Understanding Task Differences to Leverage the Usability and Adoption of Voice Assistants (VAs)

Abstract

AbstractVoice Assistants (VAs) are becoming increasingly popular, but evidence shows that users’ utilization of features is limited to few tasks. Although the literature has shown that usability impact VA adoption, little is known about how usability varies across VA tasks and its relation to task adoption by users. To address this gap, we conducted usability tests followed by debriefing sessions with Siri and Google Assistant users, assessing usability measures of six features and uncovering reasons for task usage. The results showed that usability varied across tasks regarding task completeness, error number, error types, and user satisfaction. Checking the weather and making phone calls had the best usability measures, followed by playing songs and sending messages, whereas adding appointments to a calendar and searching for information were the most incomplete and frustrating interactions. Furthermore, usability-related factors such as perceived ease of use and the interaction’s hands/eyes-free nature influenced task adoption. Nevertheless, we also identified other task-independent factors that affect VA usage, such as use context (i.e., place, task content), VAs’ personality, and preferences for settings. Our main contributions are recommendations for VA design, highlighting that attending to tasks separately is paramount to understanding specific usability issues, task requirements, users’ perceptions of features, and developing design solutions that leverage VAs’ usability and adoption.KeywordsVoice AssistantsUsabilityTechnology adoption