Chat with a Product
Lead UX Designer | March - July 2025
Designed and launched an innovative AI-powered "Chat with a Product" feature, enabling customers to efficiently find answers without waiting in a queue for a sales agent or reading through Crutchfield’s extensive research.
This initiative leverages conversational AI to redefine the online product discovery and pre-purchase support experience.
My role - Lead UX Designer
-
Research
Conducted competitive analysis and developed personas, created user tests, presented findings to stakeholders
-
Define
Defined user conversational flows and interaction patterns with chat interfaces.
-
Design
Prototyped interactive UI in Figma, led stakeholder demos, developed a small AI design system for future products.
-
Test
Conducted iterative focus group testing with 10+ internal users to ensure an intuitive and trustworthy experience.
My impact
-
Efficiency & scalability
Demonstrated significant operational efficiency gains through internal testing. Stakeholders reported sales team feedback to be “overwhelmingly positive”.
-
User adoption & satisfaction
Achieved a 100% user satisfaction rating for chat interactions through user testing, highlighting its intuitive design.
-
Innovation & leadership
Pioneered a sub-system of design components for future AI products that would exist on the site.
Research
-
What types of questions do customers most frequently ask about products before purchasing?
What specific product information is hardest to find or understand on existing product pages?
How do customers prefer to interact with product information (e.g., direct answers, comparisons, recommendations)?
What are their expectations and concerns regarding AI-powered assistance in a shopping context?
What are the key conversational "happy paths" and potential "failure points"?
-
Analysis of Customer Service Transcripts/Call Logs (to identify common pre-purchase questions)
User Interviews (to understand existing product research behaviors and AI expectations)
Competitor Analysis (of existing conversational AI in e-commerce)
Usability & Concept Testing of initial chat flow ideas and persona attributes
-
Key User Needs
While the current product detail page (PDP) is comprehensive, users often have to invest significant effort to find specific answers. They want fast, direct responses—without needing to contact an advisor—for questions related to:
Compatibility (e.g., “Will this amplifier power my existing speakers?”)
Technical specifications
Product use cases
Product comparisons
AI expectations
Users expect AI experience to be:
Highly accurate
Quick to respond
Conversational, yet neutral — helpful and human-sounding, without bias or sales language
Considerations for design
To meet user expectations, the experience should:
Avoid marketing or promotional language
Provide links to sources when available
Clearly disclose that AI may produce occasional errors
Offer a seamless handoff to a human advisor upon request
Personas
“First-Time Fiona” – enjoys a conversational way to dig into product specifics and feels intimidated to discuss specs with an expert. While "Independent Isaac" has the vocabulary to discuss technical audio gear but prefers immediate, self-service solutions for pre-purchase questions rather than waiting for human assistance.
Ideation
We initially explored integrating an AI shopping assistant into our existing chat interface (left/top). However, we quickly pivoted after realizing the current chat experience needed a substantial refactor and redesign. As a result, we scoped the chat integration as a Phase 2 initiative. Phase 1 instead focused on integrating the AI shopping assistant directly onto the page (right/bottom). By doing this, we were able to get the product out to customers much faster and quickly gather feedback on the helpfulness of the AI model.
Shaping the on-page experience
UX designers, developer, and stakeholders worked together to define technical and financial limitations of the AI model.
Testing
-
User testing
Overview: Conducted a series of unmoderated remote usability tests with 10 participants representing the target personas.
Key tasks: After reviewing the page in full, we asked users to locate the AI shopping assistant and ask at least 2 questions about the product, giving feedback on how the experience aligned with their expectations.
Key findings: User testers were asking more complex questions of the AI than we anticipated. Since this was the first iteration of the AI model, it was primarily meant to scan the page content and provide quick product information. A majority of testers also showed little interest in revisiting past questions through clicking “View conversation”.
Design iterations: We changed the title and description to reframe the purpose of the tool, and changed the location of the “View conversation” button.
-
Internal testing
Overview: We released the new tool on our internal servers and asked the Sales Team Leads and Q&A groups to help test/provide feedback over the next few weeks.
Key tasks: When looking for answers on a product page, first try entering the question in the chat box. If the answer could be improved, or is wrong, please "Flag this response" and select the reason/provide more information.
Key findings: Roughly 80% of customer Q&A could be answered with the tool. Using the tool led to a 98.3% efficiency gain for employees answering customer Q&A on product pages and social media.
Design Iterations: Toned down AI branding since it led to many advisors asking more subjective questions. Allowed customers to thumbs up/down a response. If they thumbs down a response, we directed them to start a chat with an advisor.
Before user test
After user test
Design System
Developed new conversational UI components and patterns within the existing design system, including suggested questions, feedback buttons, avatars, and loading states specific to AI interactions. This ensures visual consistency and scalability for future conversational features across the site.
Final design
Final design
After testing with over 20 users, initial qualitative feedback was overwhelmingly positive—users frequently described the experience as “sleek,” “VERY cool,” and “helpful.”
From a business perspective, the feature positioned Crutchfield as a leader in retail technology, delivering a scalable solution for complex product discovery and pre-purchase support. It directly enhanced both sales efficiency and the overall customer experience.
This work also laid the groundwork for future conversational AI applications, including intelligent product recommendations, installation assistance, and support within live chat interactions.