The pattern is so consistent it is almost a genre. Someone reads about a second brain or a personal operating system, spends a weekend building a beautiful structure in Notion or Obsidian, uses it religiously for eleven days, patchily for three weeks, and never opens it again. Six weeks, give or take.

The usual explanation is discipline. It is not discipline. The systems that die have a specific defect, and it is visible on day one if you know what to look for.

The defect: your system costs more than it returns

Every record you make has a price — the seconds and the small act of will it takes to open the app, find the right place and fill the fields. Every record also has a return, but the return arrives later, sometimes months later, when the record answers a question you could not otherwise answer.

In week one the price feels trivial because the novelty pays for it. By week six the novelty is gone and the return has not arrived yet, because a month of data is the minimum before anything can be compared to anything. That gap is where systems die.

A life OS you have to maintain is not an operating system. It is a hobby, and hobbies get dropped when the week gets hard.

So the design problem is not “what should I track”. It is “how do I make recording cost almost nothing, and make the first return arrive as early as possible”.

Rule one: the entry point must be cheaper than remembering

If writing something down takes longer than holding it in your head for the next hour, you will hold it in your head. And then lose it. This is why beautifully structured databases with twelve properties per row fail: each entry is a small form, and forms are work.

The cheapest entry point that exists is a message in a chat you already have open. No app to launch, no section to choose, no fields. You write “call Denis tomorrow” or “3 400 on groceries” the way you would text a person, and something else decides where it belongs.

If your system does not have an entry point that cheap, add one before you add anything else.

Rule two: some of it must arrive without you

A system fed entirely by hand has one failure mode: a bad week, and the record has a hole in it exactly where the interesting data would have been. The weeks you stop recording are the weeks worth knowing about.

So at least one stream should arrive on its own. A watch sending sleep, stress and heart rate. A calendar feed. A bank export. It does not matter which — what matters is that when the manual half goes quiet, the automatic half keeps the timeline continuous, and later you can see what that quiet period actually looked like.

Rule three: build the combined view, not the list

This is the part people skip, and it is the part that produces the return. A list of expenses is a list. A month of expenses against how much of the month has passed is an answer. A sleep chart is a chart. Sleep laid against the fortnight you worked late is an explanation.

  • Spending against time elapsed, not spending alone
  • Tasks against the goals they are supposed to serve
  • Sleep and stress against the weeks that were hard
  • This month against last month, per category

Every one of those needs two things in the same place. That is the entire argument for a single system over eleven good separate apps, and it is the only argument that holds up.

Rule four: decide what it is for, in one sentence

Systems that try to hold everything hold nothing. Before building, write one sentence: what question do I want this to answer that I currently cannot? “Where does the money actually go.” “Whether I am working on the thing I say matters.” “Why some months feel terrible.”

Then build only what that sentence needs. You can add the rest later, when a second question shows up on its own. A system grown from real questions survives; a system built from a template is abandoned, because nothing in it was ever load-bearing.

What the first month actually looks like

Week one: you record almost nothing and feel it is not working. Week two: you record a few things a day, mostly money, because money is the easiest thing to notice. Week three: the automatic streams have built up enough that you look at them out of curiosity. Week four: the first comparison becomes possible, and it tells you something you half-suspected but had never seen.

That fourth week is the whole point. Everything before it is deposit, not return. If you know that going in, the six-week cliff stops being a cliff.