South KoreaRetail and e-commerce2025 – ongoingCJ Olive Young
Review-based AI recommendation themes
CJ Olive Young used a language model trained on about 40 million reviews to generate app shopping themes
Problem
CJ Olive Young wanted to turn its accumulated customer reviews into product discovery features.
Approach
The retailer developed a small language model focused on its review database. It deployed an app feature that extracted review expressions and generated new shopping themes daily. Each theme could recommend up to 15 products.
Results
Takeaway
The deployment converted review language into shopping categories. No increase in conversion or sales was disclosed.
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
Electronic Times reported the launch of review-based recommendation themes in Olive Young's app. It described an internally developed language model trained on 40 million reviews and recommendations of up to 15 products per theme.
Full details are in the original sources.