5 Things Every Designer Needs to Know About Conversational UIs
Imagine you want to order a pizza – your usual – but instead of opening an app you send a message. “I want pizza” (it knows your favorite and you’re feeling demanding). Earlier that day you bought tickets to a show and ordered a car to take you there – “Get me two tickets to the X-Men movie and a car to the venue”.
None of this relies on an app’s pixel-perfect UI. Instead, the underlying service is utilized via the same messaging app. The UX then becomes less about the visible and more about conversational interactions in which a Bot, primed to respond, acts on our requests.
‘Design’ is only going to get bigger
AI personal assistant services such a Facebook M and x.ai are already making this a reality while Howdy, a Bot ‘co-worker’ in Slack helps teams organize lunch orders and schedule meetings all with chat, typed or spoken aloud.
Though Google Now has been around since 2011 the recently released Google Assistant’s intention is to create an “ambient experience” where the user isn’t focussed on the interface but the goal.
This requires a slew of new skills for the designer incorporating a knowledge of linguistics and scriptwriting to name just two. It’s not that chat works best in every context – situations where there are lots of variables (like choosing the toppings on four different pizzas) – tend to suit visual interfaces better but there are situations like those above where ii can be quicker to ask rather than open an app and select.
As the remit of the designer is extending well beyond our current idea of what constitutes design – the visual – and so the number of those identifying themselves as designers is only going to go up. It seems that as the industry matures with an increasing shift to specialization, titles such as ‘Web Designer’ become redundant only because they say so little but encompass so much.
Personas aren’t just for users
What then do we need to think about when we ‘design’ for Chat UI? The first thing is to think like a human and not necessarily one mediated by a machine.
Many of the interaction design paradigms and strategies we use stem from a time when computer interaction involved sitting in front of a beige box moving a pointer around a screen with a mouse. The desktop metaphor – storing files in folders – was developed to acclimatize users to using these strange new machines by using concepts they were already familiar with (the office).
A Chat UI becomes focussed on the phrasing and nuance of character that allows the user feel as though they are dealing less with 1s and 0s and more a human, though one that’s attendant to their needs.
In much the same way that Mailchimp uses Voice and Tone as a Style Guide for their copywriters, so too 'chat UI designers' will need to consider how a Bot will sound and what tone is appropriate for each context.
When Mailchimp successfully dispatches a customer’s campaign, the user is happy and relieved so copywriters can “Feel free to be funny”. However, in the event that something goes wrong, the tone needs to be “serious” and show concern. Don’t joke around with people who are frustrated.
What phrasing then would a Chatbot use when speaking with a 20-year-old who is ordering pizza? How would that differ from someone a good deal older who may have a question about their medication?
Our visual designs for these two types of users would differ so, of course, would the tone of the conversation they could expect. Where the former might be breezy and fun the latter should be both effortlessly calm reassuringly accurate. Or maybe something else and this would be borne out in the research.
Constructing personas for Chat UIs would seem as crucial as developing them around indicative users to guide our design decisions.
Consistency of tone is key
Many Bot start-ups employ comedians and scriptwriters to author Bots’ personas. Imagine a future startup paying Louis C.K. to help them construct a 'LouBot' that finds restaurant you'll like? This becomes much more than penning micro-copy since it’s about constructing a consistent character through various scenarios.
It’s not unlike how a TV drama scriptwriter develops a protagonist’s interior and exterior persona and considers how she might deal with the situations in each episode. How would she react and what would she say or do and even more what wouldn’t she say?
Design for humans
Faux pas like Microsoft’s ill-fated AI racist, Tay experiment aren’t limited to chatbots; they extend across any algorithmically mediated interaction. Take, for instance, Eric Meyer’s horrendous experience of Facebook’s Year in Review a couple of years back. Context is king but it’s important to ensure that empathy and common sense should never take a back seat.
Some may get left behind
As the visual UI falls into the background and underlying services become the real currency of UX many are wondering where that leaves a physical product and visual design led company like Apple who for a long time have fallen short on services. To date, their mobile business model is based around “a tightly controlled grid of siloed apps that you poke a thousand times a day on a smooth rectangle of manufacturing excellence” but for how long?
Seemingly 'non-design' disciplines like screenwriting are increasingly being pulled into the orbit of design and it’s definitely an exciting time to be a designer in this space.
Howdy have just recently released a Botkit to create your own Bot in Slack so maybe now is a good time to start designing the conversational experiences of the future.
We already tend to anthropomorphize brands – imbuing them with character traits as if they were people – and brands spend a lot of money ensuring that this character is maintained.
It’s no surprise then that many Bot start-ups are employing comedians and scriptwriters to author Bots’ personas so messaging interactions go beyond micro-copy and feel not just human but consistent across contexts.