Designing a seamless user experience for AI-Powered shopping and styling platform with a focus on user-centered design approach for both web and mobile
Mobile iOS App • Sep 2023 - Nov 2023
Perfect Fit AI
Project Overview
TL;DR
Perfect Fit is an AI-powered digital product that solves online clothing fit issues through personalized size, style, and color recommendations. By combining user measurements, style preferences, and color analysis, the app reduces returns and builds shopping confidence. As project lead, I directed user research and design strategy while coordinating a 5-person UX team to validate a measurement-first approach to fashion personalization.
THE SETTING
Timeline
3 months
Independent UX designer (research, design, prototyping, testing)
Role
Challenge
Online shoppers often lose confidence when sizing feels unclear or arbitrary, leading to frustration and returns
Goal
Build a shopping experience that not only improves fit accuracy but also reassures users through clarity and trust.
Key Insights
Fit is emotional
poor fit hurts more than physical comfort
Users express loss of confidence, avoid reorders, and feel negative about themselves.
1.
89% of users had emotional responses to poor fit - it wasn't just inconvenience, it damaged confidence.
Users wanted body-type guidance beyond basic sizing - "Medium" isn't enough.
Generic sizing fails
body-type guidance matters
Users want help not just with “medium/large” but with how clothes flatter their unique shape.
2.
Navigation copy & onboarding flow shape trust
Clear labels (e.g. “My Measurements” vs vague brand terms) + branching flows for different user motivations increase feature adoption and user satisfaction.
3.
Clear navigation language increased feature adoption from 26% to 53%
Users were willing to invest time in setup when they understood the value.
Transparency = trust
Showing why recommendations are made and how measurements are used, along with privacy information, boosts acceptance and reduces friction.
4.
🔍 How I Found That Out
User Interviews & Surveys → Exposed the emotional weight of fit failures.
What I did: 15+ interviews and 100+ survey responses across different demographics
Key finding: Users described poor fit with surprisingly emotional language - "I felt terrible about myself" rather than just "it didn't fit"
How the product changed: Messaging shifted from functional (“save time”) to confidence-focused (“shop with certainty”). UI copy was revised accordingly.
Research methods: Interviews, surveys, competitive analysis, personas
User Personas + Affinity Mapping
→ Identified three distinct onboarding needs
Deliverable: "The Realistic," "The Trend-Setter," and "The Researcher" archetypes through affinity mapping
What I learned: "The Realistic" represented our largest segment - users who wanted quick, accurate results without complexity
How I used it: Designed primarily for The Realistic's needs (simple onboarding, clear labels) while accommodating other types through optional features
Meet "The Realistic" - Our Primary User
"I just want to know it'll fit. I don't have time for complicated processes."
Age: 25-35 | Shops online 2-3x/month | Values efficiency over exploration
Key Frustrations:
Sizing inconsistency across brands
Returns are time-consuming
Complicated setup processes
Goals:
Quick, accurate size recommendations
Confidence in purchases
Simple onboarding experience
Card Sorting + Tree Testing
→ Improved navigation success rates from 26% to 53%
The deliverable: Conducted 3 card sorts and 2 rounds of tree testing with 30+ participants each, specifically targeting problematic user flows like AR try-on access and saved item purchasing.
What I learned: Users had strong mental models - they expected measurement data separate from social features and needed multiple entry points to key features. Small language choices had massive impact: "PerfectFit Passport" confused 74% of users, while clear labels like "My Measurements" immediately made sense.
How I used it: Restructured our app navigation to match user expectations rather than our internal feature groupings. Moved privacy controls to a dedicated section, created multiple pathways to AR try-on, and separated personal data management from style preferences. These seemingly small changes doubled our navigation success rates and gave users confidence to explore advanced features.
Prototype Testing
→ Validated measurement-first approach
The deliverable: Interactive prototypes tested with target users
What I learned: Users were willing to invest time in detailed setup when they understood immediate value and privacy implications
How I used it: Added transparency features showing recommendation logic and privacy controls, increasing user trust
Interactive Prototype
📊 Results and Reflections
What worked:
Emotional insight drove strategy: Understanding fit as a confidence issue, not just a logistics problem
User-centered navigation: Small language changes doubled success rates
Transparent AI: Users trusted recommendations more when they understood the logic
What I’d do differently:
Focus scope earlier: Trying to solve sizing, styling, and color simultaneously created complexity - I'd start with one core problem
Invest in measurement validation: Partner with professionals to ensure our measurement guidance was accurate
Earlier stakeholder involvement: Bring in potential retail partners sooner to understand integration challenges
What this shows about my approach
This project demonstrates my ability to:
Lead cross-functional teams while staying hands-on with research and design
Translate research insights into strategic decisions that change product direction
Notice overlooked details that significantly impact user experience
Adapt quickly when testing reveals problems with our approach