In two studies we investigate how voice shopping may provide access to meaningful data on customer mood and how retailers may use such data. In Study 1 we explores the use of a machine learning approach to predict customer mood based on customer commands in the voice shopping process. We compare it to a heuristic approach to customer mood prediction based on situational correlates of mood that that a smart speaker can access (weather, music choice, day of week, and daylight). In Study 2 we explore how a voice retailer could use the potential capability to predict customer mood. Our results provide evidence that a customer’s good mood is associated with purchases of higher-priced premium brands. In addition, retailers can use mood prediction to adapt the presentation of product information to fit customer mood, thus helping customers optimize their decisions. In a sensitivity analysis, we examine what accuracy of mood prediction could enable retailers to use the explored effects.

在两项研究中，我们调查了语音购物如何提供对客户情绪有意义的数据的访问，以及零售商如何使用这些数据。在研究 1 中，我们探索了使用机器学习方法根据语音购物过程中的客户命令来预测客户情绪。我们将其与基于智能扬声器可以访问的情绪情境相关性（天气、音乐选择、星期几和日光）的客户情绪预测的启发式方法进行比较。在研究 2 中，我们探讨了语音零售商如何利用潜在能力来预测客户情绪。我们的结果提供了证据，表明客户的好心情与购买高价优质品牌有关。此外，零售商可以使用情绪预测来调整产品信息的呈现方式以适应顾客的情绪，从而帮助客户优化决策。在敏感性分析中，我们检查了情绪预测的准确性可以使零售商使用所探索的效果。