Architectural projects

case studies

Product Design projects

Apparel Group case studies

Development Process — AI Try on me
Case study
Development process — AI Chat Feature

Development Process: AI Try on me
Case study
Development process: AI Try on me
Step 01 of 10: Strategy
Goals
Aligning what shoppers need with what the business needs.
The feature was defined around two sets of goals that reinforce each other. On the user side, the aim was to let shoppers see outfits on their own body before buying, so they could decide faster and with more confidence. On the business side, the aim was to turn that confidence into measurable results: higher conversion, fewer fit and style returns, and stronger trust and engagement.
User goals and business goals
Visualize outfits on their own body before purchasingMake faster, more confident purchase decisionsReduce uncertainty around fit and styleIncrease purchase confidenceImprove conversion rateReduce return rates related to fit and styleStrengthen customer trust and engagement

User confidence and business performance were treated as one goal, not two.

Step 02 of 10: AI x Product
Role of AI
A decision support system, not just a visual generator.
AI was used not only to generate images but as a decision support system that helps shoppers choose. Its technical capabilities were selected to serve one purpose: giving people enough visual certainty to commit to a purchase. The AI is treated as a confidence engine, not a gimmick.
AI capabilities
Body detection & pose estimationGarment overlay & fitting simulationImage processing & refinementReal-time preview generation

AI is treated as a confidence engine, not a gimmick.

Step 03 of 10: Strategy
North Star Metric
One metric to keep every design decision accountable.
The North Star Metric was conversion rate uplift driven by AI Try-On adoption, while reducing fit-related returns. Pairing the two keeps the feature honest: it has to help people buy, and it has to help them buy the right thing.
What the metric balances
Conversion rate upliftAI Try-On adoptionFewer fit-related returns

Growth in conversion only counts if fit-related returns go down with it.

Step 04 of 10: Discovery
Problem statement
Why does uncertainty keep online fashion shoppers from buying?
Online fashion shoppers face two major uncertainties: “Will this fit me?” and “How will this look on my body?” Traditional static product imagery fails to communicate real fit and proportion, especially for first-time buyers. That gap leaves people hesitating at the exact moment they should be deciding.
What this uncertainty leads to
Lower conversionHigher return ratesLower confidence in purchase decisions

Static product imagery cannot answer the two questions shoppers care about most.

Step 05 of 10: Research
Competitive analysis
Mapping the gap between styling tools and virtual fashion.
To define the experience, I analyzed leading AI-styling and virtual fashion platforms. Stylitics is an AI-driven outfit bundling and styling engine that drives outfit-level shopping rather than single-item purchases, and its strength lies in personalized outfit recommendations and bundling. DressX focuses on digital fashion try-ons and virtual wearables, emphasizing identity, self-expression, and virtual representation rather than physical garment purchase.
References evaluated
Stylitics: outfit bundling & stylingDressX: digital try-on & virtual wearables

Existing solutions either focus on styling recommendations or on virtual fashion identity. The gap is combining real-world fit preview with purchase confidence inside a native commerce flow.

Step 06 of 10: Design
Entry point (PDP)
Step 1: Start where the decision happens.
A Try It On action is placed directly on the product detail page, keeping users in the purchase journey rather than diverting them elsewhere. Shoppers do not have to leave the product they are considering to find out how it looks on them.
Design decisions
Try It On action on the PDPStays inside the purchase journeyNo detour away from the product

The best place for a confidence tool is the exact moment confidence is missing.

Step 07 of 10: Design
Consent & camera access
Step 2: Guide the capture so the AI can get it right.
Users are guided to capture a full-body image with clear instructions. Setting these expectations up front ensures optimal AI accuracy and makes the permission to use the camera feel transparent rather than intrusive.
Photo guidelines
Solo photoFull body visibleStraight postureClear lighting

Clear instructions at capture are what make the AI result accurate later.

Step 08 of 10: AI x Product
AI Try-On generation
Step 3: Make the waiting feel trustworthy.
The uploaded image is processed using AI to generate a realistic try-on visualization. During processing, the interface manages expectations and builds trust instead of leaving the user guessing.
What the user sees while processing
Progress states manage expectationsVisual feedback reduces uncertaintySystem messaging reinforces trust

Processing time is part of the experience, so it was designed rather than hidden.

Step 09 of 10: Design
Interactive Try-On experience
Step 4: Turn the preview into a decision.
Once generated, users can view the garment on their body, switch outfits instantly, adjust sizes, and add items directly to bag. Every action that leads to a purchase sits within the same surface, which reduces decision friction and increases purchase confidence.
What users can do
View the garment on their bodySwitch outfits instantlyAdjust sizesAdd items directly to bag

Seeing, comparing, sizing and buying in one place removes the friction between confidence and checkout.

Step 10 of 10: Impact
Business impact & results
Measured with Hotjar, Mixpanel and GA4.
The Try On Me feature was designed to directly influence purchase confidence and reduce uncertainty-driven friction in the buying journey. After launch, the feature demonstrated strong impact across both user behavior and business metrics. Users who interacted with it spent more time on PDPs and showed stronger intent signals, and the ability to visualize products on themselves reduced hesitation at key decision moments. Qualitative feedback highlighted ease of use, clarity, and perceived usefulness as the primary drivers of satisfaction.
Key numbers
+18%
Conversion rate among users engaging with try-on
+40%
Purchase completion for users who struggled with sizing and fit
+30%
Customer confidence in size and style selection (post-purchase feedback)
−25%
Size-related return rates, lowering operational and logistics costs
+50%
Feature engagement in the first three months after launch
Behavioral insights
More time spent on PDPsStronger intent signalsLess hesitation at key decision momentsEase of use, clarity and usefulness as satisfaction driversBehavior shifted from browsing to decision-making

The feature shifted user behavior from browsing to decision-making.

Justlife case studies