Finding it a challenge to design for voice? Try starting at the end.
By Johnny Chauvet
It’s been just over a year since I worked on my first Voice User Interface (VUI) project. In May of 2016, TribalScale was hired by the PGA Tour to design and build their Alexa Skill for the Amazon Echo. The Skill allowed golf fans to ask ‘Alexa’ questions about current tournaments, players, and standings. Currently, the Echo and other voice platforms are still fairly new to market. As a result, finding documentation and examples on how to design for them is fairly limited. A year ago, it was pretty much non-existent. Amazon did a great job of documenting best practices and providing simple examples to work from at the time, but it was not enough for me to address the complexity of PGA’s product requirements.
On day two or three of the project, after our kick-off and a bit of due-diligence, I remember opening Sketch and getting ready to dive in to creating a flow that outlined the UX. This was my first time designing something that didn’t have a screen or visual interface. I sat there and stared at my blank art-board, trying to figure out how to actually start. My initial instinct was to start with the different ways of how the user would activate the Skill, but it was challenging to envision a path forward from there that would naturally lead to the desired outcomes. I decided to take another approach–what if I started with the outcomes? Just like a math equation, you plot the knowns to solve for the unknowns.
This worked well for a couple reasons. Starting at the end allows you to consider the prerequisites of known outcomes. Figuring out the question that the user should ask is difficult when the answer to that question isn’t right in front of you. When you plot Alexa’s response first, it’s easier to ask yourself ‘what would a user need to say in order to reach this state?”. This is especially helpful when dealing with partial intents, where you need to think about what follow-up questions to ask the user in order to generate the desired response.
The other reason why you should start at the end is because it enables designers to consider the users’ needs first. By starting with the outcomes, you are effectively building empathy for users and crafting an experience that revolves around the responses that satisfies their intentions. This also makes it easier for us designers to perform user-testing for a VUI and iterate on our flows based on real user learnings.
By plotting the responses first, you can survey your target audience to see how they would word or phrase a question to arrive at the known response. Designing for voice requires careful consideration when it comes to speech and wording. There may be dozens of ways to phrase a specific question that could arrive at the same outcome. Applying user-centered design thinking and working with prospective users as you’re crafting the experience is the sure-fire way to ensure you’re using the right language.
For more on best practices when tackling VUIs, check out these links below:
Amazon Alexa — Voice Design Guide
Actions on Google: Conversation Design Tips
Johnny Chauvet is a Design & UX Lead at TribalScale. He is a pioneer in the development of voice user interfaces (VUI) and has been designing voice applications at TribalScale for well over a year. His mandate as a designer is to play a role in making people feel empowered and delighted when incorporating technology into their lives.
This post is also part of a thought leadership series by Wavefront. TribalScale works with Wavefront as a client partner to deliver innovative digital solutions.
Join our fast growing Tribe and connect with us on Twitter, LinkedIn & Facebook! Learn more about us on our website.
