Consumer decisions in virtual commerce: Predict good help-timing based on cognitive load.

Abstract

The retail sector is steadily moving toward virtual commerce (v-commerce), and the process has recently gained momentum. With the latest developments in headset technology and the rise of artificial intelligence, virtual shopping has become relevant for an increasing number of products. In this article, we present a study that combines consumer behavior research, eye tracking, electrocardiography, machine learning, and the application of virtual reality. Fifty participants were invited to experience a virtual scenario, perform multiple mentally demanding tasks, and make a purchase decision for a product from one of two different product categories. In a post hoc video analysis based on the first-person view, participants determined different points in time when they would have appreciated help from an algorithmic user assistance system or a digital human agent. Our statistical analysis suggests that the desired help-timing depends on the product category. For fast-moving consumer goods, algorithmic help was requested particularly early. Furthermore, we collected eye-tracking and electrocardiographic data to build and evaluate a predictive classification model that differentiates between three levels of cognitive load. The trained machine learning algorithm aims to classify cognitive load during decision making, which may indicate the right time to offer help. Our findings provide evidence that eye movements, in particular, allow service providers to determine a good moment to approach consumers during their shopping experience. (PsycInfo Database Record (c) 2024 APA, all rights reserved)