#SearchExperience #AIIntegration #ProductOptimization
Role:
Product Designer
Company:
Pinkoi
Year:
2026
Platform:
01
OVERVIEW
Aligning search results with shopping intent
Search is one of Pinkoi's main product discovery surfaces and a major entry point for organic traffic. But results don't always reflect shopping intent, and fixed filters can't support how differently people evaluate different products.
I proposed Visual Search Refinement and a two-layer filter strategy so people can find the right product category quickly, then narrow results with dynamic product attributes.
After launch, Mobile Web product view rate rose from 37.4% to 44.7%, and add-to-cart rate on the new iOS and Android apps rose about 14% relative to the previous version. On Web, people who clicked Visual Search Refinement went on to view products about 30% more often than those who didn't — a positive signal, not proof of causation.
OUTCOMES
02
CHALLENGE
More results ≠ finding what you want
01
One search term can stand for very different needs. When results don't distinguish between them, even a large set of relevant products doesn't help people quickly find the category they actually want.
02
Different products are evaluated differently. From material to capacity to use case, what shoppers care about shifts with the search, and fixed filters can't offer a precise enough path to explore.
03
The results page is more than a product list. It also connects shopping intent to business opportunity. Balancing product discovery, shopping filters, Pinkoi's featured services, and commercial exposure within limited screen space was the core design challenge of this redesign.
03
GOAL
Help people find the right product faster
01
Bring results closer to what people are actually searching for.
02
Offer guidance that fits the search context, reducing repeated keyword edits and product-by-product spec checking.
04
STRATEGY
Different search intents need different ways to explore
People don't always know what they want to buy at the start. Searching "matcha" may still be about browsing product types; searching "water bottle" often means comparing specifics like capacity and material.
So I proposed two complementary exploration strategies, each matched to a different search context, to help people narrow their choices more efficiently.
Strategy 01
When intent is still forming
Intuitive images and category entry points help people recognize product types quickly and find their next direction to explore.
Strategy 02
When intent is specific
Combining the search keyword with product attributes, filters adapt to the search context and surface the information that matters for the decision.
05
DESIGN
Key design decisions
01
For broad searches, I proposed image-and-text category entry points that work across devices. Pairing visuals with keywords helps people recognize where to explore and narrow their search.
For example, when someone searches "gift," they can choose a scenario like "birthday gifts," "gifts for her," or "Christmas gifts" directly. Without rewriting keywords again and again, they can focus on what they need more intuitively.
02
I reworked the cross-device filter structure so common shopping conditions are easier to reach, while product attributes adapt to the search.
On Desktop Web, the left filter panel is streamlined to prioritize key conditions. On mobile, a two-layer filter balances common conditions with dynamic product attributes. For a "water bottle" search, for example, people can narrow by capacity or material.
To make dynamic filtering work in practice, I worked with the PM and AI team on recommendation criteria, and with the Localization and Business teams to review content so AI suggestions fit real shopping needs.
03
I noticed unused space beside the Desktop Web product list and proposed a new ad placement there. It adds commercial exposure for the platform without taking space from the main browsing area.
The placement balances the product exploration flow with the platform's business needs.
06
OUTCOME
PRODUCT IMPACT
01
02
03
07
TAKEAWAYS
Rethinking how search results help people choose
01
Earlier interviews and data analysis showed that even people with clear shopping intent are often limited by the words they can search with, struggling to describe exactly what they want. Existing filters were also used infrequently and didn't narrow results effectively.
This redesign used image-and-text guidance and prompts across multiple placements to help people focus on what they need more intuitively. Desktop Web behavior showed a positive signal, suggesting visual guidance is worth continuing to explore.
03
