Talking to computers originated as a science fiction element but is now becoming a reality with the developments of Intelligent Personal Assistants from big companies such as Google, Apple, Microsoft and Amazon.This thesis aims to investigate the design aspects to get users to feel trust in voice user interfaces with the use case of performing an invoice payment. To be able to design for trust in a Voice User Interface for banks it is of utmost importance to first understand the fundamentals of a conversation between humans. Further, it is necessary to follow already existing guidelines.

A Service Design methodology was implemented with a research phase consisting of a literature study and a workshop to form an understanding from the users point of view of the problem. A mockup was made to compare two types of conversations to interact with the voice assistant.The literature study, workshop and mockup tests later resulted in a prototype created in Dialogflow, Google’s development tool for voice user interfaces. The prototype test was evaluated with interviews and a System Usability Score questionnaire. The overall opinion and grade the prototype received was that it was easy and fun to use and could add a value for when users have their hands or eyes occupied with something else at the same time. Based on the results from the workshop, mockup interviews and feedback from the users testing the prototype, design guidelines for increasing the feeling of trust were compiled. The guidelines point out the importance of a personalized application, to provide a welcoming phrase, describing instructions so the user knows what is expected from them, and for the user to be able to use catch phrases o rshort sentences when talking to the system. The system should mimmic the way the users provides their inputs and by adding a visual interface as complement to the application the user might feel more in control. These guidelines could provide a bigger sense of control and trust during the usage of the system.

Most users participating in the tests described in this thesis did not really feel entirely ready to use voice, especially not in public, yet. To increase the interest and usage it is necessary to find ways to integrate voice into the users everyday life.