VUI: How to Design for an Invisible Interface

Written by: on June 20, 2017

This article originally appeared in the Mobile Marketer.

To those in the know, VUI is pronounced “vooey,” and it stands for Voice User Interface. It even has its own bad joke. Because voice control lacks a screen, there’s nothing to touch or look at. So it’s an interface without a “face.”

If you think it’s a gimmick, however, think again. The promise of VUI is huge. If we create intelligent auditory interfaces that understand speech and context, we’ll be able to deliver what people ask for, hands-free, and with almost no effort.

Of course, we aren’t there yet, but we also aren’t as far away as you might think. Wolfram, Siri, Watson, Alexa, and Google Assistant all have VUIs that are starting to make conversational computing possible. But much like people, they are not the same, and we should know the differences between them if we want to understand the best practices for them as a whole.

For a quick breakdown, Siri and Wolfram are step-siblings with a shared lineage. Both are glorified search bars that serve users by supplying answers to queries. Siri can take on extra tasks by accessing other apps and SDKs, but everything else becomes a web search.

Next we have the conversationalists—Alexa and Google Assistant—both of which power numerous devices and applications, such as Google’s Allo and Home, and Amazon’s Echo and (soon) Ford cars. These two can, to a limited degree, talk to you and provide functionality. They both adeptly handle search, play audio, and answer questions in a conversational way. That said, there are several differences between them. Alexa currently has an SDK and is open to third-party developers, while Google plans to do the same soon. Home’s Google Assistant can use its understanding of data to be personalized and predictive. Alexa is not as user-centric and doesn’t rely on that type of personalization.

Moving forward, we should expect VUIs to become both more conversational and widespread. Which brings up the inevitable question: How can brands get into the act? How do we make experiences that are wanted and useful, rather than annoying?

Here we have an emerging set of guidelines requiring that any assistant be conversational, natural, simple, and habitual. Let’s look at each in turn.


Human beings do not usually dictate, they converse. As a result, brands should work towards making the tone of their interactions informal and conversational, while not straying too far from the personality of the assistant itself. The more you make users deviate from their normal conversational patterns, the more difficult the interaction will be.

In addition, you have to remember the limits of conversational interaction. “Seven, plus or minus two” is how George Miller put it in a classic research paper. We only have a limited number of things we can keep in mind, so individual answers to queries or interactions shouldn’t be too long.


Most VUIs require some unnatural communication, like a wake word, so that they can understand when they’re being spoken to. However, you should make sure that your application allows users to make requests that are as natural as possible. You should not only account for the most common way people ask for something, but also as many variations as you can find. If you ask an app to show you a recipe for chicken a la king, it should show you one. But if you say, “Tell me how to make chicken a la king,” or “Find me Chef John’s chicken a la king,” both of those should work too.


We all take a mental leap when we talk to a computer, undergoing a training phase in which we learn how to use an app. That’s why great voice apps always start out simple, doing one thing well. Then, with every iteration or upgrade, you can add a new function. It’s a crawl-walk-run mentality. And remember: even as an app grows more complex, information hierarchies should still be kept shallow, as users can easily get lost.


Because your app is invisible, it can easily be forgotten. Instead, make it something people can use every day. The list skill on Alexa works well because it is something you can use multiple times a day.

Creating an interface without a face may seem a complex undertaking, but with good experience design and a willingness to take baby steps, we’’ll be well on our way to achieving the promise of VUI. We will be able to give orders, receive information, and get things done effortlessly. One day. For now, let’s work smart and together, one app at a time.

Danielle Reubenstein

Danielle Reubenstein

Danielle is the Executive Creative Director for POSSIBLE Mobile, a leading mobile development company. A trained cinematographer, Danielle has a passion for bringing users through stories of all types including mobile applications. With her deep-rooted technical knowledge and design background she has been a significant force in growing the creative practice at POSSIBLE Mobile and ensuring that all customer products go into the marketplace both looking and working beautifully.

Add your voice to the discussion: