It started, as most bad ideas do, at 2 AM on a Wednesday. I was lying in bed, unable to sleep, scrolling through a Hacker News thread about a guy who had been tracking every aspect of his life in a personal database for years. Sleep, meals, mood, money, workouts, books, conversations — everything. His dashboard looked like mission control at NASA, except instead of monitoring a spacecraft, he was monitoring whether he ate enough vegetables.
"I could do that," I thought. Famous last words.
Three months later, I have a database with 47 tables, over 12,000 rows of personal data, and a growing suspicion that I know too much about myself. Here is what happened when I tried to quantify an entire human life using nothing but free tools and an unhealthy amount of determination.
The Setup: Choosing the Stack
First decision: where does all this data live? I considered a few options:
- Google Sheets — too slow once you hit a few thousand rows, and the formula syntax makes me want to cry
- Notion databases — pretty, but the API rate limits would throttle my automated imports
- Airtable — nice middle ground, but the free tier caps at 1,000 records per base. I was going to blow past that in week one.
- PostgreSQL + Metabase — this is what I went with. Free, unlimited, and I get actual SQL queries plus beautiful dashboards
I spun up a small PostgreSQL instance on a $5/month DigitalOcean droplet (yes, I could have used SQLite locally, but I wanted remote access). Metabase runs on the same server for visualization. Total cost: $5/month. Less than my daily coffee habit, which — spoiler alert — I now have a very detailed record of.
What I Track (The Reasonable Stuff)
Sleep
My Apple Watch already tracks this, but the data was sitting in Apple Health doing absolutely nothing. I set up an automated export using a shortcut that runs every morning and pushes the data to my database via a simple API endpoint. Time asleep, time in bed, heart rate during sleep, sleep stages.
After three months of data, I discovered that my sleep quality drops by 23% on nights I eat dinner after 9 PM. Not groundbreaking science, sure. But seeing the actual numbers — a clear, unmistakable correlation plotted on a chart I built myself — hit different than reading "don't eat late" in some health article for the fifteenth time.
Food and Coffee
I log every meal and every coffee. Before you call me obsessive — you're right, I am. But here's the thing: I'm not counting calories or macros. I'm tracking what I eat, when, and how I feel after. Three simple fields.
The coffee data was immediately useful. I was drinking an average of 4.3 cups per day. On days I had more than 3 cups before noon, my afternoon focus score (more on that later) dropped by about 30%. My friend Raj, a data scientist, looked at my chart and said, "Dude, you spent three months proving that too much coffee makes you jittery. Congratulations."
He's not wrong. But I cut back to 2 cups before noon and my afternoons genuinely improved. Sometimes you need a graph to believe what your body is already telling you.
Money
I export transactions from my bank weekly via CSV and import them into a spending table. Categories, amounts, timestamps. I built a Metabase dashboard that shows spending by category over time.
The insight that actually changed my behavior: I was spending $847/month on subscription services. Not a typo. Eight hundred and forty-seven dollars. When I pulled the report, I stared at the screen for a solid minute before saying out loud, to nobody, "Who authorized this?"
The answer, of course, was me. Over the span of about two years, I had signed up for streaming services, productivity tools, gym memberships (plural — don't ask), cloud storage, VPNs, and about fifteen other things I had mostly forgotten about. I canceled $400/month worth of subscriptions that afternoon.
Focus and Productivity
This one's my favorite. I use a simple 1-10 self-reported focus score, logged three times a day: morning, afternoon, evening. Takes about 5 seconds each time via a shortcut on my phone.
After 90 days, the patterns are obvious. My peak focus is Tuesday and Wednesday mornings. My worst is Friday afternoons (surprise, nobody). But here's what I didn't expect: my Monday morning focus score is consistently higher than Thursday morning. I always assumed I hated Mondays. Turns out, I actually show up pretty sharp on Mondays. It's Thursdays that are the problem.
What I Track (The Weird Stuff)
Conversations
Not the content — I'm not recording anyone. Just metadata: who I talked to, how long, in person or virtual, and a 1-5 "energy" rating for how the conversation made me feel. This sounds creepy when I describe it out loud. My partner, Lisa, called it "quantified friendship" and asked if she was going to get a quarterly performance review.
