Upstage Provides Hyper-Personalized Recommendation API to Lotte On

Upstage has signed a recommendation API adoption and usage agreement with Lotte On to provide a hyper-personalized recommendation API for app and website customers. Amid the spread of generative AI accelerating AI adoption in the retail and e-commerce industry, this is a case of strengthening behavior data-based recommendations.

The recommendations go a step further than demographic information such as age, occupation, and gender, reflecting specific behavioral data such as search patterns, purchase responses, and cart history. Upstage applied the recommendation API it unveiled last March, and cited its experience supplying a recommendation AI pack to Brandy that helped it achieve its first profit turnaround, along with its track record of papers at international AI conferences and Kaggle gold medal achievements as evidence of its technical capability.

During the adoption process, it reported better performance than competing global AI companies' solutions on key metrics such as click-through rate and transaction conversion rate. Across four rounds of testing conducted since January this year, the purchase conversion rate in the recommendation area steadily rose, and the final fourth round of testing showed a 30% improvement compared to the first round.

The two companies plan to more precisely analyze individual customer preferences based on Lotte On's 39 million member data and integrated online/offline data. At the same time, they will continue improving model performance by monitoring metrics with AI-based analysis tools, with a goal of raising the purchase conversion rate up to 50% compared to the initial level.

Upstage's recommendation AI has also been validated through accolades such as the WSDM2023 Best Paper Honorable Mention Award and a silver medal in the H&M personalized fashion recommendation competition. The Seargest (search+suggest) technology, which combines search and recommendation, has also been applied to AskUp, and this agreement will further expand its application cases in commerce platforms.