Designing Personality: Bot Persona

Industry Problems
Chatbot makers indicate a need for help building and developing the personalities for their conversational systems. For example, X.Ai’s has a chatbot named Amy, a meeting scheduler that get’s cc’ed to emails, finding the perfect time to meet through emailing the participants. In an X.Ai job posting, they seek someone to help “design and test methods to humanize Amy’s responses so that interacting with her feels as natural as interacting with a human assistant”. To achieve these tasks, X.Ai is specifically looking for people with a degree in theater or phycology (X.Ai, 2016). In addition to more tangible dialogue that must be created, designers are being increasingly asked to perform the literal design of a personality. This requires developing an emotional and cognitive framework of how a service may think, feel, and behave (American Psychological Association, 2015). Defining a personality maintains consistency across all of the interactions. The initial task in designing a conversation, as indicated previously is forming a strong hypothesis of the conversation’s goals, purpose and structure is an important step towards an effective conversation. These elements determine the why behind their conversational design. I propose by reorganizing a service’s goals and purpose in the context of empathy, teams may be able to more effectively design personalities.
Challenge for the Method
Developing personality is an odd task for most people, since it’s such a human concept, it is hard to know where to start. Additionally, creating a personality that is useful is an even more challenging task. This is measured in the degree a system allows “users to complete a set of tasks and fulfill specific goals in a particular context of use” (MacDonald and Atwood, 2014). A useful personality is one that addresses (1) the functions of the system, (2) the tasks users are trying to complete, (3) the goals users are trying to achieve, and (4) and the context in which the system is being used. What is missing in designing personality is a method that helps teams collectively address the components of utility and render a complete personality that the team can design with. One attractive method is building upon the success of the Persona.
Background on the Persona Method
Cooper designed the Persona as a concise artifact that illustrates what teams know about their users (Knox et al., 2014). The goal was to help the entire team, whether they participated in research or not, quickly empathize and understand their users. Each persona is designed by finding patterns amongst users and grouping them into their goals. By doing so, it is much easier to identify if proposed solutions meet or do not meet the goals of the identified target audience. In practice, however, many organizations create weak personas because they forget how vital effective research is into the Persona construction. As many designers know, in-depth research is often flouted, leading to the many critiques of Personas (Wong, 2015). What results is that the team becomes aware of the artifice behind Personas. Because of these experiences, the Persona Method has evolved. There are four main characteristics of the Cooper persona: Biography, Think, Feel, and Do.
Components of the Persona Method
The biography, at minimum, is a name. This is the first key input to a persona, giving a one-word shortcut to convey all of the knowledge that a completed persona conveys. The name itself has been a source of contention when building personas as teams occasionally choose unrealistic names in the spirit of conveying the knowledge behind the persona, such as the alliterated name, “Sarah the Security-Minded” (Noessel, 2014). The problem with these names is that they undercut the reality of the persona. Reducing the credibility of the name disassociates the result with the knowledge that the team has gained. The same is true for any additional ‘factual’ biographical information that is added, such as age, education, experience level, role, and so on.
Thinking and Feeling are two elements that are tightly connected, though, in the Cooper Persona, are separated. In Business Model Generation, however, they are combined, though separated from a user’s “Pains” and “Gains”. It’s important to note that the actual separating is not the most important part. Rather, these elements are selected to simply describe a user’s “environment, behavior, concerns, and aspirations” (Osterwalder et al., 2009). In Business Model Generation, to answer the qualities of Thinking and Feeling, designers are asked to understand what is important to a user, what moves her, and what may be sources of anxiety. For Gains and Pains, designers are similarly asked to understand the user’s biggest frustrations, obstacles, needs, wants, and how she measures her success. In Cooper’s persona, Pains and Gains are baked into the distinct Thinking and Feeling categories to shift focus to understanding the why of a “user’s actions, choices and decisions” (Knox et al., 2014). The categories are not as important as capturing the “needs and goals that the team designs for and they are easier to talk about, remember and get a shared view’” (Blomquist and Arvola, 2002). This intention explains Cooper’s intention to simplify the Persona process into the resulting three categories of Think, Feel, and Do. By simplifying the qualities, it may be easier for designers to fill in the respective buckets with their understanding of the user.
