Industry
Music Streaming
Duration
AUG-SEP 2024
Participation
Research, UX, UI(Sole Project)
Status
Case Study
Providing an Experience to Discover New Music Beyond Existing Preferences
Match your moment with the right music and uncover new favorites.
This project enhances Spotify’s user experience by addressing limitations in discovering music beyond existing preferences. Through surveys and interviews, user needs for broader musical exploration based on moods were identified. A solution leveraging Spotify’s audio analysis technology enables personalized playlists, expanding users’ listening horizons.
Background
Spotify is loved for its precise recommendation algorithm, yet it remains challenging to discover music outside users' existing preferences.
Action
Conducted surveys and interviews to understand users' desire for broadening their musical taste and the key factors they consider when exploring new music.
Result
Proposed a custom playlist feature using Spotify’s audio analysis technology, enabling users to discover new music through mood and element combinations.
Jump to Prototype
Challenge
Spotify excels in algorithmic recommendations but falls short in strategies for expanding user preferences.
Service Analysis
Spotify is loved for its powerful algorithmic recommendations combined with advanced audio analysis technology. (*Insights of algorithm-based music recommendation features, based on interviews with 4 long-term Spotify users.)
Daily Mix
Features
Strengths
Limitations
Playlists based on user's most listened-to artist & genre
Ideal for comfortably enjoying familiar songs
Limited in discovering new music
Release Radar
Features
Strengths
Limitations
Suggests new tracks from followed artists and preferences
Quickly identifies new releases from favorite artists
Too many irrelevant tracks make selection hard
Discover Weekly
Features
Strengths
Limitations
Weekly playlist matches user preferences
Useful for exploring new songs that match taste
Unclear criteria lead to reduced accuracy
Artist Radio
Features
Strengths
Limitations
Suggests music similar to a selected artist
Helps discover tracks from similar artists
Limited to genre and artist-specific characteristics
Up Next
Features
Strengths
Limitations
Automatically plays similar songs after a playlist ends
Highly accurate in recommending similar tracks
Only applicable when a curated playlist already exists
Competitive Analysis
However, Spotify faces clear limitations in strategies for 'taste expansion' compared to other platforms.Comparison of Service Strategies Among the Top 3 Music Platforms.
Platform
Key Strategy
Powerful recommendation algorithm supported by user and music analysis data
Editor-curated playlists, exclusive content, and high-quality audio
User-generated content and video-based library with extensive resources
Personalization
Continuous personalization through curated playlists such as Discover Weekly and Daily Mix
Recommendations based on editor-curated content tailored to user preferences
Personalized recommendations based on search and watch history
Taste Expansion
New music suggestions via Discover Weekly, but heavily reliant on past listening data
Trendy songs and new content naturally discovered through editor-curated recommendations
Flexible exploration of familiar and new music using Familiar, Blend, and Discover modes
User Research
Users seek music discovery beyond repetition, focusing on moods and detailed filtering.
Through surveys and interviews with long-term users, we identified key pain points and opportunities for music discovery.
Exploring Beyond Preferences: Mood-Based Needs
Users feel fatigued by repetitive recommendations and want music tailored to their moods and situations.
Interview
Users feel fatigued by repetitive recommendations, even if they are accurate.
"I need something fresh, so I refer to other users' playlists."
"It's nice, but all the songs are ones I've already heard."
"I don’t click on the recommended list because I’m tired of similar suggestions."
Preferred songs and songs users want to listen to in the moment can differ.
"I want music that perfectly matches a specific moment."
"It’d be great if weather or time were reflected."
"To find music that suits my mood, I have to search manually."
Survey
70% of users use user-created playlists outside of algorithmic suggestions as one way to explore music.
And, this is because they can select keywords, emotions, or situations that match the context of their listening moment.
<
Do you often use other people's playlists when exploring music?