She was joking. I think.
But the data showed something genuinely useful: I was spending about 6 hours a week in meetings that I rated 1 or 2 out of 5 on the energy scale. Six hours a week of conversations that actively drained me. I started declining the ones that weren't mandatory, and my average weekly energy score went up by 0.8 points. That's noticeable. That's an entire mood tier.
Weather Correlation
I pull daily weather data from Open-Meteo (free API, no key needed) and store it alongside everything else. Then I cross-reference it with mood, productivity, and sleep.
Turns out, I am significantly happier on overcast days than sunny ones. This goes against basically every wellness article ever written. But the data is clear: on cloudy days, my mood score averages 7.2/10. On sunny days, 6.4/10. My theory is that sunny days make me feel guilty for being indoors working, while cloudy days give me permission to be a hermit. Lisa says I'm "built wrong." She might be right.
The Dashboards
Metabase is doing the heavy lifting here. I have four main dashboards:
- Daily Summary — sleep, meals, focus scores, weather, all on one screen. I check this every morning over coffee (which I now limit to 2 cups, thank you very much).
- Weekly Trends — rolling averages for everything. This is where patterns become visible.
- Money — spending breakdown, subscription tracker, monthly comparisons. The dashboard that saved me $400/month.
- Correlations — the fun one. Cross-referencing sleep with productivity, weather with mood, coffee with focus. This is where the surprising insights live.
What Went Wrong
The Logging Fatigue
For the first two weeks, I was religious about logging. Every meal. Every conversation. Every focus score. By week three, I started forgetting. By week five, I was backfilling data from memory, which is basically just making stuff up with extra steps.
My solution: automate everything possible. Sleep data comes from my watch automatically. Bank transactions export automatically. Weather data pulls automatically. The only things I manually log now are meals, focus scores, and conversations — and I simplified the meal logging to just a photo and a timestamp. A custom shortcut handles the rest.
The Analysis Paralysis
Having too much data about yourself is weirdly paralyzing. I spent an entire Saturday afternoon building a dashboard that cross-referenced my protein intake with my gym performance with my sleep quality with the lunar cycle. (The lunar cycle was a joke. Sort of.)
At some point, you have to stop analyzing and start living. I now have a rule: I review dashboards once in the morning, once on Sunday evening. That's it. No mid-day optimization spirals.
The Privacy Moment
About a month in, I realized I had built an incredibly detailed profile of myself on a server connected to the internet. Sleep patterns, financial transactions, social connections, daily routines — if someone compromised that $5 DigitalOcean droplet, they would know more about me than my own mother.
I immediately moved the database behind a VPN and enabled encryption at rest. I also set up automated backups to an encrypted local drive. Should have done this on day one. Learn from my laziness.
Was It Worth It?
Honestly? Yes. With caveats.
The tangible wins:
- $400/month saved from catching runaway subscriptions
- Better sleep from discovering the late-dinner correlation
- 6 hours/week reclaimed from dropping low-energy meetings
- Fewer jittery afternoons from limiting coffee intake
The intangible wins:
- I understand my own patterns better than I ever have
- Decision-making feels more grounded — "what does the data say?" is a surprisingly useful question for personal choices
- I learned PostgreSQL and Metabase, which are useful professional skills
The downsides:
- It's a time investment, especially in the first month of setup
- You will discover things about yourself you did not want to know (I spend how much on takeout?)
- There's a thin line between self-awareness and self-surveillance, and I am not always sure which side I'm on
Would I recommend it? For data-curious people who enjoy tinkering, absolutely. For everyone else, maybe start with just tracking one thing — sleep, spending, or mood — and see if the insights change your behavior. If they do, add another table. If they don't, you just saved yourself from my level of obsession.
The trend of "life databases" is picking up fast — I have seen variations popping up on Hacker News and Reddit almost weekly. If you have built something similar, I would love to hear what you are tracking and what surprised you most. The weirdest correlations are always the best ones.
