Music product · listening research
A calmer way to find the next song
Most playlists ignore where I actually am. I wanted to design one where time of day and who is nearby count as real inputs, not footnotes.
The exploration kept pulling me between two forces. Suggestions that scale, and curation that has a point of view. The interface had to show both without turning every setting into a spreadsheet.
This write-up is about the narrative, the interaction, and the visual direction. The implementation will change. The trust and the explanations should not.
- Role
- Product thinking, UX, visual design
- Research
- Listening diaries, competitive audit
- Type
- Self-initiated concept
- Phase
- Concept through interactive prototype
If I link a prototype, try it first and then come back for the process notes. They are the part I care about most.
Research and listening diaries
What people say versus what they replay
I ran short diaries that captured where people listened, who held the queue, and when they skipped. Social context beat the taste model more often than I expected.
I skipped the long surveys on purpose. A couple of quick prompts right after a session told me far more than any abstract question about genre ever did.
1. Map contexts
Commute, focus, party, winding down alone. Each context sets different priorities for energy, lyrics, and risk.
2. Find the trust breakers
When someone called a pick random, I traced it back to missing context or a weak explanation.
3. Editorial gaps
I looked for the genres algorithms tend to flatten. Those gaps are exactly where a short curated set earns its place.
Information architecture
A hierarchy that protects focus
I kept the primary navigation small. Now, Collections, Live, and Profile. Everything else stays secondary or shows up in context.
I put "why this track" in the main flow, not buried in a menu. If the system cannot explain a pick in one sentence, I decided the pick was not ready to show.
Visual identity
Dark UI with warm accents and readable density
I tuned the contrast for night listening. The accent color marks live and social features without shouting over the music.
Display type carries the editorial voice. UI type stays quiet and readable so long artist notes never feel like a wall.
Motion and feedback
Micro-interactions that respect the audio
I kept control motion under 200ms so it never fights the beat. Larger motion belongs between sections, not on every tap.
I simplified the waveforms and progress bars to stay light on older phones. Felt speed matters as much as any benchmark.
Outcomes and next steps
What I would ship first on a real roadmap
A production pass would start with honest explanations and a simple playlist handoff. Deeper personalization can wait until those two feel solid.
The open questions are the real ones. Licensing, offline behavior, and how much social data people will trade for a better match.
What did I learn?
- People forgive an uneven pick when the system is honest about guessing.
- Context belongs next to taste, not in the footnotes.
- Pair the algorithm with an editor for anything cultural or time-sensitive.
Notes from diary participants
I kept skipping until the app admitted it was guessing. After that I trusted the next songs more.
I want fewer tabs. If live and on-demand live together, show me how they connect.
- Design
- Laolu James