SoundCloud UX Research
Research86% expected it. 8% could do it.
Five of us spent a quarter picking apart SoundCloud's desktop site for a grad research course - heuristic audit, competitive analysis, tree testing, moderated sessions, a full A/B test. The goal wasn't to list what felt clunky. It was to find exactly where the interface stops matching how people actually think, then prove our fixes actually worked before we recommended them.
Role
UX Researcher - contributed across all 5 phases (team of 5)
Timeline
10 weeks · UC Irvine, HCI & Design
Impact
+45% task success, validated by A/B test
Overview
This started as a class project - five of us picked SoundCloud for a research methods course at UC Irvine, no connection to the company itself. We wanted to go past "this feels a little clunky" and actually pin down where the site stops matching how people think, then prove whatever we recommended would actually fix it.
So we didn't stop at opinions. We ran a full heuristic audit and a competitive analysis, then backed it up with three separate rounds of real research - tree testing, moderated sessions where people thought out loud while using the site, and a quantitative A/B test. If we were going to tell SoundCloud to change something, we wanted a number behind it, not a vibe.
How we got there
Heuristic analysis
We went through every core flow - browsing, search, library, playlists, comments - and checked it against Nielsen's 10 usability heuristics. That gave us 17 findings, each tagged Major, Moderate, or Minor depending on how much it actually got in someone's way.
Competitive analysis
Then we looked at how everyone else handles the same problems - Spotify and Apple Music directly, plus Bandcamp, YouTube, and TikTok for how they get people discovering music and actually connecting with artists.
Tree testing
51 people navigated a text-only version of the site - just labels and hierarchy, no visuals at all - so we could isolate whether the structure itself matched what people expected, before design ever entered the picture.
Moderated usability testing
6 people worked through real tasks on the live site while talking through their thinking out loud. This is where you actually learn why something went wrong, not just that it did.
Quantitative A/B test
Finally, 85 people split between the current site and our redesigned prototype, timed and scored on the same four tasks - so we'd know for sure whether what we were proposing actually worked.
The clearest signal: playlists
One result from the tree test jumped out at us right away. 86% of people expected they could create a playlist straight from the Playlists page. Only 8% actually managed it - because the real way to do it is buried three clicks deep, inside the ellipsis menu on one specific song.
That's not something a tooltip fixes. People are walking in with a completely reasonable idea of how this should work, and the site just doesn't offer it.
Tree test, Task 1: creating a playlist. 86% expected a path from the Playlists page itself; only 8% found the real one.
What people said while doing it
Six people talked us through the same core tasks live on soundcloud.com. Three things kept coming up, over and over:
- Search was basically how people navigated the whole site - but it matched on literal words in the title, not on what type of thing you were even looking for. Typing "ed sheeran album" pulled up songs with "album" in the title before it ever pulled up an actual album.
- Playlist creation was rough on people - 18 negative reactions against 11 positive ones in this task alone. One person basically gave up mid-task: "There's nowhere obvious here for me to actually add a playlist."
- Comments and following an artist both felt kind of pointless to people. One participant figured artists get too many comments to ever notice one more; a few others just weren't sure what following someone would actually change for them.
Two very different people, same walls
We built two personas out of the research, mostly to keep ourselves honest about who we were actually designing for. Alex is new to the platform, still figuring out their own music taste and a little self-conscious about it. Joe's the opposite - he's been on SoundCloud for years, has over a thousand songs liked, and wants to start DJing, except he can't find anything in his own library anymore.
They're about as different as two listeners can be, and they ran into the exact same walls. That's usually a sign the problem isn't about being new to the site - it's built into the product itself.
A/B test results (85 participants)
We ran the same four tasks twice: once against the live site, once against our redesigned prototype.
| Task | Success | Time | Ease |
|---|---|---|---|
| Find & listen to an album | — | 74s → 58s | 22% faster |
| Adjust the volume | — | 36s → 19s | 26% fewer misclicks |
| Comment on a track | 49% → 63% confidence | — | No lift in perceived engagement |
| Create a playlist | +45% success | 2.3x faster | 2.2x more direct |
Volume and playlist creation were the clear wins, and both just needed something that already existed to be more visible. Commenting told a different story - the fix worked exactly as intended, people just didn't end up caring that much either way.
A single Create button added to the Playlists tab lifted task success 45% and cut completion time to under half.
What we'd fix first
We sorted everything by how much pain it actually removed versus how much work it'd take to ship, so nobody's left guessing where to start:
- Highest priority: add a real Create Playlist button to the Playlists tab - not buried three clicks deep in a song's menu.
- Also urgent: redesign the player bar so it's actually visible - bigger, higher contrast, with the volume slider showing by default instead of hiding behind a hover.
- Next up: rank search results by relevance and content type, not just whatever literally matches the title.
- Next up: give songs, albums, and playlists their own distinct icons, and fix the liked-heart so filled versus outline means the same thing everywhere on the site.
- Worth a look eventually: commenting alone doesn't feel like connecting with an artist to most people - social features need to go further than a text box.
- Worth a look eventually: more content variety - live sessions, video - would help SoundCloud compete for a younger audience against TikTok and YouTube.
What this was actually about
What sticks with me about this project is that the two biggest wins didn't require building anything new. The ability to make a playlist, the volume slider - both already existed. We just moved them to where people were actually looking for them. It's a good reminder that a lot of what feels like a "usability problem" is really just a gap between what's in someone's head and what's on the screen - and the only way I trust that a fix actually closes that gap is by timing a stranger doing the task, not by deciding for myself that it feels better.