Behavioral Study for Chatbot Design
The goal of this study was to learn about how, when, and why people contact businesses for help, in order to inspire the future designs of automated services and artificial intelligence (e.g. chatbots). I spoke with five people between the ages of 21-38 for 75 minutes each and heard many stories about their behavior. This project was independently run by me (not as part of a team) and I had just about 5 weeks to do it from beginning to end. This client is a Boston-based tech company that specializes in automated technologies. The project was done as part of a graduate class (Field Methods in Human Computer Interaction).
STUDY DESIGN
Research Goals
- Understand what motivates people to use (or not use automated technologies.
- Understand the contexts or scenarios in which people prefer to engage with automated technologies (e.g., a chatbot) as well as the contexts or scenarios in which people prefer to engage with a human.
Context & Dynamics
Out-of-context one-on-one interviews allowed me to deeply explore individuals' past behaviors and instances regarding AI. I was able to tailor the conversation to an individual and question deeply about behavioral and affective aspects without the influence of a specific product or others' opinions.
Interviews & Activities
I wrote a flexible interview guide that brought interviewees from broad questions about asking for help to specific questions about how and why they do it with a business. I asked interviewees to rank methods of asking for help and asked them to tell stories about their preferences. I also included a collaging activity that used 100 randomly generated images to jog additional feelings and emotions related to asking a business for help. The guide was modified slightly during each interview depending on how the responses were going in order to get at the pertinent information I needed. The interviews resulted in 5 session worksheets with the interviewees' answers. My individual notes were taken on the interview guides and on post-it notes.
ANALYSIS
I took my notes from the session and categorized each person's data (based on color) into four categories: "behaviors," "motivations," "tensions & barriers," and "attitudes / affective." After noting these categories (used to build a user persona) I grouped data from all different participants into themes that had started emerging. Some of these clusters were around Trust, The Unknown, Phone, Email, and companies being "too Sales-y." These main groups built out the beginnings of my thematic findings. Going back to the recordings was essential in making sure that I understood the sentiments in their proper contexts.
Key Findings:
- People generally prefer to solve problems on their own.
- There are very specific times where talking to a human is of the essence.
- The type and urgency of the task at hand is the primary driver of the method of communication people will use to contact a company. Convenience is the secondary driver of the method of communication used.
- Past experiences with a company or a brand’s reputation for service factor into their propensity to use a certain method of communication.
- People who work 9-5 in offices contact businesses while they are at work.