01
The starting point is not product retrieval, but understanding the recipient first.
10 style archetypes
GIFT PROJECT · DECISION DOSSIER
包治百病
Built for the luxury-bag gifting scenario, it starts by understanding the recipient before moving into brand and item selection.It translates style, personality, and relationship cues into comparable brand and product matches instead of leaving users to browse blindly through labels and silhouettes.The output is an explainable, scored shortlist: a gift decision that can actually move forward, not just another browsing session.
Product Definition
Gift Decision ProductNot a product catalog, but a gifting decision system
A decision product that understands the recipient before pushing the user into product selection.
Traditional e-commerce works best for people who already know what they want. Gifting is the opposite. Many users do not know the brands, do not understand the style language, and cannot quickly judge what will truly suit someone else.
ClaimEvidence
02
The recommendation logic is not ordinary filtering, but a style-to-persona mapping.
14 luxury brands
03
The product rhythm is test first, then matching, then browsing, then saved comparison, not information overload.
159 bag entries in the recommendation structure
Core decision flow
Decision ChainBreaking vague gifting into steps that can move forward
The mini program breaks an originally vague and subjective gift choice into a sequence of easier decisions.
01
Set the tone
A lightweight landing page frames the task as choosing a gift for a specific person, not casually browsing bags.
02
Run the style test
Five rounds of binary choices estimate the recipient's style tendency and reduce decision ambiguity early.
03
Match suitable brands
The style result is mapped to luxury brands so the user is not overwhelmed by irrelevant labels.
04
Browse representative items
Selected bags are shown as swipeable item cards with price, description, and recommendation signals.
Rec index05
Save and compare
Saved items turn casual browsing into a shortlist the user may actually choose from.
System and data structure
Evidence BoardBehind the prototype sits a real recommendation structure
The value of this project is not only in its visual playfulness. Behind it, a clear recommendation structure and product logic are already in place.
10
style archetypes
14
luxury brands
159
bag entries in the recommendation structure
01
Style-to-brand mapping
Ten women's style archetypes are mapped to brand tone and representative bag types, making the recommendation path more interpretable.
02
Structured catalog
The prototype includes 14 brands and 159 items, with prices, descriptions, and recommendation metadata partially completed.
03
AI scoring experiment
An Express endpoint was used to explore recommendation scoring from style keywords and product descriptions, so the system does not stay limited to hard-coded rules.
Rec indexVisual language
Authored InterfacePixel luxury direction
A pixel gifting game with a point of view
What makes this project memorable is that it never tries to imitate a standard luxury storefront. It deliberately mixes pixel aesthetics, retro game rhythm, and the context of female fashion consumption.
- Pixel avatars turn abstract style labels into recognizable personalities.
- The saturated purple base makes the product feel playful and unmistakably self-authored.
- Swipeable product cards keep the browsing experience light and game-like instead of dense and commercial.
My role
End to EndBased on a full consumer decision prototype
I treated this as an end-to-end product prototype rather than a pure UI exercise, connecting product framing, visual identity, recommendation logic, and implementation.
01
Problem definition
Defined the project around gift decision assistance for luxury bags instead of building another product list.
02
Product design
Structured the journey from style diagnosis to brand fit and final item comparison.
03
Visual direction
Built a hybrid visual language of pixel graphics and fashion cues, giving the product a clear visual point of view.
04
Prototype implementation
Implemented the mini program natively, organized structured data, and explored LLM-assisted recommendation scoring.
What would strengthen it next
RoadmapA more realistic and practical optimization direction
The current prototype already proves that the direction works. The next stage is to make the recommendation structure more mature and the experience closer to real-world use.
01
Expand the item database and complete metadata consistency
02
Upgrade cloud storage and backend integration beyond local-first prototype logic
03
Refine real user testing around gifting confidence and shortlist conversion
04
Turn the AI score from experiment into a more reliable explainable signal
Closing Note
Point of ViewA product idea with a very clear point of view
Gift Project shows how I isolate a concrete consumer problem, give it a distinctive visual language, and design a recommendation flow that feels more directed than ordinary product browsing.