Trust in chatbots for customer service – findings from a questionnaire study
Cecilie Bertinussen Nordheim · NORA - Norwegian Open Research Archives · 2018
Recently, there has been an increased interest in chatbots.Chatbots are software systems which interact with humans through natural language.This technology is utilised and implemented in a variety of sectors, in particular for customer support purposes.Though chatbots' capabilities for efficient interaction has increased, user uptake has been lower than anticipated.A critical success factor for such uptake is user trust.However, there is a knowledge gap concerning the factors that affect trust in chatbots.This thesis aims to cover this gap by presenting a questionnaire study, including the response of 154 users of customer service chatbots.The study consisted of two parts.First, an explanatory part mainly based on the Corritore et al. (2003) framework on trust in websites (including users' perceptions of ease of use, risk, expertise, predictability and reputation).This framework was extended on the basis of related literature to include users' perception of anthropomorphism and propensity to trust technology.The studied factors were structured according to three highlevel dimensions; chatbot-related, environment-related and user-related factors.The explanatory analysis showed that users' perceptions of expertise and risk, as well as users' propensity to trust technology, explained the most variance in trust.Second, an exploratory part where the respondents in their own words reported on what they considered to be important for their trust in the chatbot.Their responses made subject to thematic analysis.The categories identified in the exploratory analysis suggested users' trust to be affected by factors partly reflecting those of the explanatory analysis (such as expertise, anthropomorphism, low risk and not trust relevant/no trust), and partly new factors (such as fast response, absence of marketing, brand and access to human operator).Based on the findings from the explanatory and exploratory analysis, a new model of trust in chatbots for customer service is proposed.The model includes chatbot-related factors (expertise, fast response, anthropomorphism, absence of marketing), environment-related factors (brand, low risk, access to human operator) and user-related factors (propensity to trust technology).