Brief, competitors, user analysis & Problem definition

The Brief: Launch a customizable jewelry line competing with Pandora and Giovanni Raspini, enabling users to preview their compositions without the friction of multiple tabs and imagination gaps.

Initial Analysis & Constraints: Analyzing the 200+ charm catalog and mobile-first requirements (92% users), I identified critical UX challenges in existing solutions.

• Pandora: No preview functionality. Users must add a bracelet to the cart and then add charms separately, without seeing the final composition.
• Raspini: Offers a preview composer, but requires users to rotate their phones to view the entire bracelet. In horizontal view, drag-and-drop is nearly impossible—once users start dragging a charm, the bracelet moves out of the viewport.

Key Takeaways:

• Pandora benefits from strong brand recognition, allowing users to purchase bracelets even without previews.
• Raspini provides a more complete experience, including charm filtering by shape and category, but usability is limited on mobile.

Problem definition: How might we allow users to preview their composed jewelry in a simple, mobile-first experience that includes filtering charms and easily editing the preview?

User flow and jewelery composer solution

I mapped the full user flow and defined all possible actions at each step.

User Flow:

• A call-to-action in a homepage banner and a menu link lead users to the composer page.
• The guided process unfolds in five steps:
1. Choose the jewelry category (necklace, bracelet, earrings, ring).
2. Select the model, color, and size.
3. Browse and select charms.
4. Compose the jewelry.
5. Check the summary.
• Finally, users proceed to “Add to Cart” and checkout.
• A call-to-action in a homepage banner and a menu link lead users to the composer page.

User Actions:

• At each step: go back, continue, or exit the flow.
• In Step 4: place charms on the master, change positions
• In summary: remove charms, select a different master or charms through breadcrumbs.

AI introduction

As part of a strategic decision, the client decided to integrate AI. A technology partner proposed using AI to catalog charms by analyzing images and descriptions, extracting tags such as shape, color, mood, and gift occasions. Users could then complete a questionnaire to filter charms and express preferences. Based on their answers, AI generated suggested compositions.

I proposed reducing the number of questions and adding a “Skip to Composer” option to minimize drop-offs.

Key Challenges & Solutions:

• Preview Generation: Without 3D models, I studied charm and master proportions, provided detailed guidelines for asset preparation, and implemented a 2D composer with precise drop-zone coordinates.
• Collaboration: I worked closely with developers to ensure smooth integration and cart functionality.

First results

The AI composer launched successfully, earning recognition at the Netcomm eCommerce Fair and driving a 32% increase in new collection sales. Additionally, I designed and developed the front-end for an in-store kiosk used in multiple malls.

Smartlook UX and Data Analysis

Post-launch, user feedback highlighted that the questionnaire was too long, causing many users to abandon the flow.
I conducted an in-depth analysis using Smartlook:

• Data Sample: 2,952 user sessions.
• 92% of users on mobile faced significant usability issues.
• 213 users attempted the configurator, but only 88 reached checkout.
• Search function failing on basic queries ("charm" returned unrelated necklaces)
• Cart abandonment reached 58%, largely due to confusing UI and missing CTAs.
• Many users bypassed the configurator, manually adding items to the cart without previews.
• Search functionality failed: queries like “bracelet with cat charm” or “necklace with one in a million charm” returned zero results, despite these being possible to configure.

Insight:
Users weren’t searching for pre-made products—they wanted to create personalized jewelry. The rigid, step-based configurator was forcing premature choices and leaving them with no results. This insight reframed the problem and opened the door for a more intelligent, user-centered solution.

New User Flows

Building on these findings, I designed a new flow that combined flexibility and intelligence:

AIKI Assistant: Powered by AI, the assistant understood user intent expressed in natural language and generated relevant compositions.

Search Integration: Users searching for composable jewelry (e.g., “necklace with pink stones”) could now get results instead of dead ends.

Dual Approach: Users could either follow AIKI’s suggestions or use the manual composer to design their jewelry independently.