A Preferred Ride

Ride-sharing Made Delightful


CMU 05-898
Service Design
Instructor: Jodi Forlizzi


Samarth Bahuguna
Nikola Banovic
Mina Kim
Zhaorong (Jerome) Zhong

This is a class project for Service Design. We explored the problem space of cars as Internet of Things and developed a service system that matches preferences of drivers and passengers for ride-share services like Uber or Lyft.


Ride-sharing + Cars as IoT

Ride-sharing companies such as Uber and Lyft enable customers to order affordable shared rides from virtually anywhere. However, existing ride-sharing services fail to offer highly personalized experiences that best matches customers’ expectations and preferences. Not paying attention to customers’ preferences can make customers reluctant to order the service and lead to loss of revenue for the company.

On the other hand, the development of cars as Internet of Things (IoT) gives us opportunities to collect and utilize a variety of data directly from cars. We want to explore how these opportunities can make ride-sharing better.


Mapping Current State

We inquired into the current state through literature review, questionnaires, interviewing drivers and customers of ride-sharing services. Our key research findings include things that make rides enjoyable/unpleasant.

Based on our research, we mapped out stakeholders to understand the eco-system. We also created a customer journey map to identify pain-points and opportunities.

Stakeholder map
Passenger journey map


Speed Dating with Storyboards

After several rounds of brainstorming, we came up with three service ideas and made storyboards of them. We then tested the ideas using “speed dating”: showing users the sketches and quickly gain preferences and feedbacks.

Storyboards of initial concepts

From speed dating we received positive feedback about setting physical preferences and having a “quiet ride” setting. We received questions on whether it would seem rude on the driver’s side. We also need to figure out how sensors can play a more significant role in this concept.


A Preference Matching System

Based on our findings of speed dating, we went through another round of ideation and ended up with the concept of a data-based preference matching system. Using embedded sensors, the system collects data about driving style (eg. aggressive driver or not) and in-car environment (eg. AC, seat, music) and match passengers with drivers that have similar preferences. It also correlate passenger's ratings with the data to form an accurate model of user's preferences over time.

Storyboard of the new concept


Testing through User Enactment

To test our concept, we mocked up the system with prototyping tools and did user enactment with four users. We conducted think-aloud tests with our paper prototype of preference setting app, and then invited users to a ride in a car. We interviewed users after their rides.

We mocked up the app and the in-car display
Video sketch of user enactment process

Some problems we found from the user enactment include:


Find Your Perfect Ride

We promote a new service system where users set their preferences by filling out a questionnaire (with a $2 coupon as a reward of finishing). The system uses embedded sensors to learn driver’s driving style over time and log passenger’s preferences (eg. music, temperature) during ride. The system can match passengers with drivers based on both explicitly filled preferences and information collected over time.

Value flow model of new service
Passenger journey map of new service
Service blueprint of new service

We also created high-fidelity prototype of our mobile app (based on current ride-sharing applications) to demonstrate our design idea.


Discussions and Learnings

Beyond the scope of this class project, this idea still needs further development and user testing. The proposed service system has many shortcomings. For instance, it is very hard for passengers to quantify their preferred temperature, which also depends on the weather and user's dressing. This system is “smart” in the sense that it learns user behavior over time, but it is still not “smart” enough to be context aware. Yet this project as an exploration does show the potential of cars as a sensor platform and the possibilities it provides to service design once we can better utilize the data.

Our approaches of concept modeling, storyboarding and speed dating allows us to focus more on the higher level of the service, its business model, and how stakeholders interact, rather than spending too much time on polishing user interface at an early stage. Rapid prototyping is a very efficient way to test many different ideas before implementation, to fail fast and fail early.

Thanks for reading!