Live streaming commerce (LSC) refers to a marketing model in which streamers provide product information and product trials to consumers through interpersonal communication relying on live streaming, thus promoting consumers’ purchase intention (Guo et al., 2021). It can be realized in two ways, including e-commerce websites integrated live-streaming functions or embedding commercial activities on live streaming platforms (Wongkitrungrueng et al., 2020). Given e-commerce platforms are characterized by high consumer concentration, live streaming shopping predominantly takes place on e-commerce platforms.

Previous studies have discussed the commercial impact of live streaming environmental cues. These features primarily fell into two categories: social factors and technical factors. These social factors are represented by Wanghong attractiveness or Wanghong trustworthiness (Park and Lin, 2020), positive evaluations towards the vendor based on their reputation, hedonic effort, price advantage, etc., as well as a crowd perception of other buyers (Leeraphong and Sukrat, 2018). It may also derive from a relationship state with the streamer, such as para-social interactions (Ko and Chen, 2020), affective intensives (Zhang et al., 2022a), swift guanxi (Men and Zheng, 2019), etc. The second stream focuses on the technical features of live streaming, including technology enablers such as visibility affordance, meta-voicing affordance, and guidance shopping affordance (Sun et al., 2019); Learning-related affordances like coactive vicarious learning and independent vicarious learning (Li et al., 2020); And IT-mediated mechanisms such as value transmission, vicarious experience learning, and product presentation (Chen et al., 2019). Recent research tends to integrate social and technological drivers. Zhang M. et al. (2022) argued that a single technical or social perspective is insufficient to evaluate the performance of an information system. From a socio-technical perspective, they proposed that social enablers and technical enables synergistically determined consumers’ attitudes towards streamers and products. Zheng et al. (2023) constructed a stream-streamer-viewer framework, advocating that the social presence and interactivity enabled by the stream, along with the attractiveness and expertise of the streamer, stimulated flow experience and led to continuous watching and purchase intention.

Previous research has yielded valuable insights for us. However, social influence is often interpreted as the presence and the number of other viewers (Leeraphong and Sukrat, 2018; Ko and Chen, 2020; Zhang et al., 2022b), the features of streamers (Wang et al., 2018; Hou et al., 2019; Ko and Chen, 2020; Park and Lin, 2020; Xu et al., 2020a; Zheng et al., 2023), and the action and enthusiasm of sellers (Leeraphong and Sukrat, 2018). Besides, as for the internal mechanism of environmental cues inducing behavior, previous research has emphasized the mediating role of relationship status, flow experience (Dong et al., 2023), personal characteristics (Gao et al., 2022), and community-triggered emotions (Lim et al., 2020; Li et al., 2021), few studies have focused on personal emotional changes (Xu et al., 2023) and emotional subdivided dimensions. Moreover, while the uncertainty concern surrounding experience products is a critical barrier to online consumption, few studies have offered concrete suggestions on how to eliminate it. Given these considerations, this study attempts to integrate two typical social cues and media properties, as well as personal emotions and perceptions, into the interpretation of impulse buying. Exploring their specific influence mechanism and comparing their differences in efficacy is essential and interesting.

The SOR model was proposed by Mehrabian and Russell (1974), which denied the direct effect of stimulus on behavior in the “stimulus–response” (SR) model (Watson, 1913), and maintained that environmental stimulus leads to approach or avoidance behavior through the internal state of the individual. In subsequent studies, scholars tried to categorize these stimuli, e.g., high/low load environments (Mehrabian and Russell, 1974), high/low task-relevant environmental cues (Eroglu et al., 2003), as well as marketing stimuli, platform stimuli, and other stimuli (Xu et al., 2023). Regarding the internal state, Russell and Pratt (1980) initially interpreted it as an emotional state. Watson and Spence (2007) divided research on marketing emotion into three stages: the categories approach, the dimensions approach, and the cognitive appraisal approach. They advocated that the cognitive appraisal approach provides a more complete framework to explain consumer behavior. These beliefs became widely accepted, and the SOR model also evolved into one of the most influential frameworks in environmental psychology (Yadav and Chakrabarti, 2022). The SOR model provides a concise, sophisticated framework that allows us to capture the unique environmental cues of live streaming and examine the internal state of consumers.

