Milaap Case Study

An SMS carpool chatbot that connects folks without transportation in rural 
areas to volunteer drivers.

Team Hannah Johnson, Mike Cardarelli, Saransh Solanki
Duration 1 week

The MHCI+D program at UW kicks off the autumn quarter with Immersion Studio, a week long design sprint completed by teams of three. It was an intense and fun introduction into what will be a packed year for Cohort 6!

The Challenge

Teams were challenged to explore the intersection of civic engagement and technology. Specifically, how can Computer-supported Cooperative Work (CSCW) aid collaboration within communities and lower the barrier to becoming civically engaged?

We developed a design response to these questions through the lens of 
rural communities, a uniting context with personal significance for all 
team members.

Formative Research Methods

Secondary Research

We conducted a literature review of scholarly sources to become more familiar with the following topics:

  • Computer-supported cooperative work (CSCW)
  • Civic engagement issues in rural communities
  • Internet access in rural America

Primary Research

We also conducted three unstructured interviews with civically engaged individuals residing in rural communities. These conversations were illuminating for our team, and helped us gain perspective and empathy with our target audience.

Research Insights

My team and I synthesized our findings into three key insights.

  • Transportation is a main inhibitor of civic engagement in rural areas.
  • Civic engagement often serves as a retirement role for aging Americans.
  • Older people often need support in order to learn new technologies.
  • The majority of Americans who lack broadband internet access reside in rural communities.

Keeping these points in mind, we developed a How Might We statement to help us frame the rest of our design activities.

How might we ensure folks without transportation in rural areas can get involved in their communities?

Ideation

In response to our How Might We statement, our group sketched a collection of 30 design concepts. Through team discussions, critique, and grouping together similar themes among our concepts, we down-selected to four ideas that we felt aligned best with our research insights and time constraints.

If this had been a ten week project instead of a one week project, we may have more strongly considered the Election Day Carpool concept. However, we chose to move forward with the Carpool Chatbot as our final design concept for the following reasons:

  • Older residents wouldn’t have to learn new technology
  • Accessible for communities that lack internet access
  • Best fit for our project time constraint

We decided to name our Carpool Chatbot Milaap, which is an urdu word meaning harmony and ‘good mutual understanding.’

Prototyping

Conversation Tree

To begin prototyping Milaap, our team developed a conversation tree that detailed exactly how riders and drivers could interact with the service. We started at the whiteboard so we could quickly externalize and change our ideas as we discussed. The resulting conversation flow is depicted at a high level in the diagram below.

Purple boxes denote chatbot responses, orange boxes denote user responses

Testing and Iterating our Prototype

To test Milaap’s usability, we used the Wizard of Oz prototyping method. Saransh and Mike facilitated tests with participants who thought they were interacting with a real chatbot. However, I was actually on the other side of the studio replying to their messages manually while adhering to our conversation tree.

For each test, we provided a high level explanation of the service and then handed participants a phone opened to a blank text message addressed
to Milaap.

Testing our concept using this method revealed pitfalls in our conversation flows. With each test, we iterated our conversation tree for improved experience. Below is a video walkthrough of the final Milaap prototype.

Reflection

Future Work

  • Create a driver on-boarding system: Several of our usability test participants raised concerns regarding trust between riders and drivers. How could riders be assured they would be safe with their volunteer driver? Milaap currently relies on the “everybody knows everybody” mentality in some rural areas that we learned about in our primary research. However, in the future we would want to develop an on-boarding system involving background checks for drivers.
  • Explore use of Natural Language Processing (NLP): Our final Milaap prototype uses a system of numerical user input such as “Enter 1 if need a ride, or 2 if you want to drive.” It would be interesting to develop a prototype that hypothetically uses NLP to create a more conversational experience. I’d like to see if our older users would prefer this version, as it would likely feel more natural to learn.
  • Consider how Milaap could scale when areas get broadband: Milaap could effectively serve communities without high speed internet access. However, toward the end of our project we began to wonder what would happen to this service when those areas inevitably do receive better internet. This would be an interesting topic to explore in the future.

Takeaways

  • High tech solutions won’t always be the best fit: While most groups were developing high tech solutions, our team was delving into the low tech world of SMS. Because of this project, I feel better equipped to consider the full range of potential technological solutions that may best serve a specific user base.
  • Embrace ambiguity: During this project I learned that I must become comfortable with being uncomfortable. As a designer, I hope to further trust the human-centered design process when confronted with inevitable ambiguity because it allows for unhindered creative exploration and unexpected findings!