For six months I tracked my mood every day on a 1-10 scale, along with sleep duration, exercise (yes/no), screen time (hours on phone), social interaction (none/light/moderate/heavy), alcohol (yes/no), caffeine (cups), and a one-line note about the day.
I used a simple notebook. No app. Took about two minutes a day. Here is what the data showed at the end of six months.
The numbers
Total days tracked: 178 (missed a few during travel).
Average mood rating: 6.4/10.
Highest streak above 7: 11 days.
Lowest streak below 5: 4 days.
Days at exactly 6: 47 (clear default mode).
The strongest correlation
Sleep duration. Way stronger than I expected. Days following 7+ hours of sleep averaged a mood rating of 7.1. Days following less than 6.5 hours averaged 5.4.
That is a 1.7-point difference attributable to sleep. Nothing else came close.
This was a surprise because I would have guessed that exercise, social interaction, or work events drove mood more. The data was not even close. Sleep dominates.
The second strongest correlation
Exercise. Days I exercised averaged 6.9. Days I did not averaged 5.9.
One-point difference. Substantial but smaller than sleep.
What was interesting: this held even when controlling for sleep. Even on bad-sleep days, exercise improved mood. Even on good-sleep days, lack of exercise pulled it down.
The surprising weak correlation
Social interaction. I had assumed social days would be high-mood days. The data showed essentially no correlation.
What I think this means: social interaction matters, but the QUALITY matters more than the QUANTITY. Some social days were great. Some were exhausting. The variable I was tracking ("how much social interaction") missed the more important variable ("how good was the social interaction").
Lesson: not all the right variables are easy to track.
The screen time correlation
Negative correlation, mild. Days with over 4 hours of phone screen time averaged 5.8. Days under 2 hours averaged 6.7.
I do not know if this is causation. It could be that bad-mood days lead to more scrolling, rather than scrolling leading to bad moods. Probably some of both.
Either way, the days I felt best and the days I scrolled most were rarely the same days.
The alcohol pattern
Days with alcohol averaged 6.6. Days without averaged 6.3. Mild positive correlation.
BUT: the day AFTER alcohol averaged 5.7. Strong negative correlation for the next day.
Net effect: drinking made the immediate evening slightly better and the next day significantly worse. Total mood impact across two days: net negative.
This was data I did not want. But the pattern is clear.
The caffeine pattern
Caffeine had essentially no correlation with mood. Whether I had one cup or four, mood was similar. Caffeine clearly does things for me (energy, focus) but not mood.
The seasonal pattern
Mood ticked down in late November and December. Came back in February. The change was real but smaller than I expected — about 0.6 points lower in winter than the rest of the period.
I had assumed winter would crater my mood. The data showed it dipped, but not dramatically.
The day-of-week pattern
Mondays and Tuesdays averaged slightly lower (around 6.0). Saturdays and Sundays averaged slightly higher (around 6.8). Weekday spread was less than I expected.
The "Sunday Scaries" pattern was real but small. Sunday evenings rated about 0.4 points lower than Sunday mornings.
What I changed because of the data
Three changes based on what the data showed:
1. Sleep is non-negotiable. I now treat 7+ hours as a hard target. The mood return on sleep is so high that everything else has to organize around it.
2. Exercise on no-sleep days is even more important. The data showed exercise partially offset bad sleep. So I now exercise on bad-sleep days even when I do not feel like it.
3. Stopped drinking on weeknights. The next-day cost was too consistent to ignore. Weekend drinking only, and rarely.
What I learned about tracking
A few meta-observations from doing this for six months:
The act of tracking changes the behavior. Knowing I was about to write down "did not exercise" sometimes nudged me to exercise. This is sometimes called the Hawthorne effect. It is a feature, not a bug.
Daily logging is sustainable. Detailed logging is not. Two minutes a day is sustainable for years. 30 minutes a day burns out in two weeks. Track few variables.
Six months is the minimum useful period. Patterns at one month are noise. Patterns at six months are signal.
The data will tell you something you do not want to hear. Mine was the alcohol data. Yours will be different. Listen anyway.
This is not advice
Your patterns will be your own. The point of this writeup is not the specific findings. The point is that tracking reveals patterns intuition misses. Six months of two-minute daily logs gave me information about myself I could not have guessed.
Try it for a season. You will learn something.