Perceived personalization and perceived responsiveness represent the intensity and richness of streamer interaction, respectively. Responsiveness indicates the probability and speed of the streamer’s response to questions, while individualization reflects the degree to which the streamer provides personalized communication for questions to meet the needs of users (Lee, 2005). In traditional e-commerce, consumers obtain product information through self-service searches and one-way reading. Searching for products, comparing alternatives, and evaluating information requires a lot of cognitive effort. In LSE, streamers can provide personalized suggestions for them (Sun et al., 2019). Consumers’ ability to evaluate products will be improved, and their sense of self-efficacy will be enhanced, which will make them feel happy. This deep interactive experience also enables consumers to participate in shaping the live content, thus elevating mundane shopping experiences to fashion. According to the Effectance theory, individuals have an internal psychological impulse to affect the environment (Wang et al., 2019), and when it’s successfully implemented, they will experience a positive feeling. Unlike the one-on-one interaction of traditional e-commerce, in an active studio, a streamer interacts with many customers (Wongkitrungrueng et al., 2020). Those who receive timely responses seem to successfully win this attention competition. This novel experience is likely to evoke excitement and enthusiasm among consumers. Thus, we propose:

In s-commerce, Zhou et al. (2023) confirmed that interpersonal interaction with online vendors and recommenders helps to establish swift guanxi and initial trust. This interpersonal trust can translate into trust in the product (Wu and Huang, 2023). In the studio, consumers can consult streamers until they get satisfactory answers. High-quality and fruitful discussions serve to reinforce their relationship (Guan et al., 2020). Kang et al. (2021) also believe that personalized responses in s-commerce will make consumers feel cared for and respected, thus creating a strong sense of social identity and social presence, which is conducive to closer community ties. Huo et al. (2023) proposed that in rural tourism settings, interaction between residents and tourists can foster mutual understanding, form common beliefs, and build unity and trust. Thus, we propose:

Peer interaction refers to the extent to which peers’ opinions and behaviors are visible through live streaming (Chen et al., 2011). In LSE, consumers with similar preferences gathered in a studio. To encourage them to stay in the studio until the last minute, streamers often offer special discounts in a countdown following the introduction. Peers adding products to shopping carts and processing orders can create a competitive atmosphere that may evoke emotional experiences. Wu and Gao (2019) believe that peer interaction in online tutoring helps to find common opinions and differences and co-construct new knowledge. The majority of consumers do not directly engage in live interaction but focus on the public screen (Chen et al., 2019). Consumers may get the information they want from peer inquiries, discover unique problems that they have not thought about, or increase the path of unexpected product discovery (Wang and Wu, 2019). These valuable interactions may contribute to heightened pleasure levels. Since consumers can find emotional resonance from similar views, obtain a sense of belonging and self-enhancement, and obtain the effect of vicarious learning from observation to acquire more product knowledge (Li et al., 2020). Zhao et al. (2019) believe that spiritual resonance and coordination with members of the community can bring great enjoyment. Thus, we propose:

According to the Uncertainty Reduction theory, people in an uncertain environment may adopt proactive strategies, such as observing the behavior of similar others to achieve equilibrium (Knobloch, 2016). In the studio, peer entries, inquiries, thumbs up, comments, etc. are shown scrolling in real-time, providing opportunities for observation, making the transaction process more transparent, and creating a sense of social presence with others. Lu et al. (2016) believed that websites with high social interaction levels convey more information and enhance environmental transparency, thus restraining untrustworthy behaviors and making consumers feel more secure when making decisions. Leeraphong and Sukrat (2018) believe that perceived crowds will increase trust in merchants. Liu et al. (2023) also confirm that the transparency of virtual environment information can increase consumers’ trust in product safety and suppliers. Thus, we propose:

Vividness is defined as the richness of the media and the degree of sensory stimulation during the presentation of the product in live streaming (Sheng and Joginapelly, 2012). Arora et al. (2023) discovered that through the multi-sensory stimulation of vivid information in social media advertisements, users can enjoy additional happiness and generate empathy. In LSE, streamers walk around, zoom in, and explain various angles and details of the products. Consumers can obtain many non-verbal cues, such as exaggerated facial expressions, expressive body language, and vibrant tones. These visual and auditory cues may manipulate the viewer’s emotions. Yim et al. (2017) believed that media-driven vividness enriches consumers’ imaginative processes, allowing them to transcend the constraints of their physical environment, partake in the exploration of new products, and obtain adventure-like experiences. Hess et al. (2009) suggested that vividness fosters a friendlier, warmer environment. The rich sensory experience in LSE may satisfy consumers’ psychology of seeking entertainment. Thus, we propose:

