Spotify Vibe - Thesis
A thesis project reimagining Spotify's Discover Weekly recommendations with vibe-based music discovery. By visualizing user data through charts and matching listeners with similar habits, Spotify Vibe offers personalized song suggestions that capture mood over genre. The project also introduces a feed page to simplify song sharing, allowing users to share tracks with followers or privately via message.
Year
2024
Client
School
Service
Data Visualization, UX, Service Design
Duration
1 year
The Challenge
Music is always a big part of our daily life. Most people wake up to the sound of their favorite song and go to bed listening to their favorite playlist.
Music streaming platforms have become the major means of music consumption, with over 600 million subscribed users and a market worth billions of dollars. This growth not only transforms how users access and enjoy music but also fosters a diverse environment for invention and discovery.
This is where Spotify Vibe comes from. Spotify Vibe is an addition to spotify that allows us to generate a unique data-based visualization that not only allows users to evaluate their own listening habits over time but also provides a better and more accurate recommendation system.
Research
Survey Results
Results of the Preliminary Survey made among 116 participants
Which Spotify feature fo you find most effective for discovering new songs that fit your taste?
116 responses
37.1%
18.1%
14.3%
13.3%
11.2%
5.2%
1.8%
0.9%
Song Radio
Discover Weekly
Playlist enhance
Personalized Playlists
I don’t know
Jam
Release Radar
Daily Mixes
Why do users not find the ‘Discover weekly’ satisfactory when it’s a feature made to just discover songs?
How often fo you use Spotify to discover new music?
116 responses
26.7%
17.2%
10.3%
5.2%
40.5%
Daily
Less than once a week
Rarely or never
Once a week
Several times a week
Would you prefer to have more personalized insights and visualizations on Spotify to better understand your music listening habits and preferences?
116 responses
68.4%
20.5%
11.1%
Yes, I would like more personalized insights
No, I am satisfied with he current level of insight
I am neutral
How accurate do you find Spotify’s Discover Weekly playlist in recommending songs that match your preferences?
116 responses
51 (44%)
28 (24.1%)
3 (2.6%)
3
4
5
3 (2.6%)
31 (26.7%)
1
2
0
20
40
60
How satisfied are you with Spotify’s music recommendations based on your listening history?
116 responses
4 (3.4%)
41 (35.3%)
9 (7.8%)
1
2
4
5
0
20
40
60
Current Methods Used for Music Recommendation Systems
Sonic Profiles
Time frequency representation
Takes into consideration
Key, Tempo, Timbre, Loudness, Time Signature
Natural Language Processing
Word band for each song and artist
Takes into consideration
Song Titles, Playlist Descriptions, Lyrics, Song Reviews, News Articles
Filtering
Collaborative Filtering
listened by both users
listened by one user, recommended to other
similar users
Content Based Filtering
listened by user
recommended to user
similar songs
For Spotify users seeking deeper musical connections and personalized recommendations, there is a need for a feature that unveils their unique listening personalities and facilitates meaningful connections with others who share similar tastes.
Current Spotify recommendations often fail to fully align with user preferences, leading to frustration and a lack of discovery. Users would benefit from a solution that not only visualizes their music consumption habits but also fosters social engagement through shared personality maps and collaborative playlist creation.
User Personas, Empathy Map & User Journey Map
User Journey Map of Sarah
Project Development
Opportunity
2- weekly recommendations not sufficient
3- sharing songs & not recieving feedback
1- profile not personalized
How to give the users some engaging insights without taking away the hype of ‘Wrapped’?
How the New Filtering System Would Work
Sketches and Low Fidelity Prototypes
Design Choices
Selection of Data Visualization
Layout of the Feed Page
A. Only the images of users visible, chronologic order
B. Images, names and sharing time visible, chronologic order
C. Images, names visible, grouped by person, last three shares visible
Flow 1 - Profile Radar Chart Actions
Flow 2 - Artist & Song Overview
Flow 3 - Song Swap / Sharing
Flow 4 - Song Swap / Overview
Spotify Vibe
This feature is named Spotify Vibe because people are seeking more of a ‘Vibe’ when they seek similar songs, not necessarily genre or artist
Jungle
casio
nyc in 1940
ocean eyes
koca bir saçmalık
maria también
something about us
lay all your love on me
Berlioz
Billie Eilish
Jakuzi
Khruangbin
Daft Punk
ABBA
Next Steps
The next steps for the Spotify Vibe project would include validating the concept through user testing, where participants could provide feedback on the usability and appeal of the proposed features.
Additionally, collaborating with developers would be essential to explore the technical feasibility of creating vibe-based charts and integrating them into Spotify’s platform.
Prototyping these features and conducting A/B testing could help measure their impact on user engagement and music discovery.
