Making Chatbot Interactions Come Alive
If legendary actor, singer, comedian, and style icon Mae West did user experience (UX), she’d probably now be famous for a chatbot design principle, “It’s not the bots in your life that matters, it’s the life in your bots”.
Mae West in 1936 (Image in public domain via Wikipedia)
Actually, both parts of Mae’s would-be statement ring true:
- The job to be done by your chatbot is vital because it provides the reason for botifying a task in someone’s life, sure.
- But now the style, tone, and attitude of the chatbot — the personality — encountered along the journey to getting that job done really makes or breaks a great overall user experience. And a satisfying UX that resonates personally is a powerful strategy for creating more customers.
Personality: An Unbroken Series of Successful Gestures — F. Scott Fitzgerald
Yes folks, personality is the new user experience. Designing that personality for the new hotness of conversational user experience is a science as well as an art.
“It’s no coincidence that both Howdy and X.ai — a startup whose bot intelligently schedules meetings via email — each hired writers with performing-arts backgrounds to help define the UX of their products.” — The Next Phase Of UX: Designing Chatbot Personalities
Thankfully, we are witnessing the death of the hostile concept of “the user” and of “using” a system and instead see the emergence of a human design narrative of chatting or conversing with technology.
Personality, therefore, is critical to the success of that natural dialog between human and computer. It’s beyond the transaction.
Fate doesn’t hang on a wrong or right choice
Fortune depends on the tone of your voice
For UX pros what we are talking about here is anthropomorphism (not to be confused with personification). Broadly, you can think of anthropomorphism as the attribution of human motivations, beliefs, and feelings to inanimate things like computers, vehicles, robots, and so on.
Herbie the Love Bug. One of the most famous anthropomorphic vehicles. The concept of anthropomorphic vehicles has been around for a while; all have different personalities. (Image shared via Wikipedia)
Importantly, anthropomorphism does not mean the application of all of the subtle nuances and quirks of human personality into our human-computer interaction, but only the dominant, memorable, important personality traits that will be useful to the interaction in that particular context.
Ah, that UX concept again: context.
So, it is vital for the UX designer of chatbots to make use of important and contextually relevant anthropomorphic principles to create solutions that will resonate with people for the right thing, at the right time, and in the right place.
When we interact with computer interfaces we project human emotions and beliefs into the inanimate computer. When the interaction is smooth and enjoyable we attribute the pleasure to the machine in the same way we blame the machine when things don’t work out as we wish. — Don Norman, Emotional Design)
To an extent, chatbot anthropomorphism and engaging in a digital conversation is determined by the AI capability of your chatbot platform. However, designers (or artists or performers if you like) now have a central role in crafting chatbot attitude, style, and tone that reflects personality, whether it be by writing scripts prompts and messages or recording the voice of the chatbot.
Making That Personality Personal
I’ve already outlined how language crafting and conversational scripting skills are at the heart of UX chatbot design, but I’d like to call out some key chatbot personality design considerations.
- Determine the attitude, tone and style of you chatbot — the personality. This means understanding what the personality of your bot given the context of use. Writing about designing for personality is another article in its own right, but check out this Medium piece and this webinar about how you might get started in creating a tone and voice that customers will resonate with your customers.
- Introduce your chatbot conversation by saying “hello”, “hi”, etc., and respond to input greetings, asking how people are doing today, and so on.
- Acknowledge that a chatbot is a bot and not a human and say what can it do to help.
- Once the level-setting on “why you’re here” done, ensure your chatbot is different by using those additional quirks of personality to build a great user experience. True, there is some debate about whether even naming a chatbot matters, but again, that’s another article! Regardless, Billy No-name never trumps Billy No-mates!)
- Avoid repetitive circuitous replies. At some point (maybe the second attempt), admit defeat and seek clarification. Imagine if you had to talk ad infinitum to a real person who “just didn’t get it” and the emotions that would arise!
- Learn from past conversational decisions and prompt recommendations accordingly. That’s kinda like your favourite barista remembering you like a double shot espresso when you walk into the coffee house. Context is king in making a great customer experience.