The Do category is the simplest of the categories as it is the most concrete. There is a mild distinction to be made though, as Do is focused on what the user wants to achieve, rather than what they want to Do with the chatbot. Point being, the persona is crafted with a more open ended concept of what captures the behaviors of a user whether the chatbot exists or not, yet in a certain context. Again in the Cooper Persona, designers are asked to pick a particular situation to “exaggerate” the pain-points and challenges that the user may face (Knox et al., 2014). I find this characterization a bit misleading. Instead, I find picking a context important to understand how one’s more minor behaviors shift based on circumstance. A user’s more weighty needs and wants may stay fairly consistent in a relationship,  as contexts shift, their behaviors may change more readily. Thus, in a persona, I find it helpful to clarify the context that the target user is operating in, helping shape the kinds of behaviors one may take.
The Bot Persona, Modifying the Persona Method
Personas increasingly are understood within an iteration-validation process (Ries, 2011; Osterwalder et al., 2009). In Lean UX, Jeff Gothelf uses “Proto-Personas” which serves to indicate all the assumptions of the users that the team must validate. Cooper, using the same design, uses the Persona as a collaboration tool for entire businesses to converge on their shared (or conflicting) understanding of their audience. The benefit of this use of personas is that they are intentionally presented to always be in a state of change. These personas are filled with post-it notes and/or hand written. By staying in a low fidelity, hand crafted state, they communicate that they are incomplete, meant to be defined and redefined in the face of experimentation and user-testing. Yet, in their uncompleted state, personas still fill the familiar role of effectively eliciting empathy in teams for their users, providing a simple means to measure the appropriateness of design decisions.
​
Bot Persona by the author, Austin Beer
I begin by setting up the need of teams to empathize not only with their users but also with each other and the chatbot they are creating. By sharing their collective, comprehensive thinking about what they are designing, they can better understand what each other are intending to design, imparting the importance that the Bot Persona is a collaborative and iterative process.
Next, I describe the Bot Persona Worksheet that I have created for them. The persona I have developed builds on Cooper Persona to make the persona a bit more clear in its function. The center of the persona is an illustration of a generic robot with space within to write the chatbot’s name and stated goal. Above the robot is space to write the user’s goals. In an introductory test of the conversational persona, participants noted that it was difficult to understand if a goal was the chatbot’s goal, the business’s goal, or the user’s goal. Because of this, I separated the chatot’s goal from the User’s goal. This allows teams to understand how the two goals can relate to each other. A goal is also an important component taken from Improv Teachings as it provides an overall guide to dialogue structure. To the right of the chatbot are two blank sections, labeled Think and Feel respectively. To the right of the chatbot, one large section is labeled Do. I deliberately sized the print-outs in order to constrain the number of post-it notes in an effort to encourage teams to fill the space with a couple thoughts, but also to only have a maximum of three ideas per section. Because this may be many participant’s first experience designing personalities, it will be easier for teams to focus on only a few key qualities and behaviors.
After describing what a Persona is and its components, I ask experts to illustrate the process by filling in their own before the audience. In the case of the workshops I am holding, improv actors will be my experts to encourage greater divergence. After illustrating the Persona Process, workshop participants will be asked to begin filling out their own persona sheets for roughly 10 minutes. Time boxing the activity helps focus teams to the task at hand, but also to make some mistakes and assumptions.
Plan for Leading the Bot Persona Method
I begin by setting up the need of teams to empathize not only with their users but also with each other and the chatbot they are creating. By sharing their collective, comprehensive thinking about what they are designing, they can better understand what each other are intending to design, imparting the importance that the Bot Persona is a collaborative and iterative process.
Next, I describe the Bot Persona Worksheet that I have created for them. The persona I have developed builds on Cooper Persona to make the persona a bit more clear in its function. The center of the persona is an illustration of a generic robot with space within to write the chatbot’s name and stated goal. Above the robot is space to write the user’s goals. In an introductory test of the conversational persona, participants noted that it was difficult to understand if a goal was the chatbot’s goal, the business’s goal, or the user’s goal. Because of this, I separated the chatot’s goal from the User’s goal. This allows teams to understand how the two goals can relate to each other. A goal is also an important component taken from Improv Teachings as it provides an overall guide to dialogue structure. To the right of the chatbot are two blank sections, labeled Think and Feel respectively. To the right of the chatbot, one large section is labeled Do. I deliberately sized the print-outs in order to constrain the number of post-it notes in an effort to encourage teams to fill the space with a couple thoughts, but also to only have a maximum of three ideas per section. Because this may be many participant’s first experience designing personalities, it will be easier for teams to focus on only a few key qualities and behaviors.
