United StatesRetail and e-commerce2025 – ongoingWayfair
Gemini-generated customer interests for product recommendations
Wayfair used Gemini-generated interests to personalize product recommendations
Problem
Fixed preference categories missed shopping needs that customers implied through their behavior. Wayfair wanted recommendations to reflect those needs without requiring shoppers to state them explicitly.
Approach
Wayfair fed historical shopping activity into Gemini to generate free-form descriptions of customer interests. Batch jobs produced interests that were stored for use in homepage carousels and product-detail recommendations. Its search model retrieved products matching those interests.
Results
No published figures
Takeaway
Interest generation ran offline, allowing the shopping interface to retrieve prepared recommendations without waiting for a new model response.
Categories
- Size
- Large enterprise
- Industry
- Retail and e-commerce
- Business function
- Marketing and sales
- Innovation type
- Product and service innovation
- AI technology
- Generative AIRecommendation and personalization
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
Wayfair described its Gemini-based interest-generation system and deployed recommendation placements. It reported engagement and revenue gains without publishing their size.
Full details are in the original sources.