Wynde Down Well
Leveraging the Apple Watch to build sustainable sleeping habits
Project in Human-Computer Interaction for Christina Wodtke’s CS 247B: Design for Behavior Change lecture
Winter 2021, 10-week project
My Role: User Research, UX Design, and Team Organization
Project Summary
Our inter-department team of engineers and designers started from our collective desire to get higher quality sleep, despite the strain of college life. Our final product, Wynde, would help users do so with targeted and timely prompts — both additive (let’s meditate for two minutes) and subtractive (you shouldn’t have anything caffeinated).
This initial iteration leverages user perception of the Apple Watch as a health tool to encourage small shifts in behavior, helping users improve sleep quality by targeting the key problems college students face when attempting to establish good sleep hygiene.
Our Challenge
Better sleep. We want it, and so does just about everyone else — especially in our social context as university students. But sleep isn’t just about getting in bed, and it isn’t about getting eight or nine hours; it’s about how your habits throughout the day stack up, and whether they help you wind down or keep you buzzing. Talking to other students, the role of stress was a key factor in sleepless nights, but so were their actions earlier in the day — getting in a workout before bed, hanging out with friends (pre-COVID), or staying up working on assignments.
One of the major challenges for us is balancing the variety of college life. Many users don’t get great sleep because they prioritize other things in the moment — whether it’s social, academic, or interest, it’s easy to be influenced to stay up just a bit longer. Sometimes those will lead to great outcomes — less stress over an assignment, feeling secure with friends, or contentment with a side project—but consistently sacrificing sleep has been shown to lead to worse overall wellbeing, and worse performance across those fields.
Our Solution
Discovery: User Research and Literature Review
Going from the rough idea of “it would be nice to sleep better”, we started talking to other college students — in user interviews, in diary studies, in intervention tests. We found that many had tried to integrate a sleep app into their schedule, but found it challenging to remember. Many also had very little space to wind down between work and sleep, going straight from their screens to bed — and then staying up doomscrolling. They also didn’t have set nighttime routines, which gave their nights less structure and more opportunities to vary and otherwise affect sleep quality.
We conducted a few interventions — having users set alerts on their phone, or mentally linking recommended wind-down activities with standard anchors near bedtime — like brushing teeth — following BJ Fogg’s discussion of anchoring new habits to old ones in Tiny Habits. In the course of these studies, we had some success in anchoring, but found that alarms were annoying, guilt-inducing, and untimely for users, and generally unhelpful with behavior change as a result. As a result, we decided to focus on fitting into routines and leaning away from rigid prompts.
While both studies were short — five days — we got some helpful direction from both and learned a lot from our conversations with users and reflection on our own habits.
At the same time as we were conducting user research, we were attempting to learn at the problem space more broadly. We completed a brief competitive analysis — sleep is a big space — and found a gap in helping users develop long-term habits and avoid screen time. We also read about behavioral factors affecting sleep, and saw some gaps in the ecosystem that we could address as a result of that research. A key perspective coming out of the research — and part of what makes it so challenging — is that sleep quality metrics and the habits to improve those metrics are unique from person, but intention setting and prioritization leads to better sleep (and therefore a slew of other great outcomes).
Ideation
Results in hand, we moved onto ideation, thinking about how to frame our solutions to help enforce habits without forcing anything on users — how it could feel natural. Khalid mentioned leveraging heart rate tracking to respond to user context, and we liked the idea, moving forward with it after examining a few other higher-fidelity half-sheet sketches and storyboards.
Narrowing down: Form follows Function
Once we refined how we wanted the app to function a bit more, we started thinking about how we wanted it to look; Christine and I combined the mood boards everyone had done into a style tile to drive our UI development. In the process, we focused on providing a high-contrast profile that avoided the negative effects of blue light.
Next, we got to work on our actual user interface, first mocking up the setup flow, and then expanding it to cover a notification flow responding to overactivity, the flow to help users “wynde” down in the half hour or so before bed, and check stats; I focused on the “wynde” down/bedtime flow.
Finally, we conducted a heuristic evaluation to get a better idea of good next steps in developing Wynde. Key problems were:
1. Long onboarding — one of the main things we heard was that users were dismayed by their initial interaction being this long survey. One potential fix could be to choose the key questions, then let the user fill in the rest in settings, or to provide the option at the start to choose between a quick setup or a longer, more personalized one.
2. Language clarity — much of our internally used language — “overactive” and “wynd down” to talk about types of notifications, “keep it minimal” or “send me everything” for notification frequency — is not clear without context. Numerical scales were also unlabeled and unclear. Potential solutions include using more clear terms and, when possible, quantifying the language rather than using vague terms, as well as labeling or using symbols for unclear scales.
3. User control and options during setup — There isn’t much choice between options during setup, which led to testers expressing frustration at points. A full implementation would likely include more options — for example, with regards to activities to help users “wynd” down.
4. User freedom — users were not able to choose “other” options — for example, when we asked about exercise frequency and duration. It would be better, following up on (1), to have things that are non-essential in a “further setup” option in settings or the user screen that pops up when you enter the app from the main apps page.
5. Responding to prompts — it was sometimes unclear how a user should respond to notifications — what actions and in what manner. In the future, it would be good to ensure that the wording of prompts is clear and drives action.
6. Screen space — navigation buttons take up a lot of real estate, which makes some elements hard to read. A potential fix would be to have users shown that they can use swipe motions to navigate the setup and redesign the buttons to occupy less space around the text.
7. Specificity — many users don’t have a bedtime routine, so it could be helpful to let them add one in future iterations.
Our Next Steps
We were happy with the rough form factor, although there are many clear improvements to be made — both from the heuristic evaluation and feedback from classmates. That’s the point of a first draft though— we got something down and learned from it. In addition to refining our UI, it would be best to go in and conduct works-like tests with some of our key assumptions with the form factor of the Apple Watch — something we would have done once we narrowed down to it, if not for the limited time frame of this class. We would also look at monetization strategies in the sleep and Watch app spaces, developing our go-to-market strategy based on that. Another important step would be to cultivate a close mentorship with a TA (hey Sean + Matt!), professor, or someone else in our network with perspective on our challenges and approach.
