“How may I assist you today?”
“Hey, what’s up?”
“Hi! How can I help?”
All three questions open the floor to your visitors. They invite visitors to tell your conversation bot what they’re looking for. But each of the three conveys a different tone — and sets your visitors up for a conversation with a different personality.
“Your end user is a human,” Varsha Solanki writes at Humble Bits. “They look for an emotional bond everywhere.” By indicating personality through tone, word choice, and similar factors, organizations can set the mood of the conversation and create a more realistic bond between visitor and bot.
Why Your Conversation Bot Needs a Personality
At first glance, personalities for conversation bots might seem superfluous. Most users communicate with chatbots in order to complete a specific task or goal; very few do so simply to make conversation or for amusement.
Yet even users who need to complete a specific task report a better experience when the conversation bot shows some personality. “Recent reviews find chatbots to be perceived as unintelligent and non-conversational,” Tuva Lunde Smestad of the Norwegian University of Science and Technology writes. Although the projections for conversation bot use are promising, the lack of consideration for personality “make[s] one wonder whether we now are forcing users to adopt technology which they find frustrating and useless.”
Ironically, some companies have responded to users’ distaste for chatbots without personalities by replacing those bots with humans. Their thinking is that it’s easier to get a human to emulate a chatbot than it is to get a chatbot to emulate a human. “Using a human to do the job lets you skip over a load of technical and business development challenges,” says Gregory Koberger, CEO of ReadMe. “It’s essentially prototyping the AI with human beings.”
Humans don’t scale the way conversation bots do, however, which is why a bot with a personality is a must for scalable, 1:1 customer experiences. “Little by little we are experiencing the death of that hostile concept of ‘the user’ and of ‘using’ a ‘system,’” digital transformation professional Ultan O’Broin writes. “Instead, we’re seeing the emergence of a human design narrative focused on the dialog between people and machines by having a natural conversation.”
John Pavlus at Fast Company calls personality “the new UX.” It’s what stands between the natural-language-fluent, adaptive conversation bot and its open-arm welcome by the general public.
Personality Tips for Conversation Bots and the Humans Who Love Them
Like the parents of a promising child, companies with conversation bots are starting to get creative when it comes to giving their fledgling personality everything it needs to thrive.
For instance, X.ai, an AI-enabled bot that handles meeting schedules via email, is backed by a team of AI interaction designers — not coders, but professionals with degrees in psychology, theater, and mythology. The interaction designers “have helped us develop an AI assistant who is frequently mistaken for a human,” says Dennis R. Mortensen, CEO and founder of X.ai.
Another way to make your bot feel more human is to consider giving it a voice and a name. “Voice is fundamentally more humanizing than text,” says Elizabeth McGuane, content strategy lead at Intercom. We tend to attribute spoken words to a human actor more readily than text.
Yet “what is humanizing can also be irritating,” McGuane says. After a while, users can get tired of talking to their conversation bot. Here, a name without voice interaction can retain some of the human feel without fatiguing a customer who would rather type than speak.
To direct a conversation bot’s personality, you must give it a purpose, Gulraiz Khalid writes at the Unified Inbox blog. A purpose helps your conversation bot’s personality cohere around a specific task, focus on the most likely emotional state and responses of its human users, and avoid common mistakes that can make conversation feel stilted or fake.
HubSpot’s Justin Lee agrees. “Before we even began building our chatbot, we focused on exactly how it would create value in a way no other tool could for our audience,” he says. That focus quickly shifted to tasks that could most easily be carried out by having a conversation.
And not just any conversation, but a conversation made of brief exchanges. “With Facebook Messenger, or any chat apps, the general philosophy is to write copy as if you were texting a friend,” says Tam Pham at Bot Academy.
Since most conversation bots’ interfaces resemble a messaging app, the bot’s responses should look like a friend’s responses when you’re chatting with them. This helps align user expectations. Longer pieces of information can live in a blog post or article, to which the conversation bot can provide a link once it becomes clear the user needs that information.
Finally, don’t hesitate to think about how personality shapes brand awareness. Otherwise, your customers will do it for you. “If you don’t spend the time crafting that character and motivation carefully, you run the risk of people projecting motivations, personality traits, and other qualities onto your App and brand that you may not want associated with them,” says Oren Jacob, founder and CEO of PullString.
How to Find Your Bot’s Ideal Personality for Customers
Though it hails from the 1980s, Lewis Goldberg’s Five Factor Model of personality provides a useful framework for narrowing down conversation bot personality options, says Mike Jongbloet at Deeson.
These are Goldberg’s five factors:
Openness. Is your bot open to new ideas or focused on a single task?
Neuroticism. Does your bot display a sense of urgency, or is it laid-back and resilient?
Conscientiousness. Will your bot demand details in a certain order, or be willing to follow the user in whatever order they give information?
Agreeable. Is your bot cooperative, polite, assertive, antagonistic, blunt, or rude?
Extroverted. Does your bot chatter sociably or speak with more reserve?
Matching a Bot’s Personality Traits to Your Brand
No one bot personality will serve every company or goal. For instance, professional organizations like law firms may want a bot that scores higher in openness and conscientiousness and lower in extraversion, while a company that arranges adventure tours may prefer their bot to score high for extraversion and openness but low in neuroticism.
Researchers Kibeom Lee and Michael C. Ashton add a sixth factor: Honesty/Humility, which addresses traits like sincerity, fairness, modesty, and greed. In short, your conversation bot might brag a bit or might downplay its own participation in the process.
Humility is a key trait when it comes to admitting a conversation bot’s limitations to users — a feature that can actually make users much more amenable to talking to the conversation bot in the first place, Jason Brenier writes at Georgian Partners. “There’s nothing worse than a bot that pretends to be smarter than it is, particularly when a simple bot can be perfectly effective.”
Balancing these factors is essential to creating a personality that customers find welcoming, says Oscar Ibars of BotVeu. Ibars uses McDonald’s drive-through window scripts as an example: The company prides itself on efficiency, yet drive-through staff mustn’t sacrifice a cheerful greeting for the sake of productivity. Likewise, a conversation bot can be high in agreeableness without sacrificing its ability to focus on quick responses to the task at hand.
Teaching the Bot to Express That Personality
An essential related factor is tone. “Tone is a subset of voice, that shades based on the audience and where they are in your flow,” says Hillary Frazier, head of chatbot strategy and conversation design at Black Ops.
Depending how customers react to your conversation bot’s comments, the tone of the bot’s response may need to change. For example, it might need to learn to convey patience when users seem frustrated or more upbeat as a user reaches a decision point like paying for a service.
Use the map of your conversation bot’s five (or six) factors, along with tone considerations, to generate potential scripts, Tuba Tezer at Botanalytics.co recommends. “A great way to do this is to imagine you’re making a pitch directly to the target audience,” Tezer says. “What kind of language and diction would you use to appeal to them?”
Another way to think about how the conversation bot should talk to visitors is to imagine yourself introducing the product or service to a friend. Think about how you would walk them through its use.
For instance, if your conversation bot helps users schedule appointments at a salon, what would you say to a friend who needed help scheduling an appointment with the salon? Sarah M. Smart, content strategist at Adobe, calls this “prototyping the conversation.”
Although much of the conversation may not make it into the bot’s final repertoire, “prototyping the conversation in this way ‘did something magical for the team,’ according to my project manager,” Smart says. “It brought what we were all working on to life and made it feel real.”