A path from multichannel customer data to real-time personalization: Predicting customers’ psychological traits through machine learning

This study explores the feasibility of inferring psychological traits from multichannel customer data using machine learning algorithms. Partnering with a German fashion retailer, the data from 7188 customers who completed an online survey assessing their psychological traits alongside multichannel customer data from the retailer's CRM database were analyzed. The study demonstrates that domain-specific traits such as risk attitude, chronic shopping orientation, need for touch, need for interaction, need for cognition, quality consciousness, and price consciousness can be inferred with moderate to high accuracy. A comparative analysis indicates that the predictive models developed in this study outperform those models trained on alternative digital data records. The findings underline the value of leveraging multichannel customer data to accurately predict individual psychological traits, thus enabling more personalized and automated marketing strategies. This study also provides a methodological framework for practitioners and researchers to utilize psychological trait prediction from digital customer data, advancing the capabilities of marketing automation, such as psychological targeting and real-time personalization in commercial settings. • Machine learning predicts psychological traits from multichannel customer data. • Trait predicting from customer data enables psychological targeting in marketing. • Traits like price consciousness and need for touch show high predictive accuracy. • Comparative analysis reveals superior predictability over banking transaction data. • The study provides a practical framework for setting up trait prediction pipelines.