What 113 Days of Tracking Myself Taught Me (and What It Didn't)

Water beat exercise as the strongest signal for my focus, and sleep showed almost nothing. Here is the data, and why I don't fully trust it.

2026-11-20
4 min read

Share

For 113 days I logged 28 things about myself every day, as a statistics project at RKMVERI: sleep, water, phone time, workouts, meditation, mood, and how long I could focus. The only subject I could observe every day was me. Here is what the data said, and just as important, what it did not.

The full interactive version, with the calendar and charts, is in the Quantifying Myself case study.

The setup

  • 113 days from 16 September to 6 January, with up to 28 variables logged daily.
  • The thing I wanted to understand was focus, recorded on a 1–5 scale from "under 10 minutes" to "over 90 minutes" of uninterrupted work.
  • Everything else was a candidate explanation: sleep, water, phone time, exercise, meditation, mood and health.

What I found

I measured how each habit moved with my daily focus score using the correlation coefficient rr.

  • Water intake: r = +0.58 (113 days)
  • Workout time: r = +0.41 (85 days)
  • Phone time: r = −0.35 (113 days)
  • Meditation: r = +0.24 (113 days)
  • Health condition: r = +0.18 (113 days)
  • Mood: r = +0.09 (112 days)
  • Sleep hours: r = −0.10 (113 days)

Water was the surprise. It was the strongest signal in the whole dataset, stronger than exercise, and nothing I would have guessed. Phone time was the clearest negative: more hours on the phone, shorter stretches of focus. Sleep, the thing everyone expects to matter most, showed almost nothing.

Why I do not trust that headline

This is the part that matters most, so I will be blunt.

  1. One person. These are patterns in my life over one season, not facts about people.
  2. Correlation is not cause. Drinking more water might not make me focus better. On days I was already settled and working at my desk, I may simply have had a glass next to me.
  3. The sleep result is probably a measurement problem. My average was 7.1 hours, sleep was recorded as hours and not quality, and a few days had extreme values (one day logged 12.8 hours, another 2.9). A weak correlation on noisy, narrow data says "I couldn't see an effect here", not "sleep doesn't matter".
  4. Some variables have fewer days. Workout time has only 85 days of data, so its estimate is less certain than water's 113.
  5. I tried many comparisons. With seven variables, one of them will look good by luck some of the time. The honest move is to treat the top result as a hypothesis to test next, not a finding.

What the numbers did change

Even with those caveats, the log showed me things I would not have noticed otherwise:

  • I exercised on 85 of 113 days and meditated on 76, so those habits were far more consistent than my impression of them.
  • Only 39 of the 113 days reached an hour or more of focus (levels 4 and 5), which is a number I would not have guessed.
  • Water and phone time are the two cheapest variables to change, so they are the obvious first things to test next, one at a time.

The lesson for my day job

Tracking yourself is a small version of every analytics project. Decide the question before you collect, record the thing you actually care about, say how noisy your measurement is, and report the effect with the doubt attached. The habit of asking "how was this number produced, and how wrong could it be?" is the most useful thing I took from that statistics degree.

If you want to start, my earlier beginner's guide to quantifying yourself covers the tools.

Quantified self
Statistics
Correlation
Productivity
Data analysis
Focus

You may also like