Exploring Users' Preferences for Chatbot's Guidance Type and Timing

Abstract

While task-oriented chatbots have become popular recently, conversational breakdowns are still common and will often lead to unfavorable user experiences. Guidance serves a crucial role in helping users to understand how to have better interaction with chatbots. Nonetheless, questions like what kinds of guidance to provide and when to provide guidance remain underexplored. In this study, we examined users' preferences for two types of guidance (Example-Based and Rule-Based) at four guidance timings (Service- Onboarding, Task-Intro, After-Failure, and Upon-Request). Our results show that users preferred Example-based guidance, and generally preferred guidance provided at Task-Intro. Example-based guidance appearing at Task-Intro was the favorite guidance combination for most participants. Through analysis of participants' explanations of their preferences, the strengths and weaknesses of these guidance types and guidance timings are presented. The preliminary results are based on a subset of the data (n=24). Further in-depth investigation into the underlying factors that influence users' preferences for guidance, as well as the interplay effect between guidance type and guidance timing is needed.