Purposeful History
Transforming user activity into well-being insights
Purposeful is a health and well-being product centered around daily check-ins and reflections. I designed a major overhaul of the History and Progress pages of Purposeful to better deliver well-being insights back to users. As a result, usage of these pages increased substantially, helping to improve well-being outcomes for our users. Along the way, ChatGPT joined the ride..
Client
Kumanu
Roles
Lead UX Designer
Timeline
August - December 2023
Above: Progress and History on mobile
Context
Spearheading UX for a health and well-being product
Purposeful is a health and well-being app designed to help users feel their best by checking in on how they’re doing everyday and helping them develop supporting habits. Purposeful’s buyers (clients) are mainly large organizations (companies, health systems, universities) while its primary audience are the members of these organizations (employees, members, students).
Throughout this project, I was the sole UX designer at Kumanu. I worked closely with a visual designer, a product manager, analysts, and developers.
A lack of Progress
One of the main features of Purposeful is the daily check-in, where users reflect on how they’re doing with their health and track well-being metrics such as mood, activity, sleep, and more. The Progress page existed primarily to display ML-powered insights to users based on this data, but these insights were generic, unclear, and showed up at random. In addition, users could also access a historical view of their past data, but this was a rudimentary question-and-answer printout view that could only be navigated on a day-by-day basis.
Because improvements to the Progress page were user-focused and lacked buyer or business need, the page languished. It saw no improvements in two years, and was the lowest-visited and lowest-used of the four main pages in Purposeful.
Process
Setting goals and building a business need
I looked at a wide variety of products with some progression or data visualization component including Fitbit, Apple Health, Duolingo, Calm, Insight Timer, mood trackers, and even video games. Throughout the summer, I invested some time working with like-minded teammates from the development team. Together, we conducted a lot of brainstorming, sketching, and prototyping. I bounced some of our ideas off users through quick tests and surveys.
Above: Goal-setting and ideation.
Above: Some of my initial sketches.
A crucial breakthrough was the idea to integrate some form of AI — ChatGPT or a custom large language model (LLM) — into the redesigned Progress page. Since the health and well-being data users submitted on a daily basis tended to be text-heavy, an LLM such as ChatGPT would work well in summarizing that data. The technology had huge promise for Purposeful as a whole, and the Progress page could serve as an excellent test bed since the inputted data was predictable and users wouldn’t be able to directly interface with the AI model.
Hence, we were able to build a strong business need for Progress page improvements, and I received the green light to begin formal design work.
Exploring different visualization approaches
Next, I began wireframing and rapid prototyping several ideas. I focused my efforts on the Progress page — where high-level insights and information would be shown — as well as a brand new History page where detailed historical data would be displayed.
Above: Prototypes of various approaches to visualizing progress.
I explored various innovative approaches to visualizing progress for the user. Some were inspired by the competitive analysis I had done earlier, while others were inspired by Purposeful’s starry constellation identity. User testing surfaced usability issues with these approaches due to navigation unfamiliarity and lack of clarity. Further, a novel design would have increased the development cost substantially, and I could not get consensus for what the progression system would actually constitute. Thus, I pared back these designs to instead use the Progress page to simply display AI-powered insights and high level snippets.
Above: Prototypes of various approaches to displaying historical data.
For the History page, I anchored my explorations around displaying data in day/week/month formats, following paradigms established by popular fitness apps. Within each time period, I prototyped different data visualization approaches such as calendars, bar charts, line graphs, word clouds, and images. Ultimately, I discarded bar charts and line graphs due to concerns over displaying missing data. The week view was also deprioritized as we felt that it lacked relative value.
I then iterated on the day and month views many times, eliciting feedback through user testing and internal presentations. Throughout this project, I focused testing recruitment on current users of Purposeful as well as users from the general public who were active users of popular health and well-being apps. At certain points, I also validated my designs with niche populations that use Purposeful, such as seniors and healthcare providers.
At the same time, our talented developers began implementing and testing AI models on summarizing history data. Early results were promising, and I included snippets and examples in testing. Users did not realize that an LLM model was being used to provide insights; in this specific case, that was a good thing!
Designing for scale
Thanks to user feedback, my product manager, and our developers, I was pushed to design Progress and History pages that truly summarized and displayed data in a valuable, digestible, and informative way for the user. It also innately encouraged users to use the app more often to receive more value. Furthermore, the pages were structured in a modular way, able to account for both missing data as well as future increases in scale and integrations.
Outcomes
Successful testing grounds for AI
The user response was extremely positive towards the end of testing and just after release. From submitted feedback and unmoderated interviews, users shared that the Progress felt a lot more valuable, and that this encouraged them to use Purposeful more. Analytics and telemetry backed that up, as we saw that visits to the Progress page increased. However, I parted ways with Kumanu soon after release, so I’m unable to confirm if this trend will continue in the long run.
In terms of setting the stage for AI integration into Purposeful, this project was also fruitful. Our developers managed to implement a custom LLM model that worked well in providing accurate, short, useful snippets or summaries of data, and this complemented our past ML models. By early 2024, we had not received any reports of issues with the insights provided on Progress, and Kumanu had begun exploring integrating AI models into other parts of Purposeful such as onboarding and check-ins.
Making a case for the user
Personally, the biggest learning I took away was the value of building allyship with other stakeholders to push towards consensus. We saw the user needs, and by being innovative and resourceful, we were able to build a valuable business case as well. Apart from that, this was an excellent exercise in data visualization for me, and I learned that the biggest bang-for-your-buck isn’t in building overly-fancy or complicated dashboards, but in giving users the tools to connect their own dots with their data.