Since the earliest science-fiction movies and books, talking computers and robots were portrayed as something to fear. From the ominous-sounding “I can’t do that, Dave” from 2001: A Space Odyssey to the trope of the countdown to self-destruction (always stopping at 1), computers have been the perennial bad guy. Now that our devices really do talk, it’s vital that we help them shed their reputation. The best way to do that is to create a nicer, friendlier, trustworthy voice user interface (VUI). More than anything, what people want from their VUI is a more human experience designed to help them as a friend would.
We may not notice it, but as people, we look for characteristics and personality in nearly everything we interact with. In fact, we rely on speech to evaluate people on honesty, trustworthiness, overall intelligence and education (knowledge). That’s why it’s important for brands to always keep human nature in mind when designing a VUI. Even if you don’t focus on building a specific personality for your VUI, having one is inevitable—your users will naturally personify it. People dislike talking to cold and calculating robots and as a result, many innovative brands are already being more thoughtful when it comes to their voice AI’s personality.
Defining your personality should be high on your list when you start planning to build conversational experiences for your customers. It needs well-written dialogue, small talk, empathy, humor, and emotional intelligence that caters specifically to your users. The actual voice and tone of the conversation should be in sync with your brand voice, too—and that is one of the biggest hurdles in designing a VUI successfully.
Interacting with a voice assistant should be a seamless, frictionless conversation that’s comfortable and simple. These best practices can help guide you as you build your voice interface.
According to Will Hall, Chief Creative Officer at RAIN, “One of the biggest challenges with voice assistants is that they sound like humans… almost.” The almost problem is more profound than it may at first seem, though. Because the voices are almost human, we as users tend to graft human expectations onto these interfaces that simply aren’t yet up to the task of conversing intuitively, understanding emotions or building true empathy. As architects of these experiences, we must find ways to set appropriate user expectations and bridge the gap with thoughtful design.
How do you want your VUI to represent your brand? If you have a company mission statement, that’s a good place to start. What values do you want to convey? Once you have that guidance, use your imagination. If you were to personify your VUI as the ideal employee for your company, how should that employee sound? How do they carry themselves? What’s that person’s attitude when talking to customers? What kinds of words do they use?
Next, turn your attention to the users. Defining your customers’ identity is the first step toward building your VUI’s personality. Write down what you think this voice assistant’s job description is. Answering simple questions like, “Are our clients professional or leisure users?” and defining their age range, gender, geographic spread, and language can help you build that job description to meet the needs of customers better. When you build this job description and understand its purpose and audience, creating its personality will be much easier. For example, if the voice assistant provides account information, it could be more formal and knowledgeable, but if it offers kids’ entertainment, it should be friendly, funny, and use age-range appropriate dialogue.
Choosing your VUI’s gender can be tricky. Like everything in life, you should put this to the test and see how people react. Start by simulating an interaction and note what the users’ expectations are with regard to gender. While you don’t want to foster gender stereotypes, think carefully about whether your company’s voice is best represented with a male, female, or a genderless voice. Opinions vary about what’s best of course.
One school of thought urges voice designers to challenge gender stereotypes. IBM’s Jeopardy-playing AI Watson speaks with authority and a male voice. But Alexa, built to be helpful and supportive, is female. Apple’s Siri and Google Assistant offer users the opportunity to choose the preferred voice, gender, and accent of their AI, but that choice sometimes runs the risk of diluting brand recognition.
Lauren Golembiewski, CEO and Co-founder of Voxable advises clients to be ready for controversy no matter which gender you choose to represent your brand. “The intersection of a social issue and technology is always going to be rough,” she said. “The best way to be prepared for it is to have a good answer for why you chose the gender you did. Whatever your decision, if it ties into your core mission authentically, people won’t judge.”
Most importantly, design your VUI’s personality at the very beginning of the conceptual stages with well-written copy that caters specifically to the target user. That’s why it’s so important to create a persona that resonates with your audience and also with your brand.
At Soundhound Inc., Director of Product Design Michael Mahmood advises taking a holistic approach when creating voice experiences. "Know your audience, understand their motivations, and develop a deep sense of empathy for their pain points when it comes to using voice. Once the user persona is clearly defined work tightly with marketing to ensure the tone and communication style aligns with the overall branding strategy. Also, work closely with the engineering teams to make sure different uses cases are accounted for as voice and speech can get very complicated."
