Designing for voice-controlled devices: creating a conversation

On humanizing voice-controlled services

John Jones
Design Voices
5 min readApr 20, 2017

--

With the vast improvements in voice technology and natural language processing, voice is already transforming the way we interact with digital services. According to ComScore, 50 percent of all searches will be made via voice query by 2020.

More and more companies are turning to voice-controlled services as a way to ‘humanize’ their offerings. Early experiments, though, have fallen short. All too often, the experience leads to confusing and disappointing interactions as devices don’t understand us or fail to meet our expectations of simulating a conversation.

Brands need to rethink their mission — it’s not about replacing human conversations, but simplifying interactions with products and services. Rather than imitating human conversations, they should be focused on relevant scenarios that make people’s lives easier and improve human-to-machine interactions.

Our research studies and experience designing voice services at Fjord have surfaced a key stumbling block: the way people speak to voice-enabled devices is actually part of the problem. The truth is, we don’t speak to them in the same way they speak back to us. A critical factor in designing for voice is understanding more about the intent of the person interacting and the limitations of the device itself. Adhering to a few key principals will help brands to improve the human-to-machine interaction with voice-controlled devices.

Ask the Right Questions

  1. What is my brand voice? Who is the right person (or voice) for your audience? Is it male or female? Do they have an accent? Consider what voice your audience might expect to hear. An unexplored area in voice design is using familiar expert voices of real people answering questions within their area of expertise.
  2. Which questions can the device actually answer and how do we guide users to those questions?
  3. Who should use the voice service? One of the considerations we are looking at for clients is building first for customer service associates. Associates can be given guidelines for which questions a voice service can answer and can also help to train it. Using a natural language processing service requires training over time, which can be accomplished with customer service representatives before it is launched to customers.
  4. Given the desire to create two-way dialogue, how do you design for both the input and output of the device? How does the device know when to listen and when to speak? Are there certain spoken commands that will signal this shift, or can it be conveyed through things like timing and tone?

Determine the Right Level of Control

Despite our enthusiasm for voice services such as Alexa, Siri and Cortana, people still want to feel like they are in control. That can sometimes be as simple as limiting the response of a voice service. For instance, if I tell a device to turn on the lights, I expect the lights to turn on, but I don’t require verbal confirmation. Prioritizing action over information simplifies the exchange, while giving the user a greater sense of control.

Shifting control is an important consideration. If the user asks to turn the lights off, a response of “I turned the lights off” shifts control to the device and away from the person, whereas “lights off” or no sound at all leaves the user with control. (The action of the lights going out should be enough of a guide.) The use of “I” for a voice-controlled system seems like a subtle choice, but has a great deal to do with how the service is perceived. In our practice, the services we design specifically avoid using “I” in order to provide the user with control.

Voice Design Best Practices

Through our research, we’ve developed a series of best practices for voice design:

  1. Forget about the tech: When designing for voice, it’s important to test the voice-controlled interactions as two-way conversations first, temporarily putting the technology and function of the device on the back burner. Consider what a positive and successful interaction would be between a person and a service, device or assistant. You may find that a verbal response is not always necessary or preferable, as in the light switch example.
  2. Try it internally: Before bringing your voice-controlled service to consumers, try it out with your employees and develop a staff enhancement version. This will give you an opportunity to see how the service assists your employees in their day-to-day activities and will pinpoint any miscommunication that needs to be addressed. We can also let employees know which kind of questions the service can answer, establishing more realistic expectations. Giving associates this service has the added benefit of allowing them to continue face-to-face interactions with customers while they ask voice service questions through ear pieces or other mobile devices.
  3. Identify a specialty and set limitations: No single device can be good at everything or answer every question. Rather than trying to design an omni-functional device, focus on key areas that will be most relevant for your target audience. Be sure to include a design for responses outside of the device’s realm of knowledge, but don’t place too much emphasis on it.
  4. Design the Hand-Off — If the service is embedded in phone or in-person conversation, letting people know they are speaking to a service and having rules for hand-off to a human is critical. People usually don’t mind speaking to a service as long as they are quickly handed off to a live person for help if the question can’t be answered.

Giving Voice to the Future

Voice-controlled services are an output of today’s focus on human-centered design, which is why it’s important to focus on your audience first and foremost, not the technology. How do you humanize the interaction for your audience, while still ensuring the human retains control? We have explored these topics further in the Fjord 2017 Trends Report. Of course, the future of voice services remains tethered to conversation. By designing for those conversations with the limitations, control considerations and hand-off procedures in mind, your brand could be well on its way toward making people’s lives just a little bit easier.

Originally published on HOW design.

--

--

John Jones
Design Voices

Managing Director, Head of Design, Digital Innovation @ J.P. Morgan CIB — Product and Service Design