Vivid media information involves more human senses. In LSE, streamers can provide more abundant sensory cues. According to the cross-sensory compensation theory, the loss of other senses can be compensated by the visual and auditory senses (Mou et al., 2024). The vividness brought by multi-sensory channels can shape the experience of more specific simulated trial products (Huang and Chung, 2024), thereby reducing product uncertainty. Vividness can also help consumers bridge existing information and prior knowledge to form a composite picture and increase understanding of product quality (Hsu et al., 2024). Besides, in s-commerce, mental imagery vividness can enhance cognitive and affective social presence (Vazquez et al., 2023), which will provide opportunities for trust-building. Thus, we propose:

Realness denotes the degree to which products and shopping scenes are authentically reproduced in live streaming (Rubin, 1981). Consumer quest for authenticity. Live streaming makes them feel truly connected with the products. Augmented reality, as discovered by Hsu et al. (2024), imbues consumers with the sensation that products are just in front of their eyes, fostering instant gratification. The extraordinary experience brought by realness can create a flow state for theatergoers (Aykol et al., 2017), which is positively correlated with arousal (Sreejesh et al., 2020). The near-real product interaction facilitated by live streaming may enhance the intensity of positive emotions. Long and Tefertiller (2020) also believe that the more real information viewers get from live streaming, the more fun they will have. Thus, we propose:

Pictures and videos can be edited afterward, so there is often a huge disparity between the “buyer show” and the “seller show.” LSE enables the viewer to connect with the scene in real-time, mitigating the loss of information in the process of multi-level transmission (Tong, 2017). Advanced display technology enhances consumers’ ability to capture and understand the products (Hsu et al., 2024). This direct interaction with the product can alleviate consumers’ concerns regarding information asymmetry. The spatial presence brought by authenticity may shorten the physical distance of consumers and increase their trust in sellers. Wolf (2020) also believes that realness plays a pivotal role in cultivating trust when using social media for product promotion. Thus, we propose:

Pleasure and arousal represent two positive emotional states. Pleasure denotes the degree of joyful and happiness that users feel from watching live streaming (Stoyanova et al., 2015; Chan et al., 2017), while arousal means the degree of excitement, stimulation, or activity users feel (Stoyanova et al., 2015). Urge to buy impulsively refers to a sudden, spontaneous desire to buy something (Beatty and Ferrell, 1998). Mattila and Wirtz (2008) believed that a high level of excitement would destroy self-control. Because positive emotions cause people to process information in a more heuristic way (Griskevicius et al., 2010), and to reward themselves more generously. They expect instant gratification to maintain them in a positive state (Floh and Madlberger, 2013). Dhurup (2014) claimed that beyond economic motives, the desire for pleasure often drives impulsive purchase. Destari et al. (2020) maintained that people with positive emotions are more willing to spend time browsing websites, thus increasing the chances of impulse buying. Thus, we propose:

Perceived uncertainty refers to the extent to which buyers cannot accurately predict the outcome of a transaction due to product or seller uncertainty (Dimoka et al., 2012; Chen and Huang, 2013). Impulse buying still involves thoughtful decision-making. Research on young Swedish shows that online fashion consumers strive for rationality even amidst emotional influence (Lidholm et al., 2017). Dholakia (2000) argues that individuals exposed to stimuli also generate cognitive responses, trying to identify any constraints in the environment. These constraints may lead to resistance or inhibition of impulsive behavior. In LSE, if consumers are unsure whether a product is suitable for them and whether the quality of the product is consistent with the description of the streamer, these concerns may hinder their urge to buy. Thus, we propose:

To control for any potential effects on urge to buy impulsively, we added impulse buying tendency, age, gender, and income level as control variables.

The questionnaire was developed by adapting scales that have been verified in previous studies with slight modifications tailored to the LSE context. For all measures, a five-point Likert scale was used with anchors ranging from strongly disagree “1” to strongly agree “5.” Since the survey was conducted in China, we used a forward-backward-translation approach to ensure consistency with the original scale (Sun et al., 2019). The finally formed measurement scales are shown in Appendix 1.

