The Effect of Voice AI on Consumer Purchase and Search Behavior

Voice-activated shopping devices (voice AI), such as Amazon’s Alexa or Alibaba’s Tmall Genie, as a new channel for shopping, are gaining popularity among consumers worldwide. It has become important, therefore, to understand how the adoption and usage of voice-activated shopping affects consumers’ purchase and search behavior, and which types of consumers would purchase and search more (or less) as the result of adopting voice AI. We collaborate with Alibaba to take the first step toward filling this research gap. In essence, we leverage a natural experiment and a novel algorithm, instrumental forest, to estimate the heterogeneous treatment effect of voice shopping. The results indicate that the usage of voice-activated shopping leads consumers to purchase more quantities and spend more. There is substantial heterogeneity in the treatment effects. The positive impact on purchase quantity is more pronounced for high-income, younger, and more active consumers, whereas the increase in spending amount is more pronounced for low-income, younger, and less active consumers. Voice AI does not cannibalize other channels; rather it boosts purchases through PC and mobile channels. Moreover, the adoption of voice-activated shopping leads to more search in terms of both breadth and depth. The increase in search is positively correlated with previous spending, income, activeness, age, and being female. We further explore the mechanisms that explain the changes. The findings provide useful implications for both e-commerce companies and businesses that harness voice-activated shopping.