- Indicate that the chatbot is being thoughtful about an asked question rather than spewing out a rapid answer. So, those “I’m thinking” indicators are important. This is an emerging technology after all…
- Use humour (perhaps “wit” would be a better word) wisely, in context, and in ways that make sense. Although Irish people like to say “only joking” when things get messy, I can tell you invariably they’re not. And a chatbots making a joke about a transaction can be heart-stopping sometimes….
Hipmunk on Facebook Messenger “just kidding”. Your mileage may vary on that kind of joking!
- Handle the common chit-chat of chatbot interaction (“tell me a joke”, “what’s the weather”), respond to the ‘what can you do now” stuff, and engage on time critical events and topics. This seemingly “throw-away” part of chatbot UX is an important part of personality . For example, knowing “Did Dublin beat Mayo again at GAA football?” (Of course we did!), when it is International Women’s Day, or when Sinterklaas is arriving, provides an “in the know” moment with real people. This kind of banter is natural, but also a UX vehicle to engage and create a seamless path to guide the conversation back to the core use.
GrowthBot on Slack: Cartoons and jokes provided.
- Be as tolerant of chatbot input errors such as typos and grammar goofs as you would be of how people speak and write IRL (e.g., pointing out that someone’s split an infinitive or the difference between “lesser” and “fewer” as you chat is not a conversation enabler in any context!).
- Be able to process slang, abbreviations, urban lingo, emoji, and so on depending on culture and to respond that way accordingly is a black art but increasingly we communicate that way. Naturally, your chatbot platform’s capability is critical here too.
Donut.ai bot on Slack: Talking in emoji. Could emoji be considered anthropomorphic in their own right?
- Showing basic manners and things we take for granted IRL— such as saying “thanks”, acknowledging something’s done (“ah, sorted”, “deadly”), or looking for clarification by admitting chatbot limitations of understanding (“help me out here!”) is also an “only human-like after all” trait.
- Be polite and understanding. For example, Amazon’s Alexa can diplomatically tolerate a lot of inane and even rude conversation, and probably more than the average parent can in dealing with their own kids or even watching them interact with Alexa! However, a polite way of dealing with profanity or user frustration is encouraged.
VHI’s Vee on Facebook Messenger handling the bad language aspect of running frustration!
- Avoid prolonging a conversation and trying to continually sell more to a customer. Solve the job to be done problem and close the conversation and ask if your listener is done. But, if a chatbot customer wants to discover and do more, then great!
- Avoid open-ended questions; give options and choices instead. “Do you prefer Hip Hop Beyoncé or Glamour Beyoncé?” is a faster way to shape a solution than to ask, “What do you like about Beyoncé?”
PayPal Bae fashion chatbot on Facebook Messenger: Those Beyoncé options
- Steer and guide the conversation. Remember the purpose of the chatbot is to solve a problem, so keep bringing the conversation back to that jobs to be done (JTBD) decision. And when will you know when the conversation is over? When the job is done (or the problem is solved)!
VHI Vee chatbot on Facebook Messenger adds value to get that running conversation across the finish line. Until the next time!
Talk the Talk
In conclusion, chatbot design as a natural way of interacting with technology means we are now in the age of a human-centered design approach in the digital space, and not one of “user”-centered design.
Who wants to be called a “user” in any context?
Just imagine. What if Amazon “Alexa” were an “Alex” or all digital assistants were of one gender only and each had a different personality, style, and tone? Think about how this might change your experience and willingness to engage depending on the context.
Amazon Echo Alexa tells me my fitness goals are toast. Nice personality touches make for a great UX even after working out! (Audio needed)
Or would it?
You’d need to research and test out the impact of such differences, but very often a gut feeling based on guerrilla UX research and simple observations about how we live and work with real people with real personalities in real situations is a good starting point to understanding how chatbot personality can shape the digital experience.
The design of personality for chatbots is a real skill and talent; a hot commodity. This is a fast-moving space, with lots of ideas that span many domains and disciplines, so your views and comments on the topic are welcome!
Ultan O’Broin (@ultan) is a senior user experience professional with Oracle and also blogs about UX and technology. His opinions here are personal and not necessarily those of his employer. He is a member of the editorial board of MultiLingual.
All screen images are by Ultan O’Broin.