The ultimate guide to chatbot personality
A chatbot’s personality is like Donald Trump’s Tax return — everyone’s talking about it, but few have ever seen it. Indeed, your last encounter with a chatbot or virtual assistant may have reminded you of the classic meme:
A wide range of personality guides have been written as chatbot developers continue to grapple with optimizing user experience through conversational interfaces. However, despite the vast range of information available, this conundrum is clearly as confusing as ever. As I wrote in 2018, a poorly defined personality is one of the top 5 reasons a chatbot will fail, and at the end of 2019, not much has changed in that regard.
In this article, I condense my learnings from the past few years of working with conversational AI with clients in New Zealand and Europe. I establish the importance of both a chatbot persona and personality, how to define them, and highlight opportunities to let them shine. My hope is to provide a one-stop guide to developing your ultimate chatbot personality.
But before we get started…
What’s the Big Deal with Personality?
As chatbots become ever more prolific, their purpose has shifted. They have evolved quickly from a trendy display of “being digital” to an authentic extension of your brand through meaningful conversations with customers. Whether your chatbot answers queries, provides purchase recommendations, triages and resolves customer issues, or just engages in general chitchat, your chatbot must be acknowledged as a critical brand ambassador of your organization. As such, the same level of scrutiny must be applied to chatbot interactions as is given to more traditional brand experiences. Similar to an amazing interaction with a human customer service representative, exceptional chatbot experiences contribute to positive brand reinforcement and goodwill. The conversations your chatbot has with a user will, for better or worse, form part of the woven tapestry that is your brand and reputation.
Further, research shows that personality is a key driver of chatbot engagement and repeat use — both critical metrics to the success of a bot. Tuva Lunde Smestad found that chatbots with a well-defined personality scored higher “in all factors, pragmatic quality, hedonic quality and attractiveness” than a control chatbot. This intrinsically makes sense. With new human acquaintances, we are drawn in by compelling personalities and spend more time with those we find engaging. Likewise with bots, the duration of conversations and number of return visits is, all else equal, higher with chatbots that demonstrate a compelling personality than those who come across as robotic (excuse the pun), dull, or one dimensional.
Persona vs Personality, does it matter?
When building an artificial version of an intrinsic human experience, taking a human-centric approach is critical. There are a number of factors to consider when designing the character with which users will interact. These include:
● What persona will it assume?
● What personality will the bot demonstrate?
● What avatar or visual identity should it carry?
There are a ton of great blogs by experienced visual designers on the third bullet point (such as this and this), so I will limit the scope of this article to points 1 and 2.
With real human interactions, we are engaged (or disengaged) not only by the person’s profile — their interests, hobbies, job, and passions, etc., but also by the way they act and react in different situations. Likewise, with chatbots, these factors must be designed with the end user in mind to maximize engagement. Unfortunately, many online sources on this topic confuse, or disregard, the difference between personality and persona. As such, let’s pause to define these concepts, and understand why this delineation matters.
A persona is the set of attributes that define one’s societal role, whereas a personality can be thought of as the unique combination of intrinsic emotional and behavioral drivers that are visible in the way we act in different contexts. For example, a persona can include things like one’s job, hobbies, interests, purpose and passions, while personality encompasses how one interacts with the environment around them.
To help bring this difference to life: at my firm, new joiners to our team are asked to prepare a one slide overview about themselves, and present this at the first team meeting they attend. Photos of their dog, maps of their hometown, and logos from their university or favorite sports teams are common. It is the information on the slide that forms (part of) their persona; it’s the way in which they present it that is a reflection of their personality.
Unfortunately, a lot of online content encourages readers to design chatbot personalities by considering what the bot’s favorite movie would be, or what sports team they would support, for example.
I would argue, however, that these are in fact attributes of a bot’s persona, and provide very little indication how a chatbot would converse with the user. A well-defined personality provides a framework for the nature of a bot’s response to user input, for example how agreeable, extroverted or neurotic the bot should be. But in responding to the question “Where would you like to go on holiday?” (we’ve seen much, much stranger, believe me), what should the bot actually say? This is why both persona and personality are important to design, when striving for a complete and engaging user experience.
Now that we’ve clarified the difference between persona and personality, and why it all matters, let’s dive into each.
Step 1: Define its Persona
Primarily, the purpose of creating and defining the persona of your chatbot is to have a strong understanding of its character. It might seem illogical assigning a persona to what boils down to a bunch of code, but ultimately, this helps provide a complete user experience. It doesn’t matter if your chatbot acts as Level 1 IT helpdesk support, makes sales on your eCommerce site, or answers employee questions about payroll, your chatbot will invariably fall off the ‘happy path’ at some point. Like human staff, your virtual staff will get distracted and deviate from the process they were built for. Why? Because the users will undoubtedly ask them other questions. “Where do you live?”, “What is your star sign?” and “Tell me a joke!” are all incredibly common user questions, regardless of the purpose or function of the bot. Having a well-defined bot persona will enable your bot copy writers to create responses to a wide range of possible questions with purpose, rather than falling back on error loops.
