Recommendations had to make sense
Users disengaged when suggested artists felt unrelated to their taste. Recommendation quality mattered as much as the interface.

SOUNDCLOUD / GROWTH & RETENTION
A better first experience for SoundCloud listeners, designed to turn an empty feed into meaningful discovery and early engagement.
A/B TEST RESULT
+21%relative lift in Day 0 engagement, measured through likes and follows
THE PROBLEM
New SoundCloud listeners often arrived to a nearly empty Home feed. Without artists to follow, there was little to discover and few reasons to engage. Behavioral data showed that people following fewer than three accounts had substantially lower engagement and retention.
Our challenge was to help new users find relevant creators early, while staying true to SoundCloud's independent music culture.
Better artist suggestions would help people make their first connections, increasing early engagement and giving their feed a stronger starting point.
RESEARCH & LEARNING
We reviewed onboarding patterns from other music services and tested our own concepts remotely with users. Across five rounds of design iteration, we focused on making discovery feel clear, intentional, and useful.
Users disengaged when suggested artists felt unrelated to their taste. Recommendation quality mattered as much as the interface.
Search and genre browsing gave listeners practical ways to find artists beyond a single recommended list.
Testing showed that listeners wanted a way back into the flow, even after finishing onboarding.
DESIGN RESPONSE
Working with product and engineering, I refined the experience around three decisions that addressed both user needs and MVP constraints.
A closer look at the artist discovery experience.






OUTCOME
The A/B test showed that the redesigned experience helped new users interact with artists earlier in their journey.
LOOKING BACK
This project taught me how much small interaction details can matter, especially at the start of a product journey. It also sharpened my approach to working with engineers on recommendation quality and shipping a feasible MVP without losing sight of a longer-term product vision.
Today, I would pair early activation metrics with a closer look at recommendation relevance and sustained engagement after onboarding.
MORE SOUNDCLOUD WORK
Smaller initiatives, but important explorations of how to help listeners make SoundCloud their own.
MVP EXPLORATION
Churn analysis, surveys, and interviews I conducted with a senior researcher surfaced a persistent barrier: people who had spent years curating playlists elsewhere did not want to start over on SoundCloud. We believed a simpler import experience could help new listeners feel invested sooner.
I designed three MVP directions: importing multiple playlists from Home, a focused one-playlist flow from Library, and fetching playlist information upfront to reduce steps. I also explored a future state that could suggest related tracks, find alternatives for unavailable songs, and turn a practical migration task into an opportunity for music discovery.
Working with engineers, I weighed the complexity of building a native importer against the timeline. The team prioritized a third-party path through TuneMyMusic as the more feasible MVP. I left before the final implementation or impact could be confirmed. The lesson was to preserve the longer-term product vision while making a pragmatic first-step decision.






HACKATHON EXPERIMENT
I facilitated a workshop with product, design, and engineering to explore how third-party data might make music recommendations more relevant. One promising direction was to help listeners discover the artists, genres, and live events shaping the music scene in their own city.
During Hackathon Week, I translated the workshop ideas into concepts for a city selector on Home, local SoundCloud Radar pages, and event recommendations informed by listening habits. We considered automatic location access versus manual city selection, along with where this experience would make the most sense: Home, onboarding, or Library.
Early conversations with engineers revealed constraints around location and event data. We narrowed the initial experiment to location-informed music discovery using SoundCloud's existing recommendation engine, leaving richer event connections as a future direction. I left before the test concluded, so I cannot claim a measured outcome. It reinforced the value of bringing technical feasibility into ideation early.






Explore more of my product work
Next project: FORM