Purpose This study examines whether artificial intelligence (AI) personalisation depth and algorithmic transparency relate to continued engagement with streaming platforms through consumers’ psychological ownership of the recommendation stream, and over what range of privacy concern the transparency route holds. Design/methodology/approach A 14-day, two-wave panel of streaming subscribers in Türkiye (N = 450) was analysed with Partial Least Squares Structural Equation Modeling (PLS-SEM) and re-estimated with covariance-based SEM. A 2 × 2 between-subjects experiment (N = 236) manipulated both features. Findings Personalisation depth showed an indirect-only mediation pattern through ownership in the estimated model, with a direct path practically equivalent to zero within a pre-specified bound; the pattern replicated under random assignment. Ownership carried 45 per cent of the association between transparency and engagement, and a direct path accounted for the rest. The transparency–ownership link was no longer significant above a privacy-concern level of about 4.5 (7-point scale), a region holding roughly two in five respondents. Platform involvement strengthened the personalisation route; no group difference was detected for transparency. Research limitations/implications Participants were subscribers aged 18–35 in three Turkish cities, engagement was self-reported and ownership and engagement were measured on the same occasion. Practical implications Ownership measures can complement viewing and churn metrics when personalisation is evaluated, and layered explanations let users choose how much detail they see. Originality/value For interactive marketing, the study recasts the AI feed as a possession consumers co-produce, not exposure firms deliver. It tests, and does not assume, whether design features keep a direct relationship with engagement once ownership is modelled.
Live-streaming commerce has rapidly emerged as an important form of e-commerce. One of its most distinctive and critical features is streamer-viewer interactivity. However, how the key construct of perceived interactivity interacts with other cues in live-streaming commerce to influence consumer responses remains largely unexplored. One particularly prevalent linguistic cue used by live-streamers is assertive language. This study investigates whether and when perceived interactivity shapes consumer reactions to assertive language in live-streaming commerce. The authors test the hypotheses across five studies. Studies 1–3 employed a 2 (language style: assertive vs non-assertive) × continuous (perceived interactivity) between-subjects design. Study 4 employed a 2 (language style: assertive vs non-assertive) × 2 (interactivity: high vs low) between-subjects design. Web Study 1 examined a theory-driven boundary condition of this interaction effect – time pressure. This research uncovers a robust interaction effect of assertive language and perceived streamer-viewer interactivity in live-streaming commerce. Specifically, when consumers perceive lower (vs higher) interactivity, live-streamers’ assertive language enhances trust toward the product and downstream consumer responses. Trust toward the product mediates this effect on downstream responses. This study identifies an interaction pattern between assertive language and perceived interactivity, extending research on streamer-viewer interactivity, assertive language, and consumer trust in live-streaming commerce. Based on the findings, this research provides implications for live-streamers on how to adapt their use of assertive language to different levels of streamer-viewer interactivity.
Purpose As AI-generated images proliferate on social media, brands face trade-offs in using them for corporate social responsibility (CSR) communication. This study examines when and how AI-generated images versus real human photos differentially influence consumer responses to CSR posts, depending on message framing. Design/methodology/approach Two experimental studies were conducted. Study 1 tested the effect of image type (real human vs. AI-generated) on human likeness, empathy, CSR authenticity, purchase intentions, and social media engagement. Study 2 adopted a 2 (image type: real human vs. AI-generated) & times; 2 (message framing: action-oriented vs. informational) between-subjects design, measuring CSR authenticity, perceived information quality, purchase intentions, and social media engagement. Findings AI-generated images yielded lower perceptions of human likeness, empathy, CSR authenticity, and purchase intentions than real human photos, with no difference in social media engagement. Study 2 revealed a significant interaction between image type and message framing on CSR authenticity: real human photos were more effective under action-oriented framing, whereas under informational framing the gap narrowed and favored AI-generated images. CSR authenticity mediated these effects on purchase intentions and engagement, and in the informational condition, perceived information quality further enhanced these outcomes. Research limitations/implications This research advances understanding of consumer reactions to AI-generated content in CSR communication. By integrating visual and verbal components, it clarifies how and when AI-generated visuals can enhance or hinder engagement with socially responsible brands. Originality/value This study extends research on AI-generated advertising by showing that visual and verbal message elements interact to influence perceived authenticity and consumer responses.
Purpose-Digital marketers routinely decide whether products should appear in contextual usage settings or isolated on neutral backgrounds, yet platform practices and prior research offer mixed guidance. This research examines when contextual backgrounds improve or undermine digital marketing effectiveness by identifying product type and consumer experience as boundary conditions. Design/methodology/approach-The framework is tested across three studies. An Instagram content audit of 1,000 posts from top fast-moving consumer goodsbrands documents the managerial puzzle. A Facebook field experiment with 101,200 impressions tests whether product type changes contextual-background effects on real advertising behavior. A controlled experiment with 431 participants examines whether consumer experience moderates these effects. Findings-The Instagram audit reveals no clear industry consensus: 53% of product posts used contextual backgrounds, and 47% used noncontextual backgrounds. In the Facebook field experiment, contextual backgrounds increased hedonic product click-through rate (CTR) by 0.06 percentage points, a 67% relative improvement, but decreased utilitarian product CTR by 0.05 percentage points, a 56% relative decline. These effects emerge among experienced consumers but disappear among inexperienced consumers, consistent with the idea that relevant product experience enables schema-based background effects. Originality/value-By integrating schema theory with processing fluency theory, this research challenges the assumption that contextual backgrounds are universally beneficial. Product type qualifies contextual background effects, and consumer experience helps explain when they emerge. Findings offer actionable guidance: use contextual imagery especially for hedonic products targeted at experienced consumers, favor neutral imagery for utilitarian products or less experienced audiences, and guide artificial intelligence-driven creative optimization by matching backgrounds to product and user signals.
Purpose-Drawing on attribution theory and an interactive marketing perspective, this research examines how homophonic (vs. non-homophonic) store names, conceptualized as a form of offline consumer-message interaction, influence brand image through inferential and affective pathways and how informational cue type moderates these mechanisms. Design/methodology/approach-Across three scenario-based experiments, this research tests the effects of homophonic versus non-homophonic store names and examines the moderating role of informational cue types, including image-based, text-based and no-cue conditions. Findings-Homophonic names enhanced brand image via a dual-path mechanism: positively valenced persuasion knowledge, reflecting favorable inferences about merchant motives and perceived pleasure from meaning resolution. Image-based cues most clearly amplified both the rational and affective pathways. Originality/value-This research extends interactive marketing and consumer psychology literature by demonstrating how consumers construct brand meaning from offline messages and providing practical implications for tourism retailers' naming and signage strategies.
