Discovering VUI: Part 1 – User/Customer Behaviour
It’s a long time since UX was just about websites. Since I began my career in 2012, I’m increasingly as likely to be working on apps and software as a ‘standard’ consumer site. This is a good thing. Learning, and then learning some more, is one reason why I got into this gig.
In early 2017 I was geekily excited to get, not just my first Voice User Interface project (specifically for Amazon Echo), but the first start-to-finish VUI brief here at Rufus, and for a particularly illustrious client – who shall remain nameless but let’s say if you’re talking about voices, this is the one.
So no big ask, then for a novice like me. Luckily, as an industry, we’re all still novices when it comes to voice interfaces, which means what I learned is likely to be useful for you if you’re starting your own voice project.
Here are some of the things I learned while putting it together.
What I expected before I started
I knew wireframes weren’t going to play such a large role (or any role at all). I also knew that rather than working closely with the graphic designers (like I do with most projects), this was going to be predominantly working with copywriters and developers.
But there were some things I didn’t expect – aspects of voice interfaces that are quite different to GUI (graphic user interfaces) in interesting ways.
Surprise 1: VUI loves dead ends.
Going in, I assumed that, just like in good ole content-heavy websites, the principle of ‘no dead-ends’ would apply to voice devices too.
Nope. Users of VUI devices don’t want long drawn-out conversations with Alexa or whichever assistant they’re talking to. They want the content they are looking for. And if they want more, they’ll ask.
In fact I found that pushing lots of related content will frustrate your user. Now this could change in the future as users become more familiar with the devices and how to control them. But for a starter, I’d recommend using dead-ends.
Surprise 2: Lists will be fine.
Nope. It’s normal to display a list of available options on web design. Like the latest news articles, cheese recipes, Friends episodes. However, when trying to replicate this on VUI, it becomes a whole new challenge.
Unlike in GUI, the user is heavily reliant on their short-term memory when it comes to lists in VUI. The human brain can (on average) remember up to 4 things at once so I had to make sure to chunk up the results in a way that makes it easier for the user. And when dealing with long titles (like "The One Where Chandler Can't Remember Which Sister”) I found that we needed to reduce the amount in the list to 3 or 2. (Although it is highly recommended to reduce long titles if you can).
Content-based companies such as newspapers and magazines need to be very careful how they surface lists of content. It’s critical not to overwhelm the user.
And something I underestimated: Users are impatient.
True. EVERYONE is time poor (at least like to make out they are). But even though this is an exciting new technology, people still expect it to be brilliant from the off.
A good way to win the customer over is to show your humble side. Explain from the off that this is new and you’d love their feedback. That way you’re in it together. This is the 21st century. We don’t own the experience, the customer does.
This is also true when the assistant voice is speaking. Don’t go on long monologues. It will only annoy people. Keep it short and to the point. The user isn’t speaking to the device to have a conversation. They have a goal they wish to achieve (tomorrow’s weather, latest podcast episode, something to cook tonight). Most of the experience is in creating the easiest, most direct route to that piece of content - the user's goal.
So arguably the key thing I learnt when it comes to working with VUI is the same as the golden rule when creating any website or app: the user won’t blame you if they don’t like the content, but they will if they can’t find it.
Next week: We will talk about (fill in the gaps) “The pr_ct_cali_i_s of working with VUI".