Ten principles

While there are several established guidelines for GUIs, there are no equivalent principles for voice interfaces. This is critical as voice assistants are becoming increasingly popular and for some people are already an essential part of everyday life. Luckily the wheel does not have to be reinvented since Nielsen Norman Group has already created a thorough basis for GUIs – namely the usability heuristic for user interface design. Keep in mind they are called “heuristics” because they are general rules of thumb and not specific usability guidelines. Based on these rules, the following ten principles are derived and adapted for best use in voice user interfaces.

Feedback of conversation status

As soon as a voice command is spoken, it is no longer visible unless there is a visual interface. To keep a smooth user flow, the user must always know when it is his turn to speak. For a good user experience it’s important to use feedback to tell the user what options are currently available for interacting with the voice user interface, whether visually or by voice. There is a high risk of the user becoming confused due to a lack of feedback, this risk increases if the voice user interface has not understood the spoken input or recognition errors have occurred.

Speak the user’s language

Short voice commands and keywords significantly improve the flow of user tasks. To achieve a compelling dialogue, it is better if a voice user interface not only understands short, firm commands, but also more natural commands. Speak the user’s language using what is already known and without system-relevant keywords. Using simple language is the most attractive and accessible way to people with different backgrounds. It is recommended to learn from interpersonal communication to make these interactions as natural and intuitive as possible.

User control and freedom

Users often feel rushed to interact with a voice interface and worry about missing parts of the interaction. This leads to a common problem of frustration due to the lack of control over the voice interface. It is important that the user has control of every interaction at all times, but is also free to pass control of the interaction to the system.

Prevent and recover from errors

Avoid user errors when users interact with the voice interface. When designing speech interfaces, there are two types of elements that can be accessed: auditory and visual elements. Auditory elements are fundamental and required for every voice user interface. Use visual elements to significantly improve the conversation between the user and the voice interface. This avoids user errors and increases confidence in the user interface.

Recognition rather than recall

Users should always be able to remember voice commands – but they shouldn’t do that on his own. It is important that voice interfaces supports users in how they use the voice interface, for example how they interact with it and in showing which commands lead to which actions. Also, audio as the only output should not be preferred because it increases cognitive stress and the user has to remember long information.

Task efficiency

Flexibility is useful for productivity and user satisfaction with the voice interface. Due to the lack of keyboard shortcuts in voice interfaces, designers have to rethink how they can improve task efficiency for users. Think about how a user can perform an action in several ways instead of following a prescribed path.

Minimalism in design and dialogue

Large information typically affects the user’s ability to remember what has been said earlier through the voice interface. Do not give the user all the information at once. Provide only the most relevant information. Remember that minimalism is all about letting the unnecessary go.

Allow users to recognize and recover from errors

Let the user recognize errors and provide a fallback option. Allow the user to avoid errors by understanding how they occurred and how the user’s intentions must be properly performed so that the errors do not repeat. In the event of errors and missing information, make sure that you specify exactly what information is understood and which information is missing – try to ask the user only for the missing pieces instead of a standard saying like: “Sorry, I didn’t get that.”.

Providing help and documentation

If the user no longer knows how to proceed, always give help. At best, provide contextual help within the interaction. Users shouldn’t have to leave the current conversation to get help. Instead, help should be a built-in function available in the active conversation. This always provides quick access and support to the user, but still only when needed.

Considering how context affects speech interaction

Users often feel uncomfortable speaking on their phones in front of other people, as this is not a common form of a natural interaction. To ensure the best voice interaction, think about when and how the user will use the voice interface. Or whether it is best to avoid a voice interface utterly in some cases.