This research launched an online survey through the “https://www.wjx.cn/,” inviting users who have participated in Taobao Live to support the survey. IP address restriction was set up to prevent duplicate responses. To ensure respondents have the same understanding of LSE, we introduced Taobao live at the beginning of the questionnaire and provided video links and screenshots with annotations for respondents in need. A total of 458 samples were received. According to pre-screening questions, abnormal response time, regular answers, and reverse-coded questions, 403 valid samples were screened out for further analysis. The sample size meets the criterion of Chin (1998).

Table 1 lists the demographic information of the respondents. The statistical results demonstrate that the sample is representative. Consistent with the research of Xu et al.’s (2020b) research, young and highly educated people are the dominant force in social commerce consumption. The gender distribution closely aligns with the data released by the Foresight Industry Research Institute (Chen, 2021).

SPSS25.0 and SmartPLS3.2.8 (Ringle et al., 2015) were adopted for data analysis. PLS (partial least squares) is superior to multiple regression. It has a minimum requirement on sample size, can process non-normally distributed data, and deals with second-order formative variables (Sun et al., 2019). Streamer interaction is a formative higher-order construct, so its parameters are estimated by a repeated indicator approach and two-stage approach (Ringle et al., 2012). The repeated indicator approach is used to get the latent variable scores of two lower-order constructs, then these scores serve as manifest variables for the measurement of streamer interaction.

This study aims to reveal how social and media cues induce impulse buying in LSE. It proved that these impacts are mediated by consumers’ emotions and perception, and many interesting differences were discovered regarding the efficacy in the subdimensions of these cues.

As for social cues, contrary to expectations, streamer interaction cannot evoke an arousal experience. This may be because over-responding leads to core rigidity and learning myopia (Zhou et al., 2018). Perhaps streamers failed to provide unique content, as consumers become more experienced with watching live, they may expect more novel product knowledge and more entertaining engagement. It may also be that Taobao Live encourages shopkeepers to act as streamers, which, unlike those platforms characterized by the web celebrity model, cannot achieve the fan effect. Besides, unlike streamer interaction, peer interaction cannot effectively reduce perceived uncertainty. This is inconsistent with the discovery of Pandey et al. (2018), who argued that people’s sharing behavior on social media helps reduce risks and eliminate uncertainty. This may be because the text chat box allows large-scale participation, “noise” or the danmaku information updates too fast, limiting the depth of communication. Another possibility is that some sellers’ hiring of “water army” or fake fans undermines the credibility of peers’ information.

In terms of media cues, vividness cannot alleviate perceived uncertainty, although consumers can obtain more product information (Cai and Wohn, 2019) and have a closer psychological distance (Wu and Gao, 2019). This may be because previous research experimentally manipulated vividness (Vonkeman et al., 2017), producing information focused on the product itself. In LSE, streamers, products, viewers, multi-objects are involved in the studio (Zhang et al., 2022a). Vividness independent of the product may cause distraction and interfere with cognitive formation. Flashy elements, such as rockets or streamers’ overly flamboyant “performance” also tend to make the thematic argument get lost in the irrelevant background, which damages the signal transmission and hinders the formation of trust. Besides, realness cannot provoke excitement and passion as effectively as vividness. This is inconsistent with Verhagen et al., (2014) study, in which virtual mirrors caused product likability (including excitement and enthusiasm). One plausible explanation is that without the support of innovative technologies (such as AR), authenticity only met consumers’ most basic requirements for information quality. Alternatively, authenticity is a multi-dimensional concept encompassing streamer, product, and scene (Sun et al., 2022), where explicit authenticity at the visual level cannot inspire deeper psychological involvement.

Regarding the organism, emotions play a more important role in impulse buying than perception. This means that akin to social live streaming platforms, people expect relaxation and emotional enjoyment from their LSE consumption. Another possible explanation is that LSE predominantly involves experiential products, and complex shopping tasks encourage consumers to rely more on heuristic information processing patterns. What’s more, pleasure exerts a stronger impact than arousal, indicating that consumers tend to avoid overstimulation, which makes them feel burdened (Pelet et al., 2017).