Field Guide for Interacting with Voice User Interfaces

By Antonio Holguin, Associate Principal Designer

The current forms of Voice User Interfaces (VUI) can be found right in our hands, on our nightstands, or on our kitchen counters, ready to be called into action. Generally, these VUIs are the user’s connection to a simple artificial intelligence (a.k.a., computer) that responds to commands and perform simple tasks: turning lights on, running search queries, or adding to a shopping list, as a few examples. Our current set of simple, and readily available, VUIs (Amazon’s Alexa, Apple’s Siri, Google’s Assistant) are not quite fully “conversational.” To be more conversational – rather than simply order-based – they would need to fully understand nuances and inflections in human speech, as well as be able to creatively and proactively draw conclusions, ask questions, and provoke further responses based on context.

Even with this current limit to conversational abilities, VUI devices are becoming increasingly common in our everyday lives. Their availability and use will, most likely, only continue to grow. Yet, VUI design is only on the cusp of becoming a popular subject. Now is a great time to get ahead to learn and explore how to design and build VUIs. To help you get started, we’ve put together some universal concepts to keep in mind when building your voice enabled apps. (Note that each platform (Google’s Assistant, Amazon’s Alexa, Apple’s Siri) have specific guidelines for responses, errors, and flow, so be sure to keep those as reference.) Understanding these concepts now will help us, as VUI designers, be prepared for future Conversational User Interface artificial intelligences.



Recognize Variances of Speech | There are many ways to say the same thing. Through commands, questions, sentence structure and form, speech can be incredibly nuanced between individual people. Be sure you try to recognize as many ways to speak as you possibly can. Have a few people try out you app to get to know how people are speaking to it.

Listen for AmbiguityExpect that some commands from the user might not be clear or precise. Our everyday speech with others can be limited to very simple or ambiguous phrases. The most helpful voice apps pick up on generalities and respond appropriately; either asking for clarification or, if possible, being able to interpret user intention as accurately as possible.

Anticipate Needs | Expect that your users will be asking certain things. Be ready to have a response for any current or upcoming scenario. If necessary, queue up any information that might come next based on the context.



Try to Be Human | When designing your VUI responses, try to imagine what you might say to a friend in the same situation. Try to create your VUI to sound as relatable as possible. Test by having multiple people read your responses out loud to others. Get a feel for what sounds natural, and tweak–or add to–responses.

Be Brief, Be Clear, Be Relevant | Respond with appropriate and relevant content to the information requested. Concise responses reduce cognitive strain on human users. Don’t over explain, making even basic information unintelligible. Keeping responses short will also keep your users engaged.

Acknowledge and Cooperate | Let the user know you understand their commands or requests. Give your users useful updates to the inner workings of the system, even when things aren’t working. Don’t fail silently.

Be Polite | A little “please” and “you’re welcome” goes a long way to an AI feeling more human.

Match Formality | Try to match the same formal (or informal) language with which the user is speaking. For instance, if the user says, “thank you”, respond with “you’re welcome”, “thanks” with “of course” and so on.

Be Helpful | Ask the user if more information or work is required before exiting your app. At all times, especially when next steps can be vague, give the possible options for next steps.

Use Common Language | Speaking in everyday vernacular (in the language of your user) helps to connect your user and VUI. For example, in English, not using contractions can sound stale or robotic.

Use Gentle RemindersIf a user utters a command that is not recognized, gently remind them what they can do at that point.

Don’t State the Obvious | If your VUI controls a physical device, say lights, be mindful that the user may be in immediate vicinity and can witness the lights turn on or off for themselves. Stating the obvious, “lights are on,” would sound redundant and unnecessary. If your device has further control, a more apt response might be “should I change the brightness?”

Give Enter and Exit CuesLet users know when your app opens or closes. They may not be able to see the device to understand any visual cues.


Download our Field Guide for Designing for Voice User Interfaces

These concepts will be most useful when defining and polishing the interactive responses in your VUI app. Yet, they are also useful to take into consideration at every point in your design process–from rough draft scripts to prototyping to final, production-ready development. To help you along the way, we’ve distilled this guide down to a handy “field guide” you can print out to reference while designing your next VUI-based app.


Related Posts

Hi! Let's stay in touch.
Sign-up for our newsletter!