01
HMW make algorithmic outfit inspiration meaningful for real users?
Understanding how people construct outfits
We began with generative research to close insight gaps, studying how users interpret outfit suggestions, where they expect control, and what helps them trust recommendations.
Key insights
- Outfit planning is mental work and often guesswork
- Context drives outfit decisions.
- Anchors vary by person (item, occasion, mood)
- Anchored on owned item increases perceived value.
- Poor first suggestions destroy trust
- Visualization quality drives confidence
Hypothesis
Pre-selected context helps users grasp outfit intent before editing
Pre-selected contextual anchoring, combined with a rolodex-style outfit view, helps users quickly understand outfit intent and coherence before exercising control.
Surface UX
Landing view
View item details
Systems thinking
Designing outfit inspiration as a scalable system
Early concept tests showed users could quickly understand outfit intent when suggestions were anchored around context.
The next step was designing a system that could scale, expanding the pairing matrix while giving users ways to steer the algorithm.
Through collage
Increase the pairing matrix
Introduce accessories, bags e.t.c
Allow user feedback on style
Based on what we learned from the concept test, users want a way to steer the algo away from looks they'd never consider.
Outfit visualization
Some more stuff to come .....
Detailed case study available on request.