Google A120: Ghost Fitness
PRODUCT DESIGN
PROTOTYPING
How might we improve health outcomes for everyone with novel approaches to resistance training?
Company | Google
Role | Senior Product Designer
Links | Google's Area120
Overview
DesignMap had an ongoing relationship with Google's startup incubator Area120 until January 2023. We would work with Google founders and teams to take their startups from zero to one, and were brought in in partnership with Google's UX Research team to ensure the triple diamond process was followed closely, and the products pitched to Google partners were user-centered. I worked with Ghost Fitness to build a strength training mobile app and AI-powered coach, with the goal of getting absorbed into YouTube or Fitbit.
The Problem
Only 22% of adults meet the American Heart Association’s recommendations for muscle-strengthening activities. The challenge? Many just don't know where to begin, what routines to follow, or how to progress when diving into strength training.
In order to get up to speed, I conducted various discovery exercises to ramp up on strength training, existing research, and the team.
____
Empathy & System Mapping
____
User Insights from FitBit
We were able to pull a lot of research from FitBit regarding motivation and behavior change, and synthesize that information into what might be useful to Ghost Fitness. There are usually some "struggle loops" when people are on their various health journeys. In order to help people combat these loops and push past plateaus, we needed to consider the types of motivation that help people progress and get unstuck.
____
In-Person and Remote Research
We were able to conduct research on DScout for remote interviews with people interested in strength training and fitness apps, but the most enjoyable sessions were the ones in person with trainers and coaches. I synthesized the user interview findings into insights.
____
Landscape Audit
I mapped out the flows of >10 popular fitness apps people use for strength training, as well as some apps that used chatbots or AI coaches. Then I did a deep dive into selected key competitors for insights. From this activity, the team understood the existing solutions in the fitness and coaching space better, and we were able to identify competitors, their pros and cons, and possible opportunity areas for Ghost Fitness.
The Solution
Discovery never ended, we were continuously learning. But through that process we decided to build an AI-powered fitness coach that would provide the knowledge needed to begin a strength training program, guide users through their workout plan, track their activity, and motivate them to progress past plateaus.
Why AI? An AI coach can process vast amounts of data rapidly, detecting patterns, inconsistencies, or plateaus that might be harder for humans to spot. This constant and real-time analysis allows for immediate intervention and feedback. AI coaches are available around the clock, and can handle a vast number of users at once, ensuring consistent quality of coaching regardless of the number of users. Coaching becomes more affordable, accessible, and scalable.
While AI coaches offer unique benefits, it's essential to understand that they don't replace the empathy, experience, and nuanced understanding that human coaches bring to the table. Instead, they can be viewed as complementary.
____
Scope
There were many tracks of work happening simultaneously, including the development of:
High-fidelity prototypes of target state flows
Vision work geared towards YouTube fitness influencers
____
Initial Flowchart
Based on the themes uncovered by the landscape audit, I guided the team through an initial flowchart of the Ghost Fitness experience. Iterating on this chart with the entire team, including engineering, was instrumental in creating the most intuitive flow for the product.
____
Mockups & Prototypes
The first prototype included a chat that mimics an AI-powered fitness coach, workout plan information, and progression tracking. Since Ghost Fitness had meetings with high profile YouTube fitness influencers, we also built prototypes specifically for presentations with their teams. When creating a creator-agnostic version of the UI, we transitioned to Material Design in order for the app to be developed more quickly by the engineering team.
In addition to the mobile prototype for creators, we also built some mockups for what those programs would look like within the YouTube Creator Studio backend on desktop, as well as within the YouTube browser for subscribers.
Results
____
Testing
After conducting interviews and testing our prototypes with creators and their teams, we achieved 100% buy-in! Influencers with millions of subscribers were excited to partner with Ghost Fitness in the following ways:
As a tool to communicate more seamlessly with fans.
As a method to broaden their reach.
As a dedicated home within YouTube for fitness content.
"I love YouTube but I built my own app because I want to be able to help my fans find the right solution for them, and see engagement to help them progress. A tool like this would be a dream to have."
— Anna, YouTube Fitness Creator with 2.2M subscribers
____
Takeaways
Despite rave reviews, the project ended abruptly in January 2023 amidst Google's shift in focus away from Area120. Despite the ending, the work was fast & scrappy, incredibly fun, and I learned a ton about strength training. Some key takeaways:
Rapid Iteration: Working in a fast and scrappy environment allowed for rapid prototyping. This meant quicker feedback loops and an ability to pivot designs based on real-world usage.
Collaboration: Given the pace, there was an even greater emphasis on cross-team collaboration. Efficient communication between designers, developers, and product managers was crucial to avoid misalignment.
Scalability: Working with a tech giant like Google meant that the design had to be scalable from the start. This experience taught the importance of thinking globally and ensuring that the app can cater to a diverse set of users.
Balancing Automation & Human Touch: How can we strike the right balance? While AI offers unprecedented efficiency, not all automatable aspects necessarily should be. How do we ensure AI complements the invaluable insights, nuance, and expertise that come from human experience? Since our project's end, I've seen a wave of new apps pop up with LLMs and AI coaching. Watching their approach to this balance will be fascinating!