Project details
- Team
- Customer Reviews Experience
- Platforms
- iOS, Android, Web
Some details are generalized. Read confidentiality note.
This case study focuses on my role and design approach. Illustrative data and reconstructed visuals are labeled where they appear.
Turning thousands of reviews into a signal that customers can trust.
Role UX Design Lead
When a product has hundreds or thousands of reviews, shoppers need a way to identify the themes that matter to them.
To help shoppers get to the point faster, Amazon introduced AI-generated summaries and attribute tags, like “Ease of cleaning” or “Noise level.” My role as UX Design Lead was to refine these attribute tags so they didn't just look like static text, but felt like clear, interactive signals shoppers could trust and explore.
I led baseline usability research, compared customer-feedback patterns across other platforms, and explored sentiment cues and interaction styling for the attribute tags.
Explore the interface: select an attribute to see the review information behind it.
Select an attribute above.
I stepped back to ask what these tags were actually doing.
Before jumping into visuals or interface tweaks, I wanted to step back and ask three fundamental questions about what these tags were actually doing.
How should an attribute tag instantly communicate whether an attribute is positive, negative, or mixed?
How do we make it obvious that these tags are clickable actions rather than static badges?
How easily can a shopper jump from a high-level AI summary to the real, unedited customer reviews that back it up?
- SummaryA short overview of what reviewers mention about a product.
- Attribute tagA topic from that summary, such as “Sound quality.”
- Supporting reviewsThe individual reviews that mention that topic.
I made sure we weren't designing in a vacuum.
I grounded the work in baseline usability research and a competitive audit. I ran usability testing on the existing experience to understand whether shoppers noticed the tags, recognized that they could select them, and found the supporting review information.
I compared how other platforms presented customer feedback. Walmart placed topic tags beneath a review summary. Best Buy used topic chips with counts and sentiment colors. TikTok Shop grouped review keywords into chips without per-keyword counts, while AliExpress placed counted review tags above individual reviews. I used these examples as references for exploring tag presentation and interaction.
Understand where shoppers got stuck in the existing experience.
Compare familiar customer-feedback patterns across major platforms.
Walmart
Best Buy
TikTok Shop
AliExpress
I evaluated every idea against three core principles.
The visual language had to mirror the sentiment and tone of the content. The interface needed to invite interaction without feeling cluttered or pushy. And the iconography and color choices couldn't conflict with how symbols were used elsewhere across Amazon.
I compared how checkmarks were used elsewhere in the shopping experience: to confirm an order or an add-to-cart action, present a Subscribe & Save offer, and accompany a message about lower return rates. Those different meanings gave me a reason to consider the tag's sentiment cues as a connected system.
The visual language must directly mirror the sentiment and tone of the content.
The interface should naturally invite interaction without feeling cluttered or pushy.
Icons and colors cannot conflict with how symbols are used elsewhere across Amazon.
I considered the meaning of a symbol alongside its use elsewhere in the interface.
Order confirmation
Subscribe & Save offer
Lower-return message
Add-to-cart confirmation
Connecting the summary to supporting reviews
I explored sentiment cues and interaction styling for the attribute tags. The reconstruction below shows the connection between a selected tag and supporting review information: mention counts, a sentiment breakdown, and review snippets.
Cooking performance Select any tag in the phone to open its details.
I explained the trade-offs behind the tag explorations.
I connected the sentiment and interaction proposals to the questions they addressed and to the surrounding shopping interface. I advocated for focused comparisons so changes to sentiment cues and interaction styling could be evaluated separately.
The attribute tag had to adapt across multiple environments.
These examples show review information in several shopping contexts: return-related feedback on a product page, attribute tags in an Alexa Shopping response, and a compact product preview. They illustrate differences in available space and presentation; they do not establish that I designed each surrounding experience.
Review information across shopping contexts
Product Detail Page Return feedback
Alexa Shopping Attribute tags in a mobile response
Long-Press Preview Review information in a compact product view
Search Results Topic-based recommendations
Define what a tag needs to communicate before trying to paint it.
This project reinforced the value of defining what a tag needs to communicate before choosing its visual treatment. I would keep sentiment, interaction, and access to supporting reviews connected throughout research, exploration, and evaluation.