Founders and creatives5 min read

Keeping an idea journal when you're building a company

Not a notes app and not a task list. A record of what you decided and why, which turns out to be the thing you cannot reconstruct later.

Founders are not short of capture tools. There is a notes app, a task manager, a doc somewhere, and four hundred messages to yourself. What is missing is almost never the ability to write something down.

What is missing is the reasoning. You will remember what you decided. You will not remember why, and the why is the part that turns out to be load-bearing.

The specific thing worth recording

Six months from now, someone — possibly you — will reopen a settled decision. The conversation goes: why did we do it this way? And the honest answer is a reconstruction. You remember the decision and you rebuild the reasoning backwards from the outcome, which means you rebuild it wrong, in a flattering direction.

A decision log fixes this, and it is a small, specific practice:

  • What I am deciding. One sentence.
  • What I am choosing, and the alternative I am rejecting.
  • Why. The actual reason, including the unimpressive ones — we are out of time, I do not want the argument, this is reversible so it does not matter much.
  • What I expect to happen, specifically enough to be wrong about.
  • What would make me change my mind.

That is five minutes, and the practice is well described elsewhere in more structured forms. The last two lines are the ones that pay. An expectation written down before the outcome is the only honest calibration data you will ever get about your own judgement, because memory adjusts expectations to match results and does it invisibly.

Why this is not your notes app

The distinction that matters: notes are reference, a journal is evidence.

Reference material is organised for retrieval of things you know you want. You go looking for the spec, the API key, the meeting notes. Fine — that is what Notion and Obsidian are for, and they are good at it.

A decision log is for something you do not know you want yet. The entry that matters in eighteen months is one you cannot identify today, which means no amount of tagging helps, because you can only tag what you have already named.

It is also why a journal in a structured database tends to decay — the schema has to be guessed up front, and the dimension that turns out to matter is the one you had no word for when you built it.

The second thing: the honest version

There is a register of writing founders almost never produce, which is the account that does not go in the investor update, the changelog, or the all-hands.

Not the dramatic version. The accurate one: what is actually working, what you are avoiding, which of your own stated priorities you have not touched in a month.

This has two uses, and the first is immediate — writing it down tends to settle it. The second is the real one. Over a year it becomes the only record of what building this was actually like, as opposed to the story you will tell about it afterwards, which will be considerably tidier and partly false. Narrative sets hard around a company fast.

The third thing: the recurring frustration

This is the one that justifies the whole practice, and it only works in aggregate.

You hit the same bottleneck in March, in June, and in September. Each time it reads as a bad week. Written down and seen together, it is not three bad weeks — it is one structural problem you have been routing around for seven months without ever addressing it, because it never once presented as a problem worth a decision.

Nothing but an archive catches this. You cannot notice it in the moment, because in the moment it is indistinguishable from an ordinary bad week. Your co-founder cannot tell you, because they see the same three bad weeks. It is only visible as a pattern, and a pattern needs a record.

Finding it by hand is the hard part — you described it differently each time and never gave it a name. That is the specific job of reading across entries by meaning, and it is most of why this is worth doing at all for builders.

Keeping it from dying in week three

Founder journals fail for predictable reasons.

Make it fast. No template, no required fields, no title. Open, type, close. Any friction is fatal, because this competes with forty other things and it is the only one with no deadline.

Write short. Four sentences. The idea that an entry should be substantial is the single most common reason people stop.

Separate it from work tools. Not because of secrecy, but because anything living in the work surface acquires the work register — you will write for an imagined reader and the honesty will go.

Accept gaps. A fortnight missed is not a failure. The apology entry is reliably the last entry.

Write on ordinary days. An archive of only crises has no baseline, so nothing in it can be compared to anything. This is the step everyone skips and it is the one that makes the record usable.

What to do with it

Monthly, ten minutes. Read the month. One question: what did I write more than once?

Quarterly, thirty minutes. Read the decisions from the previous quarter against what actually happened. Not to assign blame — to calibrate. You will find you are systematically wrong in one direction, and knowing which direction is worth more than any individual decision.

Annually. Read the year properly before concluding anything about it. Memory of a year is a poor source and the review built on it inherits the problem.

When reopening a settled decision. Read what you wrote at the time, before re-litigating. Half the time the original reasoning holds and the conversation ends in four minutes.

The honest case

This will not make you a better founder in any month. The entries are boring and the practice feels low-value while you are doing it.

What it does is make one specific thing possible: being able, in two years, to see what you were actually doing, in your own words, before you knew how it went. Nobody else will have that record and you cannot reconstruct it afterwards.

Most of what you will want is not the wins. It is the stretch in the middle where it was genuinely uncertain and you kept going — which is exactly the part the retrospective story erases.

A journal that reads itself back

Ember reads across everything you write and surfaces what keeps returning. Bring an old archive with you, or start tonight.

Keep reading