GENERATING CONSUMER INSIGHTS FROM BIG DATA CLICKSTREAM INFORMATION AND THE LINK WITH TRANSACTION-RELATED SHOPPING BEHAVIOR

E-Commerce firms collect enormous amounts of information in their databases. Yet, only a fraction is used to improve business processes and decision-making, while many useful sources often remain underexplored. Therefore, we propose a new and interdisciplinary method to identify goals of consumers and develop an online shopping typology. We use k-means clustering and non-parametric analysis of variance tests to categorize search patterns as Buying, Searching, Browsing or Bouncing. Adding to purchase decision-making theory we propose that the use of off-site clickstream data—the sequence of consumers’ advertising channel clicks to a firm’s website—can significantly enhance the understand-ing of shopping motivation and transaction-related behavior, even before entering the website. To run our consumer data analytics we use a unique and extensive dataset from a large European apparel company with over 80 million clicks covering 11 online advertising channels. Our results show that consumers with higher goal-direction have significantly higher purchase propensities, and against our expectations - consumers with higher levels of shopping involvement show higher return rates. Our conceptual approach and insights contribute to theory and practice alike such that it may help to improve real-time decision-making in marketing analytics to substantially enhance the customer experience online.