Reimagining
Search
Reducing decision fatigue on a streaming platform
through mood-based discovery
Ever spent 30 minutes
choosing what to watch...
then watched nothing?
Decision fatigue is killing streaming engagement
Users are overwhelmed by choice, endlessly scrolling through content libraries without committing. The search experience was a dead end: cluttered, passive, and unhelpful.
of users browse for 10+ minutes
before selecting content
Internal analytics, Q3 2024
What we discovered
Search Landing
The primary entry point for search and the central discovery hub for all content.
Categories and content competed for attention. Users didn't know where to start. The experience was passive, not guiding.
Search Input
Once a user taps search, they're met with a blank canvas and their own search history.
Just a blinking cursor and old queries. No guidance, no inspiration. Users had to already know what they wanted.
Search Results
After typing a query, users see a flat list of matching content with minimal context.
Results were flat listings with no confidence signal. Users couldn't tell which result was best, so they scrolled endlessly.
What if we asked
how you feel,
not what you want?
Refining the experience
Reorganized by context
Sorted by language, sports, moods. Cleaner hierarchy, but the core interaction remained browse-and-hope. Users still lacked a reason to commit.
Mood-first discovery
"What do you feel like watching?" became the opening question. Paired with a confidence card that surfaces one best match, users felt understood and decided faster.
The complete experience
Mood discovery
Mood-based chips replace the blank search canvas. Users tap how they feel instead of guessing what to type.
Confidence card
A high-confidence match rises above generic results, giving users a clear recommendation with one-tap access.
Episodic discovery
Once mood matching gets users watching, episodic discovery keeps them watching. Related episodes and clips surface automatically for returning viewers.
I finally stopped doom-scrolling.Beta tester, usability study
I just pick a mood and it gets me.
What I'd track
Time to Content
The core problem was browsing fatigue. Measuring median time from search initiation to play would validate whether mood-based filters actually reduce the decision burden.
Browse-to-Play Conversion
What percentage of search sessions end with the user selecting content? A higher conversion rate means the discovery flow is guiding, not just displaying.
Search Refinement Rate
How often users modify their initial query or mood selection. Lower refinement suggests the first set of results was relevant enough to act on.
Return Search Usage
What percentage of returning users engage with mood-based search again? Repeat usage separates novelty from genuine utility.
This project was validated through usability testing but did not ship to production.
What I learned
The biggest shift wasn't in the interface. It was in the question we asked. Reframing search from "what do you want?" to "how do you feel?" changed everything downstream.
What surprised me was how much work went into edge cases: what happens when a mood has sparse content in a regional language? We built fallback logic that gracefully widens the results rather than showing an empty state. It's invisible when it works, but it took more iteration than the mood chips themselves.
If I did this again, I'd push for a longer testing period. 18 participants over 2 weeks gave us directional confidence, but I'd want to see how mood-based behavior changes once the novelty wears off.