(168 Respondents)
Reasons for Using Other People’s Playlists
(117 Respondents, multiple choices allowed)
53.8% (63)
34.2% (40)
32.5% (38)
29.1% (34)
17.1% (20)
14.5% (17)
3.4% (4)
5.1% (6)
1
To choose keywords, emotions, or situations for each moment
2
Becauses songs are unified by emotions / themes
3
To break away from algorithm and discover songs
4
To experience other people’s tastes
5
Because each playlist has a interesting topic
6
Because overall quality of playlist is high
7
Not knowing how to use another playlists
8
etc
How Users Discover Music: Genre vs. Sound Details
In music exploration, users prioritize 'genre' as a key factor, while some also focus on specific song features like tempo and instrumentation.
Survey + Interview
Survey results show that users tend to consider 'Genre' as an important criterion when exploring music.
>
The ranking of keywords used by people
when searching for playlists (123 respondents, multiple choices allowed)
However, existing algorithms often suggest too broad or irrelevant genres.
"Genres I have no interest in sometimes appear in recommendations."
"I only listened to Taemin, but now all K-pop keeps getting recommended."
"Recommendations mix songs in different languages without clear reasoning."
"I prefer EDM with lyrics, but it keeps suggesting instrumental tracks."
Meanwhile, some users expressed interest in filtering by song features rather than genres.
Focus
Genre
Individual Song
Features
Beat/rhythm style
Cultural/historical background
Production style
Tempo
Key
Instruments
Rhythm
Vocals
Lyrics
Example
Hip-hop, K-pop, American country, Latin pop, Afrobeat
A fast-tempo, bright-key song with upbeat rhythm and guitar focus
Usage
Facilitates broad classification and easier exploration
Allows detailed exploration tailored to mood or context
Solution
Utilizing Spotify’s audio analysis technology to build a mood and song trait filtering system.
Utilizing Spotify's Audio Feature Data
Spotify leverages its world-class audio analysis technology, Echo Nest API, to precisely analyze the detailed characteristics of tracks, forming the foundation for personalized recommendations.
Tempo
The speed of a track, measured in beats per minute (BPM).
Danceability
How suitable a track is for dancing based on rhythm and beat stability.
Valence
Reflects the emotional positivity of a track (e.g. cheerful = high valence).
Acousticness
Likelihood of a track being acoustic (e.g. higher values = more acoustic).
Explore Songs Based on the Mood of the Moment
Combine keywords like weather, time, and context to find the perfect mood for the moment. Spotify’s audio and lyrics analysis system ensures consistent mood categorization across tracks. This provides a fresh way to discover music beyond similar genres and artists typically recommended by existing algorithms.
Filtering by Genre and Song Features
For users who want more control, additional filters allow adjustments by genre and specific song traits, complementing the mood-based system. These filters can be skipped if desired, letting the system recommend tracks aligned with the selected mood and the user’s existing preferences.
Create Playlists with Daily Updates
Like Spotify’s other playlists, the custom mix updates daily with new tracks. Songs users mark as favorites are fed back into the algorithm, helping Spotify refine and expand its understanding of their preferences over time.
Key Features
sunny
rainy
cloudy
Snowy
windy
stormy
sunny
rainy
cloudy
snowy
windy
stormy
spring
summer
autumn
Dawn
Morning
Afternoon
Evening
Night
spring
summer
autumn
Dawn
Morning
Afternoon
Evening
Night
Romantic
Focus
Workout
daily
Relax
festival
Party
Cinematic
Select Mood
Custom Mix
Next
#1
Explore Songs Based on the Mood of the Moment
#2
Genre and Song Features Filtering (Optional)
Explore various genres by language and view description with artist lists.
*If skipped, recommendations will be based on user's existing preferred genres.
Filters individual song features using Spotify's audio analysis technology.
Original Data
Renamed Title
Tempo
→
Tempo
Very Slow ~ Very Fast
Valence
→
Brightness
Dark ~ Bright
Acousticness
→
Sound Texture
Acoustic ~ Electronic
Danceability
→
Dance Vibe
Low ~ High
Great job!
Your playlist is ready!
User can always check and set the presets again.
#3
Create Playlists with Daily Updates
Like other Spotify playlists, Custom Mix updates its tracks daily.
After the Project
As a passionate Spotify user, I loved the platform but often felt something was missing. Through this project, I turned those frustrations into actionable insights by validating them with others and uncovering clear design opportunities. I also learned to align solutions with brand and technology, ensuring they were meaningful to both users and the product’s core strengths.