When defining your bot’s persona, first reflect on the following:
● Your users — what are your customer segments and which ones will interact with a chatbot? What are their interests, hobbies, passions and other attributes? While everyone is unique, it is important to align the bot persona to resonate with your main user groups. Just see how successful you’ll be trying to sell jeans to teenagers through a bot that sounds like their mother. My guess is: not very.
● Your brand — what does your brand represent? If your brand was a person, what would they be like? This will likely overlap somewhat with the point above, but is an important extra check in the process. You wouldn’t want your bot to admit to illegal substance consumption, even if that’s how most of your users spend their Saturday nights, for an exaggerated example.
● Your bot’s purpose — what is your bot’s raison d’être? What role has it been created to fulfill? As a balanced check to the above two points, consider what type of human role your bot is simulating, and ensure this reflected. The persona of a bot answering neighborhood noise complaints should have a different persona than one trying to up-sell rock concert tickets, for example.
You should strive to articulate your bot’s persona succinctly, in 1–2 slides. This clear understanding of who the chatbot is will act as reference material, enabling the team building the bot to create a consistent experience across all conversation flows and topics. Further, it will contribute to creating meaningful, engaging conversations with customers.
Below is a simplified example of a chatbot persona:
A few tips and tricks for developing your chatbot’s persona:
● Always start with the persona — Define what your bot is about first, before designing its visual identity or agreeing its personality.
● Often genderless — There are certain contexts in which a gendered bot is more appropriate or effective, however I’ve found that in most situations, an undefined gender is more effective. Make sure your bot’s name and visual identity reflect this.
● Bring the brand to life — Have you ever been asked what color you would be, if you were a color? This process is almost the exact opposite to that, in the sense that you need to create human characteristics from your brand.
Step 2: Define its Personality
Much of what I’ve read online regarding chatbot “personality” leads to the development of what I consider to be a persona. So how do we develop a personality for a bot?
There are many frameworks that have been developed to try and quantify or define one’s personality. My team and I like to think about chatbot personality through the lens of the Five Factor Model (FFM), or the Big Five (read a psychologist’s perspective here). The FFM outlines five factors that drive our behavioral and emotional responses. People can be scored as high, low, or something in between on all five factors. This combination of five scores defines our fundamental personality.
In the FFM overview below, each factor description describes behaviors of those that score high in that factor.
Scoring low in a particular factor does not mean the described behavior is muted, but rather results in the opposite behavior being true. E.g. those scoring low on ‘Openness’ gain fulfillment through perseverance and are characterized as pragmatic and data-driven.
When designing your bot, it is important to understand how you want your chatbot to score in each of the five factors outlined in the FFM, as this will determine how your bot would react in different situations. For example, when reflecting on the persona in the previous section, our bot Charlie could react and respond in many different ways to the same question, depending on how he scores on the FFM.
While exaggerated, the examples above demonstrate how different factors result in different types of responses all else being equal. These in turn will affect different user experiences when extrapolated across an entire flow or conversation. It is likely that your chatbot will not be so extreme, and will demonstrate a combination of personality factors with high and low scores for each.
Step 3: Let it Shine!
In the sections above, I have outlined how to define the persona and personality of your chatbot, and demonstrated how this influences the ‘what’ and ‘how’ of your chatbot’s responses. So in addition to general tone of voice, where can we demonstrate this personality or persona in an obvious and engaging way?
● The Welcome Message — While it is important to outline here what your bot can help with, the welcome message is a great opportunity to let the personality shine. Does it welcome users with a ‘Good evening’ or a ‘What’s up’? Does it reference current events, comment on the user, laugh at itself, or get straight to the point?
● The Error Message — With the current state of chatbot technology, it is inevitable that your chatbot will not recognize a user’s question or intent at some point. Instead of responding with the generic and infuriating “can you rephrase that?”, this is a good opportunity to delight the user with an explosion or personality, while still achieving the desired rephrasing of input.
● Unrelated Questions — Similar to the point above, users are guaranteed to ask questions outside the functional scope of your chatbot. However, there may be some questions you decide to respond to with an interesting or humorous answer rather than just an error loop, such as “where do you like to go on holiday?”. Particularly if it aligns with your brand, this is a great opportunity to engage users and plug your brand while you’re at it.
● The Last Message — In conversations with a natural end point (where the bot can bid adieu before the user closes the chat window), make sure your bot closes the user’s meaningful brand interaction on a high. Make it memorable, make it resonate, and make it distinctly <insert your company name here>.
In Summary
As deploying chatbots to engage with customers increasingly becomes standard practice, chatbot designers must give consideration to how the user experience can be maximized. A chatbot’s persona and personality that resonate with users are two key factors in ensuring high engagement and returns, but also extending your organization’s brand through meaningful conversation.
The distinction between persona and personality in chatbot articles is often blurred, however it is important to define both. Both answer the question ‘who is my chatbot’, however, the persona is the ‘who:what’ while the personality is the ‘who:how’. Defining the persona will provide engaging substance to your bot’s answers, while its personality will determine how it responds to different situations. Of course, there is no inherently ‘good’ or ‘bad’ persona or personality for a chatbot. Rather, the ideal combination should reflect your brand, resonate with your users, and align with the purpose of the bot.
If you enjoyed this article on chatbot personality, you’ll find my article on chatbot desirability really useful:
