Scaling the Coaching Experience With Design

Product design

Type of work
Product design

Languages and Tech
Sketch, Framer, HTML/Javascript

March 2017 - April 2017

Project Link

During my 2017 winter term, I was given the opportunity to work as a product designer with a health startup called Noom based in New York. Their main offering, Noom Coach, is a program that focuses on improving the health outcomes of people diagnosed with chronic illnesses, such as obesity, through a regiment of logging and real-life coaching. The program is offered as a 16-week bootcamp that guides these users to better health through personalized action plans, curated content, and coach feedback from the activity they log through the Noom Coach app.

An unprecedented growth

By the end of 2016 and the beginning of the new year, their business-to-consumer stream saw an unprecedented growth of user sign-ups that strained our coaching resources. Coaches were overloaded with the large volume of new enrollees from this surge while managing the relationships of their existing user load. As one of their responses, Noom sought an ambitious acquisition of new coache hires to balance the users amongst a larger resource pool. Screening hundreds of coach applications proved to be arduous for recruitment, while on-boarding new hires did not match the pace of the user enrolling in our product. Another short-term strategy assigned incoming users to one coach account shared between all coaches as an attempt to balance the workload, but this ultimately elicited negative user feedback once the trick was discovered.

"How do we scale our dashboard to empower coaches to manage 170 users while maintaining quality service?"

As user growth continued to soar, we asked ourselves the important question that many startups address: how do we scale? More importantly, we challenged ourselves to find solutions that can empower coaches to manage 170 users while maintaining the human quality our product excels in. Ambitious and determined, my design team began the investigation in coach scalability through an intensive 5-day design sprint to find sustainable product solutions in a short period of time.

User-oriented Research

Our team began the sprint by interviewing a select group of 5 coaches, both remote and in-office, to map out a general coach journey with our enrolled users from the start to end of their 16-week program. These coaches had diverse experiences ranging from an ex-competitor employee to an in-office coach supervisor. We sought to understand a universal process every coach underwent with their users, and their testimonies revealed underlying pain points with the current dashboard and the program in general.

Notes on defining the design sprint goals

Notes on coach interviews

A couple pages of notes on problem definition and user interviews.

From the information gathered, the coach-user journey could be decomposed into 4 primary phases:

  • Pre-assignment: the user is not assigned yet to a coach is asked to complete questions to collect user insights
  • Creating an action plan: from the pre-assignment insights gathered, the coach is assigned to the user along with creating a strategy to meet their health goals
  • Main loop: this phase comprises most of the coach/user journey and consists of routine activies such as analyzing activity logs, holding chat conversations, and adjusting the user's action plans.
  • Post-core: after a user completes their 16-week program, they can continue to interact with their coach in sustaining their health outcomes

Talking with the coaches confirmed that most of the time spent in this journey comes from the main loop, and our team focused the dialogue to specifically target the coach's experiences from their day-to-day interactions with the user. We noted key insights arising from these conversations, rephrasing as "How might we" questions to see if their concerns pointed to a strategy for scalability:

  • How might we end a conversation with a user?
  • How might we reduce the user onboarding process while collecting critical user insights?
  • How might we optimize routine food logging feedback?

It was particularly observed that messaging consumed much of the coach's working time, and often conversations between their users shared enough redundancy to be optimized. Coaches would copy and paste canned messages to certain inbox alerts that would arise frequently such as a check-in or welcoming a new user. Their process to these events currently requires them to switch between the dashboard and a word processing application to paste these messages, resulting a higher cognitive load for a coach. Of their overall working time, we found that coaches were spending 55% sending reactive messages (responding to a user's conversation) while 18% of was spent on proactive messages (starting a conversation).

From this discovery, we decided to integrate suggested messages feature into the coach dashboard as a tangible strategy to help coaches scale. The goal of this feature was to greatly reduce time spent on responding to common inbox alerts with automatically canned messages without sacrificing the human quality provided through their coaching style.

Prototyping with Chrome Extensions

Part of the structured design sprint challenged us to validate our learnings with a prototype by the end of the week. We considered various approaches on how to test our hypothesis such as making a Marvel prototype for coaches to give feedback on. Given my additional background in front-end development, I was able to design and build a Chrome extension that fully integrated with the existing dashboard in one day. The interface design for the feature was primitive: wide call-to-actions appeared at the top of the dashboard to elicit action from the coach. Clicking the suggested response then copied the message into the user's respective chatbox for the coach to review and edit before sending it off.

This approach worked out successfully and enabling coaches to immediately test out the feature by the next day. Conceiving a working prototype helped us test our product feature against real user communication and elicit more genuine end-to-end coach feedback.

Using the learnings from the Chrome extension prototype as my launching pad, I began my dive into realizing a more thoughtful iteration of the suggested messages feature. I asked myself where the placement of the feature should go along with their respective design consequences. Placing the suggestions as a list inline the chat interface seemed the most intuitive for a coach to locate them, but the chat's current physical constraints would have cluttered the interface. Conversely, the placement below the user's profile from the Chrome extension allotted the design with more space, but it would have felt disconnected from the chat experience. The end result placed the feature inline the chat window while increasing the height of the conversation history to compensate for the added space.

I ended up prototyping the final version as a high-fidelity HTML page here.

UI options for an inline solution

For solutions within the chat window, I considered several UI options for display a list of message suggestions (vertical scroll vs. horizontal scroll vs. paginated). Horizontal scrolling and paginated ultimately didn't feel intuitive on a non-touch desktop interface, so I proceeded with a vertical list for future iterations.

UI options for outside the chat window

I also considered placing the suggested messages right below the user profile. This option allowed for the feature to be prominently visible and encourages coaches to interact with it more easily. This option was discontinued later in the process since it felt disconnected from the chat experience.

Final mockup of suggested messages feature

Final mockup of the suggested messages feature. The top left label indicates the context triggering the message options suggested. Clicking the ... icon toggles the display of the panel.

Form for adding custom messages

Final form for adding custom messages. Previously, coaches used a Google Doc with canned responses they already prepared, but allowing for custom messages under this feature would eliminate that extraneous step to their workflow.
Learning Outcomes

Through the design sprint, I was able to help conceptualize and deliver a product feature that coaches at Noom are using today. The user interviews conducted at the beginning of the process proved to be a crucial step in helping inform the decisions I took into the end design of the suggested messages. Researching how coaches interacted with their users uncovered a shared, universal process that they were facilitating, and through this understanding, we were able to focus on the major time-sinks and ideate solutions to minimize them. This project also helped me realize the effect of my personal bias, and applying feedback from my mentors helped guide me to an objective, intentional design.