Overview
A faster connection between viewers and creators
Conceptual UX project. I redesigned the way users discover content on YouTube, using data the platform already has but doesn't use for discovery.
The problem
Deciding what to watch: choice overload
Every day, over 20 million videos get uploaded to YouTube. To pick one, users have almost nothing to go on: a thumbnail, a title, a few seconds of doubt. The decision comes before the enjoyment — creating friction at the exact moment that matters most: the first one.
Key insight
"Most Replayed" for Discovery
YouTube already knows exactly where each video hooks people: the peaks on the progress bar, "most replayed." That data tells the platform which moment of a video makes people stay. Today it's only used inside the video you've already chosen. What if it were used to help you choose?
With Gemini now integrated into YouTube's recommendation engine, this idea is even more viable: multimodal analysis already identifies, frame by frame, what makes a video hook viewers — which would directly reinforce automatic selection of these moments.
The solution
A feed that starts playing before you decide
When you open the app, the player starts automatically with content from your creators — not the full video, but a preview built from their highest-retention moments. If it hooks you, you keep watching. If not, it automatically moves to the next one.
Go to appWho it's for
Two sides, one shared problem
YouTube Compass connects both sides faster: less friction to find, more chance to be found.
They waste time deciding what to watch, and sometimes don't even know what they're looking for.
They compete for attention among more than 3 million active channels in the YouTube Partner Program.
In process...
Things I am working on
A hypothesis, not an answer. YouTube Compass starts from a simple idea: the data to improve discovery already exists inside YouTube. What's missing is connecting it. A project I'd love to test with real platform data.
-
Ads/revenue impact — What happens to the monetization model if the player starts with "trailers" instead of the full video (fewer pre-roll impressions, for example).
-
Relationship to Shorts — This doesn't replace or directly compete with Shorts.
-
What this project doesn't solve — A conceptual project, with real limits: no real data, no validation against the real recommendation engine.
-
API access — YouTube's API doesn't expose the data I'd need to actually test Compass.
Real time comments
Reactions, at the moment they happened. Jump straight to what people said when it mattered. Comments tied to the video's key moments, not buried at the bottom.
Go to app arrow_forwardOne topic, endless content
What if your interests became their own channel? Collections group everything you care about into themed cards built just for you.
Go to app arrow_forward