Why you need personality as a design framework

The importance of persona has always been a hotly debated topic. It’s time to settle the score once and for all.

When I started out as a conversation designer, about 4 years ago, most clients would say they wanted a chatbot without a persona. Personality was associated with cracking jokes or giving answers to silly questions, like ‘what’s your favorite color?’ or ‘who made you?’. It was perceived as frivolous — and serious companies with serious reputations are not frivolous.

So, every time we started a new project, we spent considerable time educating clients on why persona is essential. Over time I grew better and better in explaining it, whilst simultaneously gaining a deep understanding of how integral to conversational AI it really is.

Rooted in science

Let me begin by saying I didn’t come to any of the following conclusions on my own. A number of people shaped my views over the years and I'd like to share with you what I learnt from them.

Kicking things off with Wally Brill, I had the pleasure of attending a Google Voice Workshop he hosted back in 2018. Wally Brill is a veteran in the conversational AI space and one of the core people responsible for creating the persona of the Google Assistant. Hearing him speak completely changed the way I viewed things. Not only did he make the case for why persona is so important in a supremely eloquent manner, it was also rooted in science.

You see, humans view the world through a uniquely human lens. It’s the reason we see faces in the sky and why we like Disney characters so much. Research shows that when we hear a voice, any voice, we automatically assign a personality to it. We make assumptions about age, gender, and locale. We decide if we like the person and whether or not they are trustworthy. It happens in a matter of seconds, even if that voice is artificial. The same goes for chatbots.

No such thing as ‘no personality’

Credits for this line go to Cynthia Marcucci. She’s a colleague working with us in the US and is an absolute expert when it comes to persona. Her adagium is that there is no such thing as ‘no personality’. Meaning: when language is the interface, persona is implied.

When you design with a clear persona in mind, generally, your assistant will be perceived as more likeable. However, without this ground anchor in place, it will not end up being seen as neutral — no, it will most likely be viewed as confusing, unreliable, off putting, and all-over-the-place. As people we pick up on these subtleties, we just do.

Personality is multimodal

Language is only one part of the equation, though. Jason Gilbert, a multimodal designer at Intuition Robotics, recently explained to me the careful consideration they go through when designing experiences for ElliQ.

ElliQ is an in-house companion robot for elderly people. Because she has a body, she can move. She can lean in when curious, put her head down when sad, and dance when she’s in the mood. She is also equipped with personal sound effects – your classic beeps and boops. They are what oinks are for pigs and meowing for cats. 

She also communicates through LED lights. It helps with turn taking, signalling she’s listening or thinking before answering. The light pulsates soft and slowly when she guides someone through a mindfulness exercise, and when she’s happy she lights up and lets herself shine bright. 

In chatbots, we only have to worry about one modality: text. Although we can convey a great deal of personality from text alone, adding more modalities provides us with unique opportunities to build digital creatures that we can bond with, really bond with, like we do with our pets.

Design systems for conversational AI

In summary, words are never just words. Words convey personality. And personality is amplified when more modalities are introduced. 

What are the implications for us as designers? My dear colleague Jasper Klimbie puts it as follows: we should view personality as a design framework. A set of rules, descriptions, and practices that help us create more cohesive, consistent experiences. This is of the utmost importance if you work on a single product as a team, regardless of whether it’s a chatbot or a multimodal experience.

What that looks like and how to manage that, is a whole different topic. A path that leads us to designing ‘design systems’ for conversational AI. But let’s save that for another time.