The SoundHound design team is continuously learning and listening to their users as the voice revolution moves ahead. "We at Soundhound Inc. know we're just scratching the surface when it comes to the potential power of voice, and we're excited to lead the way with meaningful voice-powered experiences for our customers."
“If there’s anything that will connect your users with your voice assistant, it’s small talk,” advises Rachel Batish, vice president of product at Audioburst and author of Voicebot and Chatbot Design. “Either as side-interactions or as part of the main discourse, it’s the small talk that people will talk about and remember.” Human-like chatter is a feature that both demystifies AI and warms people to it. Users become aware that they can simply talk normally to a VUI, and eventually come to depend on it. But VUI, like any technology, isn’t perfect, so designing humorous or witty quips and come-backs are valuable in keeping the user smiling when the system can’t deliver the desired information or action. “The more chatter sentences and events you put into your VUI the better it is,” Batish said. “While it’s not a huge investment for you, it makes a huge difference for the user.”
SoundHound Inc. worked closely with writers on the marketing team and found that people like a little bit more fun than robotic responses. We found that being friendly, casual but polite, and not too verbose goes a long way. Mahmood is inspired by tiny touches used by other companies that keep the user encouraged even if the system can’t provide instant gratification. “Think of that little dinosaur you sometimes see in Google. It’s telling you it can’t find what you’re looking for, but you don’t mind because it’s so cute. You just try again. That’s what we aim to do with voice. We want to make even a bad experience delightful. That helps users to be a little more forgiving. We don’t want to discourage them from getting better at (using voice assistants). Of course, we also want to be helpful. We try to anticipate users’ needs. We might provide an answer and a follow-up question like: “Is this what you’re looking for?”
While it is important for a VUI to be approachable from the very beginning, it should not repetitive. It’s important to provide your VUI with the ability to anticipate moments and keep a conversation going during the interaction flow with users. It’s essential to add variety to its language and shape its ability to retain information so it can constantly adapt to the user’s requests.
In person-to-person interactions, building up a sense of empathy is essential in establishing a proper and well-meaning relationship. For the sake of your user base, your VUI should be no different. In many cases, a general tone of positivity and affirmation goes a long way in keeping your users engaged even in worst-case scenarios. For instance, if the VUI is unable to fulfill a user’s request, it should take full ownership for the lack of understanding and then follow up with corrective suggestions. This helps to build confidence and trust with the user, encouraging additional interactions and queries because there is now an increased belief that the outcome will be beneficial versus having no value.
Also, remember to have your VUI remain sensitive to the user’s answers. Adjust subsequent options based on how your user answered previously, anticipate particular errors based on functionalities that your VUI does not have, and acknowledge any difficulties within the conversation flow to encourage the user’s patience and perseverance.
At the very base of a VUI is the way it speaks and responds to the user. The more personable its tone, the more it helps to increase brand affinity. When it comes to writing the dialogue, it’s best to opt for the most “human” interactions possible. This means including occasional small talk as well as humor and even general reference to pop culture among the VUI’s many other functionalities. Jokes or witty remarks help the user to become more comfortable and acquainted with speaking and interacting with the VUI, potentially increasing further use and engagement. In fact, according to a study by Adobe Digital Insights, 53% of U.S. smart speaker owners use their smart speaker to ask “fun questions.”
The Mercedes-Benz infotainment system MBUX’s ability to crack a joke every now and then allows it to be more entertaining and “human.” Senior Engineer Mihai Antonescu advises brands to “never assume that you have enough utterances as that’s a mistake that you can easily make in the beginning. Don’t just think about “How would I ask this?” Because if you ask your friend, he will probably ask it very differently. So keep asking people about how they would ask for one specific intent and then constantly try to improve your language model,” he adds.
You can also take this small clip of Mercedes-Benz’s MBUX voice system for instance. Providing occasional moments such as these throughout the VUI experience encourages users to be more forgiving during moments in which the interface is unable to complete tasks or answer certain questions. The power of natural, friendly conversation should not be underestimated.