Overview
I led the design of AI-powered filters to improve product discovery across our apparel rental platform. The initiative not only addressed user frustration with limited filtering but also enabled internal teams to more easily create and manage collections from the backend. The new system increased filter adoption by 84% and laid the foundation for scalable merchandising.
Problem Statement
Users reported challenges in finding desired garments due to limited and confusing filter and sort functionalities. The existing system offered basic categorizations like "Dresses," "Tops," and "Bottoms," with sort options that lacked clarity and relevance, such as an unclear "Featured" option and unordinary sort options, like brand names and colors, which were sorted alphabetically.
At the same time, internal teams lacked flexibility. Adding or updating collections required manual workarounds, slowing down launches and limiting our ability to merchandise effectively.
Our challenge: design a filtering experience that improved product discovery for users while streamlining backend operations for scale.
Project Role
- Led UX and UI design across research, wireframing, prototyping, and final specs in Figma
- Partnered with product managers, engineers, and data scientists to align user needs with backend AI capabilities
- Defined new filtering patterns that balanced clarity, control, and trust for users while improving operational efficiency
Process
Research and Insights
Qualitative user interviews revealed that the existing filters and sorting mechanisms were inadequate, leading to difficulties in product discovery. Users expressed the need for more intuitive and meaningful ways to navigate collections.
Collaboration with Data and Engineering
Concurrently, our data and engineering teams were developing algorithms to classify inventory using AI. This technology enabled the categorization of garments by occasion, season, style, and weather suitability.
Design Solution
Integrating user feedback, AI capabilities, and our internal team's need for improved collection management, I designed a unified sort and filter interface accessible via a single button, opening a drawer with comprehensive options.
The user was then able to sort by price, addressing users' price sensitivity, and filter by occasion, style, length, and weather in addition to size, color, and brand, and our internal team was able to easily add, edit, and remove collections given the new structure.
Outcome
The implementation of AI-powered filters led to an 84% increase in filter adoption rates and enhanced product discovery, contributing to higher conversion rates. Additionally, the internal teams adjusted collections directly from the backend without engineering dependences, improving speed and scalability.
This project reinforced the importance of designing for both user and operational needs. By combining user research with backend AI capabilities, we delivered an experience that was not only intuitive for users but also scalable for internal teams. It was a reminder that the best design solutions solve for the full ecosystem.
Continuing the utilization of CaaStle's AI and improving user experience, my colleague Connie Mach designed the AI-powered search function on The Ensemble.
