U-NEXT is Japan's #1 streaming service — a catalog of over 30,000 pieces of content spanning series, dramas, movies, and books. This project redesigned the autosuggest search experience across Web, iOS, and Android, transforming a basic text-matching utility into a smart, ML-powered content discovery surface.
A 5-year roadmap blocker
What looked like a simple search improvement turned out to be tangled in conflicting business expectations, engineering constraints, political friction between teams, and user frustrations all pointing in different directions. A full rebuild would take years — maybe five. I saw a different path: what if we worked within the existing tech stack and focused on high-impact improvements first?
Research and cross-platform design
Conducted desktop research benchmarking autosuggest vs. autocomplete patterns across streaming and e-commerce. Facilitated ideation workshops with core and extended teams. Designed cross-platform explorations for Mobile, Web, Responsive, and early TV concepts. Created paper sketches and Figma prototypes, and documented handoffs for all platforms.
5 years → 4 months
Compressed 5 years of roadmap inertia into 4 months of focused execution. The A/B test was a success, unlocking full implementation across Web, iOS, and Android. Also opened the door for autosuggest on the TV platform — a new initiative I helped design, delegate, and oversee. Immediate buy-in from business stakeholders, PO leads, and Dev leads.
Having empathy for business stakeholders and developers gave us insights that shaped the right strategy. Sometimes the best innovation is simply using what you already have — well. Due to complications with TV and other projects, we couldn't launch everywhere at once, but this became an opportunity to grow as a delegator and a design lead.
Future improvements I'd advocate for: query intent modeling — distinguishing "I know exactly what I want" from "I'm exploring" — and voice search integration, especially relevant for a platform consumed across TV, phone, and tablet. The next frontier is search that feels less like a lookup and more like a conversation.