work
Walmart Case Study
Empowering item discovery with Gen AI summaries
Overview
I shaped the vision for integrating Gen AI into Walmart and took sole leadership of a key initiative. Partnering with product, engineering, and data teams, I ensured alignment with business goals, user needs, and scalability. I also co-led the creation of a scalable design system to maintain consistency across AI-powered experiences.
Through collaboration with 20+ teams, we delivered AI-powered tools in two months, improving item discoverability and reducing cognitive load for shoppers.
Goals:
Provide concise item overviews with easy access to details.
Reduce cognitive load and guide users to relevant information faster.
Create a scalable and PR-worthy use of AI to enhance Walmart’s digital innovation.
Role
Responsibilities
Collaborators
Timeline
Q3 2023
The challenge
Business Need
Use AI to help shoppers quickly evaluate and decide on items.
User Need
Streamline overwhelming item and review information while maintaining trust and transparency.
Item page
In person kickoff
Our team conducted a two-day workshop to identify user pain points and opportunities to simplify the shopping journey using Gen AI.
We identified three focus areas that would ultimately evolve into three sections of the overall E2E flow:
Awareness
Enhance and streamline search functionality to improve item discoverability.
Consideration
Provide tools to compare and narrow down options quickly.
Decision
Summarize item and review information to reduce cognitive load and enable informed purchases.
problem to solve
Users struggle to quickly understand item details and reviews, leading to decision fatigue and missed opportunities for informed purchases.
quick_reference_all
Research
Gathering insights
Emerging themes
Essentials to know
Users need concise, high-priority item details at a glance.
Amazon: Summarizes the reviews, highlighting positive, neutral, and negative feedback.
Zoom: Summarizes calls and chats for easy review with AI.
Wirecutter: Grades key features and highlights what matters most.
personalization
Tailor recommendations based on user data and preferences.
Wirecutter: Recommends items based on specific priorities or budgets.
Google Maps: Assigns match percentages based on past visit history.
trust & transparency
Build user confidence with clearly labeled AI content and linked sources.
Google Search: Provides answers and recommendations with source links for deeper exploration.
Amazon: Links review highlights to original content for verification.
Amazon
Zoom
Wirecutter
Google Maps
Google search
Key insights and themes
These insights drove the prioritization of features and ensured alignment with Walmart’s goals.
1
2
3
how might we
How might we leverage AI to highlight relevant item and review details, enabling users to make confident and quick purchase decisions?
lightbulb_2
Ideation
Multiple iterations
To meet tight deadlines, we prioritized rapid iteration, starting with high-fidelity mocks. Collaborating with Product, we refined key areas for the item page:
iteration 1
iteration 2
iteration 3
iteration 3 cont.
iteration 4
Co-design credit: Melina Sisaro
group
user testing
Testing the solution
Item summary learnings
Percentage match
Insight: Users found this feature less helpful for decision-making.
Action: Remove this section and expand the "Top Picks" feature.
Item details at a glance
Insight: Users often skipped this, perceiving it as a marketing tool.
Action: Explore different layouts and section headings to improve clarity and engagement.
Item specs
Insight: This section was well-received, with users most likely to pause and read here.
Action: Keep this feature and ensure scalability across all item categories.
Gen AI Module (Top half)
Gen AI Module (bottom half)
Reviews summary learnings
What reviewers said
Insight: Users still value reading full reviews, as not all users trust AI-generated summaries.
Action: Make navigation to the full ratings and reviews section more intuitive from this card.
What reviewers said
Insight: Users did not immediately recognize the summary as AI-generated.
Action: Clearly label the summary as AI-generated and indicate that it provides an overview of all reviews rather than highlighting a single one.
Reactions to "Generated by AI"
Users had mixed reactions to AI-generated content, praising its efficiency but raising concerns about authenticity, accuracy, and ethics.
Positivity around efficiency & innovation
“AI is making our lives easier tasks, more streamlined.”
Concerns about human authenticity & accuracy
“Sometimes if I see reviews are generated by AI, I am like, is this actually a person that reviewed this? Did they, you know, use something to generate the review? Things like that... it kind of gives me a feeling of like, it might be a fake review.”
Concerns around ethical effects & data privacy
“They're underpinning their employees in my opinion, and then also cutting jobs, trying to cut down costs when they're already making a big profit there. Um, so if I saw that Walmart was using AI to power these contents and results on their website, that yeah, that would give me a negative view.”
compare_arrows
aligning e2e
Ensuring visual consistency
We initially debated making AI content stand out but shifted focus to integrating it seamlessly for a cohesive experience.
I co-led the color strategy, ensuring alignment with Walmart’s design system while maintaining a distinct, recognizable identity. I explored scalable color approaches, tested variations, and presented recommendations to stakeholders, balancing consistency with differentiation.
Light color strategy
The palette prioritized simplicity and trust, ensuring seamless alignment with Walmart’s design language.
Item summary
Review summary
Dark color strategy
This option used a bold, innovative palette with a gamified toggle to reinforce transformation.
Item summary
Review summary
We chose the light color strategy to balance familiarity with innovation.
To enhance visibility while preserving harmony, I opted for a slightly darker blue than our primary tone.
editor_choice
Final Design
The MVP flow
The final MVP delivered a cohesive AI-powered shopping experience that seamlessly integrated into Walmart’s platform.
Full E2E Gen AI integration flow
Search results
nudge appears on scroll
refinement applied
Item Page
Item summary
Item summary cont.
review summary
aspect summary
Highlights & quick description
The highlights and item summary provide relevant, contextual details to help users make quick, confident decisions.
For those needing more information, a link directs them to full item details, which are not AI-generated.
At a glance
Displays up to six key specifications of the item in an easy-to-digest visual format. To ensure clarity in a text-heavy section, I worked closely with illustrators to rapidly create 60+ custom icons in time for launch.
Users can access the full, non-AI-generated specifications through a link at the bottom of the card.
Reviews summary
Helps users identify critical review insights, classified by sentiment, to aid informed purchase decisions.
A link directs users to the full reviews if they want more detailed feedback beyond the AI-generated summary.
Aspect summary
Provides an overview of user sentiment for specific features or attributes, with a filtered list of relevant reviews for deeper understanding.
Search initiative
Gen AI tailors the top-of-page categories to user search queries, adding educational content to help users better understand their options.
+3.62%
search ATC/visitor
+2%
item click-Through-rate
-1.81%
Referred item ATC/visitor
Design credit: Lauren Glazer
Design credit: Franklin Huynh
Comparison initiative
Introduces a comparison tool for evaluating multiple items on the search page, reducing the need to use the cart as temporary storage, enhancing efficiency and reducing back-and-forth navigation.
+2.62%
add-to-carts/visitor
+1.14%
new buyer conversion
-1.81%
removal of item from cart
monitoring
impact
The results
+
1.1
%
-
0.93
%
-
0.63
%
What I learned from this project
rocket_launch
thumb_up
handshake
work
Next Case Study
Simplifying review management with a centralized reviewer profile, personalized dashboards, and transparent feedback systems to foster contributions and trust.
0 to 1 Design
loyalty program
Web & App
