Spotify Feature Addition
Case Study
Spotify offers several AI-driven features such as the annual Spotify Wrapped feature which provides users with insights into their listening habits for the past year. This design expands on the idea of the Spotify Wrapped feature and offers users with more frequent and playlist-specific listening insights allowing users to track their habits and curate their music libraries more effectively.
Project Overview
Project Goals
For this project I needed to propose a feature addition for the audio streaming platform Spotify. This allowed me to work on my UX/UI design skills, UX research skills, and exercised my ability to produce a user-friendly design whilst continuing to work within a companies design guidelines.
Explore what AI features Spotify users currently find useful.
Work within the limits of Spotify’s current brand design.
Design a useful feature for users rooted in AI and music discovery.
Process
Research
Research Plan
When beginning this project, I had several ideas I wanted to explore relating to artificial intelligence and music discovery. With this in mind I created my research goals around these concepts and let my user interviews and secondary research decide which idea to go with.
Research Goals
● Explore ways to assist music listeners in music discovery
● Better understand the role AI can play in an individual's music experience.
● Research what the experience of discovering new music is like for different users.
● Discuss user’s feelings surrounding AI and its convergence with music discovery.
With these goals in mind I began to conduct secondary research to get a better idea of the AI features of Spotify and other third-party companies.
Secondary Research
Spotify and AI
Spotify already has some AI integration mostly based in algorithms. These features include Discover Weekly, Spotify wrapped, Spotify DJ and some curated playlists which are tailored to individual users' music taste.
Spotify has made several investments in AI over the past few years such as the purchase of “Sonatic”, an AI startup that converts text into a realistic speech format. It also held its first “Machine Learning Day” back in 2018. Since then Spotify has continued to make additions to its platform using AI.
Third-Party Platforms
Maroofy is a platform that utilizes AI to find songs similar to others. Users are able to search for songs they already like, and Maroofy will provide other songs it thinks the user may enjoy.
Moodify is the only third-party app listed here with integration to Spotify’s platform. Users are able to sign in using their Spotify account information and find music similar in mood to the current track they are listening to.
Imfeeling is a platform that allows users to type in a specific feeling and uses AI to match users to songs related to that emotion.
Gnod offers two tools on its website for music discovery. Both of these tools utilize user input to provide suggestions. One tool uses three artists to provide recommendations related to those three while the other uses one artist to build the user a “Music Map”
User Research
Empathize
User Personas
With research complete I wanted to try to identify more with potential users of the feature and put my research into a humanistic perspective. To do this I began by creating two fictional user personas, both relating to the potential needs a user might have.
Pov’s & HMW’s
User Flows & Task Flows
After working on my HMW's & POV's it was starting to become clear which direction this project should take. Still, I went ahead and created two separate User Flows and Task Flows. One was for the playlist cleanup feature while the other was for a general AI feature.
After revising these flows several times, it really became obvious which idea was the most cohesive and would be most useful. The playlist cleanup feature made the most sense for the path this project needed to take.
Design
Mood Board
With my path on this project more clear, I finally felt it was time to begin to develop ideas around what this feature’s design might look and feel like. I first started by doing research into Spotify’s design and branding.
Once I felt I had a good understanding of Spotify’s design and branding, I took a step back and tried to get a better understanding of the design of AI features on other platforms. I took all of these resources and put them together on a sort of mood board to reference back to while sketching the playlist cleanup icon and screens.
Sketching
Once I had an idea of the icons related to AI, I began to sketch the icon of the playlist cleanup feature. Between these sketches and the mood board I had put together. I felt ready to start sketching out screens.
Role:
UX/UI Design, UX Research
Timeline:
8 Weeks
Tools:
Figma, Google Docs
Many of the insights I got from the user interviews mostly revolved around users stating they did not feel they had enough input into the Spotify AI features and so therefore the music recommendations they were getting were rarely useful.
One user brought up a particularly good point in regards to this and library curation that played a large role in the direction this project was going to take. This user noted that some sort of Spotify Wrapped but for playlists might allow a user to better curate their library and also receive more accurate song recommendations.
With that, I began to include that idea amongst the other AI features I had come up with and started to move onto the next phase of this project.
Usability Testing
Testing Results
Iterations
Want to see more projects?
With the first task, 2 out of 4 users took longer than expected to locate the “Playlist Cleanup” icon. Given this is an unknown feature, it is understandable for users to show slight hesitation with the feature and its corresponding icon. With that being said, it never hurts to take another look at visibility to be sure the icon is noticeable.
The second task revealed no issues. All users were successfully able to navigate to the “similar songs” section of the playlist cleanup in a standard amount of time.
The third task proved to be the most problematic task throughout all of the usability tests although 2 out of the 4 tests were particularly difficult. There is an overall issue with consistency between the “find similar songs” and “review skipped songs” sub features in the design. Users are able to view the overall list of the recommended songs whereas the “Skipped Songs” takes a user directly to the song often skipped. This inconsistency caused users to feel there was something they were missing with the third task.
In terms of iterations that should be made, either another screen needs to be added to the design or one needs to be removed for a better sense of consistency between the “find similar songs” and “review skipped songs” sub-features. For improved consistency throughout all of the application, it would be most ideal for another screen to be designed. Given the time constraints of this project, further wireframing may be difficult but ultimately I believe this iteration should take top priority. This new screen should be similar to the screen that users initially see when navigating to the “find similar songs” section of the feature. Another revision to consider relates to the visibility of the playlist cleanup icon. It would be ultimately more beneficial to the user and overall design to spend extra time ensuring this icon is just as visible as the other icons next to it.
After doing several projects where I had the opportunity to determine the overall design guidelines for the product, it was nice to finally work on a project where these guidelines were set for me. This project allowed me to work heavily on my design skills, specifically the skill of working within a preexisting design. Another skill I was able to work on during this project was my ability to build user and task flows. Throughout the User/Task flow step of the process, I found myself gaining a better understanding of the value of good user and task flows. For this project, my user and task flows ultimately determined the direction that the project would be going in. If I had more time, I would like to retest these new screens on more users. I would also like to spend more time exploring other data and sub-features that the playlist cleanup feature could provide to the user.
During user research, I aimed for my interviews to feel more like a conversation. I sat down with four participants in total and gained invaluable information regarding listening habits, AI, music discovery, and more. I then took these results and transferred each insight into Figjam to build an Affinity Map. This Affinity Map gave me a good idea of any repetition between insights or ideas to be pursued from the user interviews.
In addition to building these personas, I began forming HMW questions and POV statements to help me get a grasp on how this project could best serve users.
Prototype and Testing
With the results, I designed another screen for the “skipped songs” feature. I also adjusted the playlist cleanup icon several times to try to improve visibility, but ultimately decided to leave the icon as it was because any improvement to visibility made it stand out poorly compared to the icons next to it.