Consumption is continuing to grow exponentially, with global B2C e-commerce sales expected to reach USD 1.92 trillion in. 4 The number of online consumers has never been as high: in 2013, 191.1 million US citizens browsed products, compared prices, or bought merchandise online at least once. 5 In this context, there is a crucial need for companies and brands to help consumers in their website visits. Online virtual agents (OVAs)–computer-generated characters designed to interact with users by simulating human appearance and behaviors through artificial intelligence 6 (Payne et al., 2013)–lately have been recognized as an integral element for managing more effective and efficient online one-to-one customer interactions. Recent innovations in virtual agent technology enable retailers and companies in the service industry to offer customers the possibility of interacting with such embodied conversational agent that assist them online when they are in the need of specific information (Slim Ben Mimoun and Poncin, 2015). Such virtual agents are able to fulfill online consumers' desire for personal attention from salespeople (Cox et al., 2005) and provide them with improved self-service capabilities and web navigation experiences that go beyond basic frequently asked question (FAQ) interactions. At the same time, OVAs also offer benefits for the companies, enabling them to maintain relationships with customers, capture market intelligence, and increase cross-selling opportunities and revenues (Köhler et al., 2011).

Virtual agents are augmenting and replacing human support personnel, and research recently has started to investigate the impact of OVAs on online consumer behavior. Virtual agents have been shown to exert positive effects on important variables, among which are trust (Keeling et al., 2010, Nan et al., 2006), attitude (Sivaramakrishnan et al., 2007), positive emotions, and purchase intentions (Holzwarth et al., 2006). Research also indicates that physically attractive and OVAs differ in their effects, with attractive OVAs being more effective at moderate levels of product involvement and online agents being more effective at higher levels of product involvement (Holzwarth et al., 2006). The social cues inherent in OVAs have also been examined and have been shown to influence website socialness, consumers’ affect, flow and patronage intentions, leading to higher firm-level performance, especially for newcomers on the website (Köhler et al., 2011).