Disruptive information technologies, including artificial intelligence (AI), blockchain, big data, and cloud computing, have reshaped marketing in unprecedented ways, which has made digital marketing a prominent topic in both academia and practice. However, despite the upsurge in the number of publications, related topics remain fragmented. Drawing on the antecedents-decisions-outcomes (ADO) framework and adopting an interaction perspective, this paper systematically reviews 693 relevant articles and develops an integrated framework for digital marketing centered on the logic of “interaction antecedents‒interaction processes‒interaction outcomes.” Building on this analysis, the paper further identifies several critical issues in current literature, including the AI paradox, the personalization-privacy paradox, the impact of digital technologies on customer behavior, the measurement of digital marketing performance, and theoretical frameworks for digital marketing interaction. Finally, it outlines multiple avenues for future inquiry, offering valuable implications for both theoretical development and practical advancement.

This paper investigates the effects of bullet-screen content and behavioral characteristics on consumer purchasing behavior in live streaming e-commerce, as well as the moderating effect of streamer-product relevance. Based on the Elaboration Likelihood Model (ELM), this paper uses bullet-screen data from the Douyin platform and consumer data from the Huitun platform. By combining text mining with zero-inflated negative binomial regression, this paper investigates the factors influencing consumer purchasing behavior from two perspectives: bullet-screen content characteristics as the central route and behavioral characteristics as the peripheral route. Grouped regression analysis is further employed to examine the moderating role of streamer-product relevance. Information richness, social interaction degree, and the number of bullet-screen comments positively affect purchasing behavior. The effect of the sentiment polarity of bullet-screen comments on purchasing behavior follows an inverted U-shaped pattern. Compared with live streaming rooms with low streamer-product relevance, those with high streamer-product relevance exhibit a broader range over which the sentiment polarity of bullet-screen comments positively affects purchasing behavior. This paper selects bullet-screen comments from only one live streaming e-commerce platform, and its findings therefore lack generalizability. Analyzing the factors influencing consumers’actual purchasing behavior from the perspective of bullet-screen comments can provide practical guidance and recommendations for effective communication between merchants and consumers and for improving sales effectiveness under the live streaming e-commerce model.

The widespread application of artificial intelligence, such as machine learning, deep learning and ChatGPT, has sharply reduced the incidence of new product shortages and stockouts. However, it has not eliminated them or their consequences. Although there has been abundant prior research on product out-of-stock and compensatory consumption respectively, few studies have explored how attributions for new product out-of-stock impact consumer compensatory consumption, especially how the effects of algorithm-error attribution and human-error attribution differ on consumer compensatory consumption. This gap has held back the development of theory on new product outof-stock and compensatory consumption and has also limited the practical guidance available to manufacturers, consumers, and regulators. Drawing on the theory of perceived control and conspiracy theory, this paper develops a moderated mediation model to explain how and under what conditions attributions for new product out-of-stock influence consumer compensatory consumption. The model is tested through a field survey and three experiments. The field survey examines the direct effect of attributions for new product out-of-stock on compensatory consumption. Experiment 1 tests the robustness of this effect and investigates the proposed mediating mechanisms. Experiment 2 examines the boundary conditions of these mechanisms and rules out plausible alternative explanations. Experiment 3 further establishes the robustness of both the mediating mechanisms and their boundary conditions, while confirming that other potential explanations do not account for the findings. The results yield three main conclusions. Firstly, attributions for new product out-ofstockinfluence consumer compensatory consumption: algorithmic-error attributions are more likely to trigger such consumption than human-error attributions. Secondly, lack of control and belief in conspiracy theories jointly mediate this relationship. Thirdly, consumers’ situational and dispositional persuasion knowledge each moderate the effects of attributions for new product out-of-stock on lack of control and belief in conspiracy theories. Persuasion knowledge also determines which mediating mechanism dominates. Among consumers with greater persuasion knowledge, lack of control is more likely to be the dominant mediator; among those with less persuasion knowledge, belief in conspiracy theories is more likely to dominate. None of psychological envy, task completion, psychological disgust, or anticipated regret provides a viable alternative explanation. These findings deepen and extend theories of new product out-of-stock and consumer compensatory consumption, while also contributing to the theory of perceived control, conspiracy theory, and persuasion knowledge theory. They also offer important managerial implications for manufacturers managing new product out-of-stock, consumers responding to them, and regulators overseeing the market.