After describing what a Persona is and its components, I ask experts to illustrate the process by filling in their own before the audience. In the case of the workshops I am holding, improv actors will be my experts to encourage greater divergence. After illustrating the Persona Process, workshop participants will be asked to begin filling out their own persona sheets for roughly 10 minutes. Time boxing the activity helps focus teams to the task at hand, but also to make some mistakes and assumptions.
Assumptions and Testing Criteria
The success of this method will be measured by its utility in helping teams understand the different components a personality requires, the degree in which it it helps teams collaborate to develop a personality and if teams are encouraged to later revise their work after testing their ‘Proto-Persona’. I do have initial concerns with this method. One of the common failures of personas is that, when filled with arbitrary information, they can be viewed less credibly. The problem with my method is that these personas in a workshop setting are complete fabrications. My assumption is that teams will rely on characters and people they have previously encountered to ‘fill in the blanks’. Unfortunately, this “political act of making a chatbot who can inform about a cultural group has the often unintended consequence of creating the context for stereotyping” (Marino, 2014). I intend to include some time for team reflection to help expose these issues, though this ethical issues deserves greater attention and discussion. Another concern with a less credible persona is that teams will ignore the work they have done when they are more engrossed in developing conversations. Later in the workshop, I plan on encouraging teams to revisit their persona, after building initial conversations. I am less worried about this possibility as this method is simply a tool to help teams share their thinking and reflect on their group knowledge.
​
Bibliography
American Psychological Association (2015) Personality. Available at: http://www.apa.org/topics/personality/ (Accessed: 1 February 2016).
Blomquist, Å. and Arvola, M. (2002) ‘Personas in action’, Proceedings of the second Nordic conference on Human-computer interaction – NordiCHI ’02, . doi: 10.1145/572020.572044.
Busche, L. (2015) Brand Voice and UX: A Starter Guide. Available at: http://blog.invisionapp.com/brand-voice-and-ux-a-starter-guide/ (Accessed: 2 February 2016).
Gothelf, J. (2013) Lean UX: Applying lean principles to improve user experience. Edited by Josh Seiden. United States: O’Reilly Media, Inc, USA.
 Knox, N., Brazen, T., Noessel, C., Thompson, S., Cooper, A., Cooper, S., Winchester, C. and
Samadzadeh, S. (2014) Persona empathy mapping. Available at: http://www.cooper.com/journal/2014/5/persona-empathy-mapping (Accessed: 2 February 2016).
MacDonald, C.M. and Atwood, M.E. (2014) ‘What does it mean for a system to be useful?’, Proceedings of the 2014 conference on Designing interactive systems – DIS ’14, . doi: 10.1145/2598510.2598600.
Marino, M.C. (2014) ‘The racial formation of Chatbots’, CLCWeb: Comparative Literature and Culture, 16(5). doi: 10.7771/1481-4374.2560.
Noessel, C. (2014) Barry the Blog post.. Available at: http://www.cooper.com/journal/2014/4/barry-the-blog-post (Accessed: 2 February 2016).
Osterwalder, A., Pigneur, Y., Smith, A., Clark, T., van der Pijl, P., Alex, O. and er, P.Y. (2009) Business model generation: A handbook for visionaries, game Changers, and challengers (delux.. 2nd edn. Amsterdam: Alexander Osterwalder & Yves Pigneur.
Ries, E. (2011) The lean startup: How today’s entrepreneurs use continuous innovation to create radically successful businesses. New York: Crown Publishing Group, Division of Random House.
Spool, J. (1996) Branding and usability. Available at: https://www.uie.com/articles/branding_usability/ (Accessed: 2 February 2016).
Tassi, R. (2008) Stakeholder Design Activities. Available at: http://www.servicedesigntools.org/taxonomy/term/17 (Accessed: 2 February 2016).
Wong, J. (2015) The problems of using Personas — Masters of Experience. Available at: https://medium.com/masters-of-experience/the-problems-of-using-personas-341da959b826#.gsbbosx8x (Accessed: 14 February 2016).
X.Ai (2016) Careers Page. Available at: https://x.ai/jobs/ (Accessed: 1 February 2016).