A few obvious caveats first: yes, clearly no conclusions can be drawn from two days of data. But the differences are fascinating, and can be correlated with both objective data (via the Sleep Cycle sleep analysis app) and subjective data (I feel great today.) Some notes:
The Sleep Quality metric, as defined by the Sleep Cycle app, is pretty vague and ought to be taken with a grain of salt, but a solid 24% jump is not to be ignored. Both nights were slept in the same room, the same temperature, and the amount of darkness; variables that I failed to control include step count and the time I went to sleep (strangely though, I slept better with fewer steps, and after staying up later). Note the time in bed as well: both nights were slept without an alarm clock, as I'm a believer in waking up when one feels ready to. According to the Sleep Depth chart at the top of the app, the supplemented sleep was much deeper.
My initial conclusion is to continue with magnesium citrate supplementation on a nightly basis, as there seems to be zero downside. (It's important to note that most Americans are deficient in magnesium anyway, so supplementation wouldn't hurt even if sleep quality was unaffected.)
The purpose of this article was also to briefly highlight the importance of n=1 experimentation. This is a great use of the Quantified Self approach: track a metric, toss in a variable, and see the change. Remember, tracking data without any informed approach or end goal is close to useless. Experiment, try things, measure what happens. Some life-changing habits can be formed with this approach.