Experimenting by Talking: Improv

Industry Problems

After defining the goals and purpose of a conversation, conversation design begins when we start sketching possible strategies that can structure the conversation. What is the beginning, middle and end? What are the key inputs we need and how can we get them from our users?

Elaine Lee is a designer at Large, a chatbot that helps offices order anything that was needed through a chatbot. One of the services is ordering coffee. Lee was challenged to design a conversation that could figure out the perfect coffee for a particular office in one or two questions. Her insight was that what kind of coffee machine an office uses closely reflects how much the office cares. Also, many of the machine types require different grinds or beans. For example, an office with a percolator coffee maker uses different coffee than an office with fully manual Italian import espresso machine. After understanding what kind of coffee maker the office has, Large can already provide a couple options reflecting some more nuance. This conversation’s strategy is short, and focused on the brewing method.

Goal for the Method

The challenge for conversation design is that it’s easy to quickly get caught into one, common sense conversation rather than really exploring the divergent options. A discussion about the weather can easily be “How’s the weather?”. But there are many others, “Do I need an umbrella today?” “How should I dress?” “Is there anything that might delay my plane?” “What’s the surf report?”. Thus it’s important that a method for divergence encourages teams to experiment with different perspectives and rational. Additionally, teams need a method that quickly tests and validates each possibility in as realistic of a situation as possible. Lastly, real life conversations can help teams better empathize with their users.

Inspiration for the Method

This problem of exploring conversations iteratively, is very similar to the process of Improv Acting. In Improv, the process is a spontaneous outpouring of information, with clear rules and structure, namely “Yes, and…” (Fowlie, 2016). Actors always accept another person’s contribution of information, then add more information. When an actor adds some information, this starts building context. As each actor continues to add information, character, context and goals become solidified. Improv’s collaborative spontaneity helps people to discover engaging, innovative, and complex situations. Simsarian describes the Role-Playing method, nearly identical to Improv, as participants take on a character role in some context (2003). When participants are “in the moment” and use their entire bodies to explore ideas, there are a wide range of benefits, including: “maintaining group focus on the activities at hand; bringing teams onto the ‘same page’ through a shared vivid experience that involves participant’s muscle memory; deferring judgment while building on other’s ideas; building deeper understanding grounded in context; the ability to viscerally explore possibilities that may not be readily available in the world” (Simsarian, 2003).

Components of Improv

Common human-centered design practice often involves developing user personas, providing a summary of the audience to test designs decisions against. This Improv method can help explore how we can imagine not only different users, but also different manifestations of our service. We are able to explore conversation strategy, different personalities, pacing, and question types. In a less constrained format, I would expect to improvise the multiple kinds of users a chatbot may interact, to test whether certain conversation structures or personalities better work for some sets of users than others. Yet, I anticipate that this process, while wonderfully entertaining, can also become very complex to deal with all of the combinations of user personas and service personas. Because time is limited in the anticipated workshops, I am planning on asking participants to only explore different dimensions for the service-side of the conversation. By doing this, we may be able to place greater emphasis on careful design of how our services are perceived and accepted by an average user. Another important component to improv is how active listening to how a scene is developing helps actors find inspiration for how to contribute. Active listening is when people are devoting their complete attention to comprehending and processing what each other is saying (Bryant, 2009). By seeing one set of actors attempt an interaction, the next actor can try something completely different, defined by the contributions that came before. The group as a whole then are exploring the possibilities by being present in the conversation. The cycle of participation and observation allows the group of actors to quickly explore a wide variety of possibilities that the situation can hold.

Modifications for the Improv Conversation Method

I am modifying the Improv method by reinforcing the reflection component so that observations and insights can be more easily saved. While performing the “Improv Prototype”, I introduce an additional role where at least one participant is focused on recording observations of what people are doing. These observations are merely to understand what interesting choices the other participants make while engaged in the scene. Individually, one set of observations does not indicate the best choices. As the group rotates roles, the combined observations from the entire session allows for many data points from which the group can later discover patterns.

Plan for Leading the Improv Conversation Method

In the workshops, I introduce the Improv Method by picking a random context and role-play the conversation with a workshop participant. After our characters have reached some agreement, I ask the audience for feedback on what helped the conversation, interesting words or phrases that came up, and comments on our personalities. While receiving feedback, I note what they say then repeat back the audience feedback, teaching the audience how to make effective observations. Teams are then instructed to stand and begin role-playing their own conversations. For example, in a team of four people, two are acting out a conversation, one is recording observations, and the other is engaged and waiting. Every couple of minutes, or when a conversation closes, team members rotate roles. After several rotations, all of the observations are gathered and teams reflect on the observations, identifying patterns and opportunities to shape the conversation’s goals, purpose, and structure.

Assumptions and Testing

The improv method intends to help teams experiment with all aspects of a conversation. This method will be successful when teams build greater empathy for the diversity of conversations that can be had around a particular topic and a growing understanding of their conversation’s goal and purpose. In reflecting on their Improv conversations, teams can better revisit their hypothesis of how to address a problem. A major assumption I am making is that teams will be able to engage in Improv with a divergent and open mindset. Jimmy Fowlie, a lead Improv Teacher at the Groundlings, cautions that the first problems new students to Improv encounter are fear and hesitation (Fowlie, 2016). His suggestion is to elevating the energy of the participants by acting animatedly and to demonstrate to audiences what is expected, by having trained improv actors demonstrate the process.


Bryant, L. (2009) ‘The Art of Active Listening’, Practice Nurse, 37(6), pp. 49–52.

Fowlie, J. (2016) ‘Interview with Jimmy Fowlie’. Interview with 24 January.

Lee, E. (2016) Product design and creative direction by Elaine Lee. Available at: http://www.elaine-lee.com/#large (Accessed: 24 February 2016).

Simsarian, K.T. (2003) ‘Take it to the next stage’, CHI ’03 extended abstracts on Human factors in computing systems – CHI ’03, . doi: 10.1145/765891.766123.

Last updated