top of page

Hyper-Personalizing Online Shopping Recommendations With Personality Analysis

Online recommendation systems use algorithms to help customers find their preferred items based on a customer's profile -- a composite of data that includes among other things their navigation history, lookalike analysis, and previous shopping and purchase behaviours. Research has proven that insights into customer personality can further improve the effectiveness of recommendation engines, thereby increasing customer satisfaction and revenues for online retailers. However, uncovering individual personality data has been difficult, until now.

In this study, the authors explore inferring a customer's personality traits from their review texts, which are publicly available and can provide insight into the customer's personality. They use the Receptiviti API to analyze the customer's personality and feed it as input to a recommendation system to test whether it improves the recommendation performance. They construct two new datasets in the music and beauty domains and conduct empirical experiments, finding consistent improvements in the recommendation system's performance, from 3% without personality information, to 28% by incorporating personality information.


Trusted by industry leaders:

Subscribe to the blog

bottom of page