Senior product designer with over six years of experience in leading 0->1 products. Also a maker, coder, photographer, and more.
Made in: 2025
Made of: Design Strategy, Interaction Design
Made with: Chloe Sabo (Product Manager), Emmerline Wu (User Researcher)
The brief we received left many questions open. We were set to explore and define how far we want to push guest insights in the next two years. This was also the first initiative where UX took the lead on shaping product strategy and producing a playbook other teams could follow for similar challenges.
The outcome of the project was a two-year experience strategy that outlined how we test and iterate from quick wins to bigger swings. The prototype outperformed the current experience in every confidence category and every key experience signal, boosting guest trust, clarity, and decision-making.
3.88 → 4.5
Confidence in purchase decision3.81 → 4.88
The experience is considered human3.5 → 4.38
The experience is considered affirmingScores are averages out of 5 (current vs. prototype)
01
Synthesis vs. specificity
The reviews section is guests’ go-to place to find what people are saying about the product. Fortunately, there is no shortage of review content, with many products having over thousands of reviews! At the same time, it makes it a time-consuming task to assess the overall sentiment.
With the emergence of AI, we saw an exciting opportunity to do the detective work for our guests. But the trade-off of specificity and trust is challenging. We need more than just dumping a text summary. We need a digestible, relevant, and transparent representation of the review that aids decision making.
Research shows that shoppers typically read reviews to answer two questions:
- Is this product overall good quality?
- How does it fit on people like me?
Through a series of lo-fi explorations, I broke down the AI summary so that it speaks to those key questions and shows the proportionality of each sentiment.
In the user test, users responded well to AI summary of reviews, especially the clear, topic-based breakdowns. However, some were confused how these topics connected to the filter buttons below.
While exploring alternatives, I noticed a shared thread between the summary and the filters: both map naturally to the key questions. Instead of presenting the summary at the top and filters separately below, I consolidated the AI summary and review filters into a single, cohesive view built around the key questions guests trying to answer.
In a follow-up test, users found it more intuitive and easier to navigate. The AI insights now appear right where users are focused, without overloading them upfront.
02
For those who dig the dirt
Second to the image filter, the “1-star review” filter is the most engaged. People enjoy “digging the dirt” to uncover potential issues. While it may sound counterintuitive, these “dirt” reviews don’t necessarily diminish purchase confidence. In fact, the presence of negative reviews signals trust and authenticity, and sometimes the bad experiences aren’t even relevant to other shoppers.
When guests filter to read the 1-star reviews, they will see an AI summary that breaks down the top issues. Where appropriate, we will recommend an alternative product, or inject product education to contextualize the negative feedback.
03
Reviewers become the educators
Compared to the product description from the brand, reviews offer a unique perspective to the product from peers. It’s the voice of lived experience, it’s relatable, and it speaks in guests’ language. Sometimes, guests rely solely on reviews to learn about the product.
Seeing product features, review content, and product images together made it easier for users to connect what reviews were referring to. They especially valued not having to scroll back and forth, calling out the benefit of getting bite-sized, contextual information right where it was relevant.
The final design
“
I get the feeling that lululemon really loves its products, and it wants us to love them, too. So it’s really trying to show me all the things that will be important to me to make a decision.”— User test participant
This project was a true testament that impactful design lies at the intersection of user needs, business goals, and technical feasibility. Unlike online marketplaces, we had to balance the business implications of unflattering reviews with the value of authenticity that users deeply care about. Many of the concepts wouldn’t have been possible without close partnership with our data scientist, to craft an AI summary that’s not just what’s feasible, but what’s meaningful and desirable for our guests.
Ultimately, this project reminded me that great design isn’t just about crafting beautiful interfaces, but about finding harmony between truth, trust, and impact.
Let’s chat