Human-Likeness Realism Of Digital Assistants’ Interface: Effects On Customer Experience, Trust In Digital Assistants And Purchase Intention
Companies are increasingly incorporating digital assistants on their websites to improve the customer experience in their online frontline services. However, despite their functional advantages, digital assistants have yet to provide optimal customer experiences and rewarding interactions for firms and customers. For example, customers are prone to stop interacting with digital assistants, which affects sales processes. Emotional and social aspects in service encounters provided by human contact are essential for excellent customer service, which is relevant in online services because online platforms limit human contact. In this context, designing an optimal digital assistant interface with adequate human-like characteristics is necessary to enhance digital assistants with emotional and social cues. Although several studies have found relationships between human likeness and concepts such as customer experiences, trust, and behavioural intentions, to the best understanding of this study, more research is needed to observe the effects of different types of human likeness realism on a wide range of types of feelings, thoughts and beliefs; to look at them from a multidimensional approach; and to use a combination of affect-as-information, information-processing, social presence, human-likeness realism and uncanny valley theories. A question identified is whether making a digital assistant interface look more human-like improves its acceptance and use by providing a more pleasant experience and making it look more trustworthy.
This dissertation aims to answer the following research questions: (1) To what extent do different levels of human-likeness realism (HLR) in digital assistant interfaces influence affective and cognitive customer experiences, willingness to trust in digital assistants and purchase intention?; (2) To what extent do (a) affective and cognitive customer experiences formed from the interaction with digital assistants and (b) willingness to trust in digital assistants influence the intention to purchase from a digital assistant?; and (3) To what extent does HLR moderate the relationships between affective and cognitive customer experiences, willingness to trust in digital assistants and purchase intention?
A conceptual model was built to help identify: (1) the effects of low, medium and high HLR characteristics of digital assistants on positive and negative affective customer experiences, cognitive customer experiences, willingness to trust in digital assistants, and intent to purchase from a digital assistant; and (2) the process that explains the role of customer experience and willingness to trust in digital assistants in customers’ intention to purchase from digital assistants with human-like characteristics.
A between-subjects experiment was designed to test the effects of low, medium and high HLR in positive and negative affective customer experiences, utilitarian, immersive and social cognitive customer experiences, willingness to trust in a digital assistant and purchase intention. One-way ANOVA tests were used to test the differences produced by each type of digital assistant interface tested. Linear regressions were used to test the relationship between the variables.
This study contributes in four ways. First, High HLR is presented as a breaking point where the perception of affective CX, TRU, and PI turns into negative outcomes. Second, it was observed that, although human likeness is a crucial element in creating positive feelings and thoughts about the digital assistant, low to medium HLR characteristics did not appear to produce significant changes in affective and cognitive customer experiences, willingness to trust in digital assistants and purchase intention. The non-significant differences are potentially explained by the presence of a human likeness ceiling in the effects of HLR on dependent variables. Third, affective customer experience is a major driver of cognitive customer experience, willingness to trust and purchase intention. Fourth, although interaction with digital assistants creates a state of immersion, this does not appear to convert into higher trust nor purchase intentions. These findings suggest to marketing practitioners that digital assistants should be designed with low or medium HLR characteristics in opposition to high HLR characteristics to produce optimal interactions.