After making sure that Sleepy works well at the lab environment it was time to test it on real users!
We gathered for a session and scheduled with a few people to come by and try Sleepy.
The whole thing was very exciting! We enjoyed watching people engage with Sleepy.
Some of the insights:
The feeling that sleepy is breathing with you and following your breath is pleasant and fun. The users mentioned it's pretty exciting and a bit like a game that they were checking to see if it really follows them, and it did ;)
The users mentioned that it is intuitive and feels right that the light dims during inhalation and becomes bright during exhalation.
The light colors (green and blue) are nice, strength of the light is good as well.
Some of the users didn't feel that sleepy became the Leader. The switch from Follower to Leader was completely seamless.
Some of the users though mentioned that they did feel the change and that is surprised them.
Sleepy's breathing pace was too fast sometimes at the Leader stage.
When sleepy went into a second calibration the light was still blue so the users didn't know what's going on (we need to change it to green).
Overall, our first real user testing went very well. Our tech team will work on the bugs that we found during the testing in order to make Sleepy better.
User testing demo:
Tech Work and Progress After User Testing
Blue instead of green when re-calibrating.
At the end of the first calibration Sleepy will now dim so it'll be noticeable that the calibration is ending.
The Leader is too fast when starting the Leader - This bug was really serious but after working on it, now it works great. We need a stable and long enough time to be able to sample the breathing qualitatively, so we had to increase the Follower time in case of a re-calibration. What we did is such a thing: We constantly measure and count breaths. If Sleepy goes into calibration, we simply ignore the measurements made in 10 seconds of calibration, and also ignore the previous 10 seconds of calibration (which were insufficient measurements, that actually caused the re-calibration). For each calibration, we increased the follower time by 10 seconds.