Designing a New Feature Through User Testing

What other kind of alarm could guarantee waking you up?

I don't think there's any surer way to wake up than moving your body. Right now, Alarmy has the Step and Shake missions, but what if there was an even more powerful mission? What if you could turn off your alarm not just by moving, but by actually working out?

That thought is exactly what led to Alarmy's Squat mission. You have to do a set number of squats to turn off the alarm. (Right now it's only released on iOS, and Android is coming soon.)

It's easy to think that it'd be great if the alarm only turned off when you're forced to exercise. However, turning that thought into reality isn't easy. Usually, people just think about it and easily give up, saying it'll never work, or they tend to think of other workarounds.

But the Subs (Subscription) squad materialized the idea and carried on with technical research. Because you can't know if it's actually implementable until you try it. Instead of avoiding it, we took on the challenge and ultimately achieved both a launch and results.

How are other fitness services doing it?

I looked into a lot of fitness-related products just in case there was something I could reference. However, even if there were similar products, it wasn't easy to reference them because they all differed in everything from their purpose to their user base. If we proceeded by referencing the researched content just because the feature was similar without considering this point, we could end up with a user experience that doesn't fit our own product.

I went ahead with the design having missed this fact, and belatedly realized it through a few rounds of user testing. So, rather than referencing other products, I redesigned it by placing more emphasis on the characteristics of our product.

The Nth Draft and the Nth User Test

I deleted the entire draft I had first worked on and started over, building it out through the feedback I received after conducting user tests. The following are some of the pieces of feedback I got on that first draft.

• It doesn't feel like a preparation process for dismissing the mission. When it said 'Start,' it was confusing because I didn't recognize what I was starting.

• When the phrase "Stand up!" was shown to the user, it was conveyed as opening their eyes in bed (as Wake up instead of Stand up...). It actually meant to get up from their spot, but that wasn't conveyed.

• The information is scattered across one screen, making the delivery of information unclear.

The biggest problem was that preparing the right posture for a squat and then doing the squat was unfamiliar to the user, and the interface guiding them through it was unfamiliar too. Up until the middle of the drafting process, we were providing a guide upon alarm dismissal, so we only briefly introduced the Squat mission on the alarm settings screen. However, for the reasons above, and because you can't know how to turn off the alarm with this mission unless you encounter the alarm beforehand, we decided to provide more detailed information delivery before they actually try the mission. We also drastically revised the posture guide.

After some thought, we determined that shooting and providing a video would be the most surefire way for information delivery. So we filmed a quick internal demo and ran another UT (User Test).

We introduced the mission with a video instead of photos, and the result was a victory for the video. Providing the improved guide while they had somewhat recognized how to position themselves through the video led users to adopt the posture and start the mission much easier than before. We improved the guide in two major ways based on the existing feedback.

• Improved the phrasing on each screen

• Instead of making them meet the right conditions for the posture, guided them to strike a pose that would naturally produce those conditions (naturally fulfilling the conditions)

Usability improved significantly just by improving the phrasing on each screen. After changing "Stand up" to "Get up from your spot," users began to actually get up and get into the squat position. Also, when we presented conditions for the posture and made them match those conditions, users found it universally difficult to match them, but conversely, when we gave them a guide, they easily adjusted their posture and started squatting.

However, there were still a few things that needed revising in the guide section. One of the main pieces of feedback was that it was hard to recognize that they had to place their fingers on the screen. So we changed the touch area to an intuitive drawing of fingers, and for clearer feedback, we made it change colors when a touch was recognized and vice versa. Besides the mission introduction, we also improved it to provide a video when actually doing the squats so the user could easily follow along.

Then we ran another UT with that draft. But things only went smoothly up to the touching process, and the part where they had to stick their elbows to their torso to make the squat posture (the 5th screen in the draft above) wasn't executed smoothly. After the test ended, we came up with various hypotheses about what the problem was.

Based on the hypotheses, we modified the button's design and even tried putting in a drawing of the squat posture. We even added an arrow above the button telling them to press the button. We revised it through UTs numerous times, but getting them to press the button still remained a conundrum.

During user interviews, we asked why they had trouble pressing the button, and we found out that the difficulty they experienced was because they had to click the button again while keeping two fingers attached. However, since we were already on the verge of launching, it ended up being left as an area for future improvement.

I wanted to solve this problem somehow and decided to run a UT utilizing a simple prototype. My hypothesis was as follows: we made them grab the device with both hands at first, but then told them to press a button, so they couldn't press the button while keeping their hands attached. However, getting them to press the button was ultimately just one of the methods to move on to the next step after having them strike the squat posture, so we would just need to change the method.

So we kept the text as is and removed the button. And we changed the design so that they would touch the screen one more time while holding the device with both hands.

We simply implemented that draft utilizing Figma's prototyping feature and tested it with people the next day. Fortunately, the results were positive, and we launched the Squat mission with that design reflected in the final implementation.

Link to the Figma prototype I made at the time

Fast Launch and Results

By repeating this fast improvement process through UTs, we were able to proceed with everything from planning to design and development in a short amount of time and launch the Squat mission. Of course, if we hadn't run UTs, we could have launched even faster. But doing that ultimately prioritizes the launch schedule over the user experience. For the Alarmy team, user experience is the top priority during product development, so even if the launch schedule was delayed a bit, we ran UTs multiple times to build a product that hit the mark for user experience.

We also worked as minimally and quickly as possible so we could pull the launch schedule up as much as we could. The squat guide video before launch was simply shot in the office, and for the design, we initially worked by focusing on the minimum features.

If we had tried to provide the perfect video and guide from the beginning, the launch date of the Squat mission would have been pushed back by a few more weeks. And thanks to these efforts, the Squat mission was able to become the second most popular premium feature after its launch, and Android is scheduled to be released soon too 🙂

Wrapping Up

Before joining the Alarmy team, the only UTs I had experienced were done internally, and all it was was asking simple questions to just around ten people. But the UT we ran to launch the Squat mission was much more full-scale. We ran UTs targeting various age groups, and to obtain untainted data, we even searched for users who had never used Alarmy before. It goes without saying how much I learned here. And even when running UTs, we recorded the user's process from waking up to actually completing the mission in text and video, and similarly recorded the draft's process of change accordingly. The fact that I can write a post like this right now is all thanks to a lot of data and meticulously recording it. Going through this process, I could feel the Alarmy team's affection for the product and the users, and on top of that, I could feel my own affection for Alarmy growing even bigger. That's probably why, even in the final step, instead of leaving the usability issue as an area for improvement, I wanted to dig deeper and solve the problem.

To me, the Squat mission isn't just a popular premium mission feature, but a feature that laid the groundwork for me to be able to say that if I had to pick one factor that determines a product's completeness, it's affection for the product.

Of course, this isn't the end. The algorithm for the Squat mission is scheduled to be further refined, and it will continuously be improved so users can use it more easily. If you struggle to get up in the morning, I really hope you give the Squat mission a try.