’s focus on that as a goal.
User Segments
It’s hard to come up with a magic bullet that solves everything for everyone so one of the ways I like to narrow scope is by looking at different user segments and choosing one to prioritize:
- Musicians – In addition to playing music they also listen to it obviously. Expert level understanding of music and a better sense of what is “good” music.
- Top 20 Fans – Generally like catchy songs and different genres, whatever the flavor of the month is
- Fan Boy / Girl – Zealous followers of an individual artist or group, aka Swifties or BTS Fans
- Genre Experts – Individuals who are really passionate about a specific genre of music, for example hip hop heads
Out of the above user segments, I’m going to suggest we focus on the genre experts. This particular user segment offers a solid combination in total number of users and time spent listening to music per user which will give us a sizable opportunity to increase the amount of time users spend listening to music on Spotify which is our goal.
Pain Points
Now that we’ve chosen the user group we’re going to focus on, Genre Experts, let’s go ahead think through what might be some of their pain points or areas of frustration with our intial song recommendations:
- The song is not in their genre of preference
- The song is way too overplayed / they’re sick of it
- The song doesn’t match their current setting, listening to music at the gym is very different from the library
Solutions
With the above pain points in mind, let’s brainstorm some solutions that might address or alleviate those problems:
- Preliminary Music Taste Survey – As part of the onboarding process users could optionally fill out a survey regarding their music tastes and preferences which we would then use to make our initial song recommendations.
- Import Listening History – We could let users import their listening history from other platforms like Pandora and Apple Music which would then give us a very good sense of what they would like to listen to.
- Friend Music Preferences Informed Algorithm – When users sign up, they could upload their contacts and mark which of their friends with spotify accounts they think have good taste in music.
- Grant Access to User Personal Data – After signing up, users could give us permission to access some of their data from other apps like Facebook and we could make informed music recommendations based off of that. For example, if we saw that you liked the Metallica FB Page it would be a pretty strong signal to put Metallica songs in your play list.
Prioritization
In all likelihood we won’t have the time nor bandwidth to build out all four solutions in parallel so we should choose one to prioritize. To help guide that conversation I’m going to use a comparison matrix:
Ease of Implementation, Impact on goal
1. A, A -> Would be easy to implement and would definitely impact our goal, only concern would be how likely are users to fill this out.
2. B-, A+ -> 100% would make our recommendations better, but I can imagine importing music from other platforms being hairy. I doubt they would want to make that easy for us to do.
3. B, B- -> Would be a cool feature, but people might be wary of granting access to contacts and I’m not sure how closely an individuals music preferences are going to closely mirror their friends.
4. C, A- -> I imagine people would be skeptical of granting us more access to their data and other apps and some other platforms may be tricky to work with. Perhaps we could circumvent this by working with a third party data provider.
Based off of the above, I’m going to suggest we prioritize building out the Preliminary Music Taste Survey for new users.
Summary
In order to increase the amount of music Genre Experts new to the platform listen to, we are going to offer users a Prelmininary Music Taste Survey during the onboarding process. This will ensure that we don’t recommend a song to the user in a genre they don’t like.