Who’s Telling the Story? The Effects of Identity Cues and Advertisement Type on Chatbot Advertising
Yuan Sun (Ph.D Candidate); Jin Chen (Ph.D Candidate); Feier Chen (Ph.D Candidate)
Dr. S. Shyam Sundar
Chatbots, as one type of AI that businesses see as a game-changer, have a real potential on customer experience. While chatbots are widely utilized to deliver promotion messages to target customers, how to enhance user experience and the consequential acceptance of chatbot advertising has been a challenge. The present study looks into how to enhance the effectiveness of chatbot advertising by coordinating chatbot identities and advertising content.
RESEARCH QUESTION / HYPOTHESES
RQ: For chatbot users, controlling for product involvement, what is the relationship between chatbot identity cues, ad types and chatbot evaluation, attitudes toward the adverting and patronage intention?
H1: Chatbot with human name identity cues will lead to higher levels of perceived homophily.
H2: Higher levels of perceived homophily through chatbot identity cues will lead to more favorable attitudes towards (a) the chatbot, (b) the push-in ads and (c) greater patronage intentions for the product.
H3: Narrative ad pushed by a chatbot will lead to greater level of (a) transportation (b) message involvement.
H4: Higher level of transportation through ad types will lead to greater positive attitudes towards (a) the chatbot, (b) the push-in ad and (c) greater patronage intentions for the product.
H5: Higher level of message involvement through ad types will lead to greater positive attitudes towards (a) the chatbot, (b) the push-in ad and (c) greater patronage intentions for the product.
H6: Narrative ad type of chatbot push-in ads from a chatbot will lead to less perceived intrusiveness.
H7: Higher level of perceived intrusiveness through ad type will lead to less positive attitudes towards (a) the chatbot, (b) the push-in ads and (c) greater patronage intentions for the product.
H8: Chatbot profile and ad type will jointly influence people’s evaluation of the chatbot through expectancy violation.
We conducted a 2 (Chatbot identity cues: machine name vs. human name) × 2 (Ad types: factual vs. narrative) between-subjects online experiment (N = 94). After giving consent to participate in the study, participants were randomly assigned to the four conditions and interacted with a chatbot via Messenger. Regardless of the question order, after asking the chatbot about the “day”, “date” and “weather”, participants were exposed to a print ad of a digital camera, followed by the link of the second questionnaire after one-minute exposure. Participants then finished the questionnaire and completed the whole study.
No significant main effects of chatbot identity or ad type were found with the study sample. While transportation was found to mediate the effects of ad type on attitudes toward the ad and patronage intention, narrative ad was found to lead to lower level of transportation, which were opposite to our hypotheses (see Table 1). Exploratory analyses generate interesting findings. First, identity cue and ad type interactively affect perceived intrusiveness, suggesting a chatbot with a human (machine) name is rather congruent with narrative (factual) ads (see Figure 3). Second, males in general report greater social attraction of the chatbot, ad transportation, message involvement, and patronage intention than females. Furthermore, the three-way interactions between identity cue, ad type, and gender show that males are more involved in the message than females when a chatbot with a human (machine) name pushes factual (narrative) ads (see Figure 4).
The interaction effect between identity cue and ad type on perceived intrusiveness show that chatbots with human (machine) names may be more congruent with narrative (factual) ads, leading to lower perceived intrusiveness when pushing the congruent ads. The results echo the language expectancy theory which suggests that people’s expectations of experiences and expressions in interpersonal communication are shaped by communicator characteristics.
Our exploratory findings showed that gender not only has main effects on perceived social attraction, transportation, message involvement, and patronage attention, but also interactively influences message involvement together with identity cue and ad type. Males report more favorable evaluations of the chatbot (i.e. social attraction), the ad (i.e. transportation, message involvement), and the product (i.e. patronage intention). Males are also more involved in reading the message when identity cue and ad type are not congruent. Since males and females are known to adopt different information processing strategies, males tend to rely on heuristic processing and base their perceptions on cues that are salient in the context. Thus, they might be more influenced by the information presented in the stimuli including the chatbot conversation and the ad. The other explanation for the gender effects is that gender influences people’s attitudes toward technology usage. Females’ technology adoption decisions are more influenced by subjective norm and perceived behavioral control, while males’ decisions are more influenced by their attitude toward using the new technology.
For more details regarding the study contact
Dr. S. Shyam Sundar by e-mail at email@example.com or by telephone at (814) 865-2173