It's easy to end up with a note app you never actually chose. You inherit one, and then everything goes into it: half-formed ideas, meeting minutes, the thing they promised to send on Thursday. Six months later, finding anything means scrolling. This is for the people who have switched apps twice and gotten the same result both times.
The verdict
The app is rarely the problem. The text you capture does three different jobs, and it's easy to run all three through one undifferentiated pile. Once the jobs are separated, the differences between tools matter much less. Leave them merged and no tool will save you, because every search returns a mix of things you were supposed to do, things you wanted to remember, and things you were just thinking.
The three jobs:
| Job | What it holds | How long it lives | "Done" means | Symptom when it breaks |
|---|---|---|---|---|
| Fleeting capture | Raw thoughts, quotes, half-questions | Hours to days | It got sorted or deleted | An inbox that only grows |
| Reference | Things you look up, not act on | Indefinite | You can find it again | You re-google what you already saved |
| Committed work | Anything with an outcome and an owner | Until it ships | It leaves the list | Promises buried inside long notes |
Fleeting capture: an inbox, not a library
Capture exists for one reason — to get a thought out of your head before it escapes, in as few steps as possible. Anything that slows down the moment of writing defeats it: choosing a folder, picking tags, deciding on a title.
The test: if I never open this again, do I lose anything other than the idea itself? If no, it's capture.
Capture has one hard requirement that people skip. It has to be emptied. An inbox that is never processed isn't a capture system, it's a landfill with a search bar. Emptying means one pass where each item goes to reference, becomes a committed task, or gets deleted. If you end up deleting a lot of it, that's a reasonable outcome, not a failure.
Reference: judged on retrieval, not on writing
Reference material is what you consult. Specs, how you configured something last time, the client's billing address, the recipe for the deploy script. You don't act on it; you look it up.
The test: if I lose this, do I have to reconstruct it? If yes, it's reference.
Reference is the only one of the three that should accumulate, and it's the only one where the writing effort pays off later rather than now. That flips the standard for quality. A reference note is good if you can still find it months later, when you have forgotten the exact wording, by searching a word you would plausibly think of then. It is not good because it's beautifully formatted. If you find yourself organizing reference material more often than you retrieve it, the system has become the hobby.
Committed work: it has a date and it leaves
Committed work is anything you have promised, to yourself or to someone else. It has an outcome attached and a person responsible.
The test: if this never happens, does someone notice? If yes, it's committed work — and it does not belong in a note.
This is where mixed systems do the real damage. A task written inside paragraph four of a meeting note is invisible. It has no date, it doesn't appear in a list, and nothing surfaces it on the day it matters. You only rediscover it when you reopen the note, which is usually after the deadline. Committed work needs a view that shows only committed work, sorted by when it's due, with nothing else in it.
Why mixing them fails, specifically
Three failure modes cover most of it.
Capture merged with reference. Every fleeting thought is treated as something to keep, so nothing is ever deleted. Search quality drops as the pile grows, because half of it is noise you wrote at 11pm.
Commitments merged with reference. Tasks live inside documents. There is no single list, so your sense of what you owe people is a memory exercise, and you re-read whole notes to check.
Reference merged with capture. You know you saved the thing, but you saved it as a two-word fragment. Finding it takes longer than looking it up from scratch, so eventually you stop trying.
Notice that all three are the same error in different clothing: one container being asked to satisfy contradictory requirements. Capture wants speed and disposability. Reference wants structure and permanence. Commitments want dates and disappearance.
What this means when you shop for tools
One app can hold all three, as long as it gives you three distinct surfaces. Judge candidates on that, not on feature counts:
- Capture speed. Count the steps from "I have a thought" to "it is saved." A global shortcut or a widget is worth more than a hundred templates.
- Retrieval. Full-text search that finds things by words in the body, not just the title. Whether you use tags, folders, or links matters far less than whether search works.
- A commitments view. A list you can open that shows only things with dates and owners, with reference material excluded.
- Export. You will change tools again. Being able to leave with your text intact is worth more than any single feature.
If your current app can do those four things, switching is unlikely to fix anything, because what you're missing is the weekly pass that sorts capture into the other two buckets. No app does that for you.
Who should skip this
- Your setup already works. If you can find things and you're not missing commitments, restructuring is procrastination with a nice interface.
- Your system is imposed from outside. If commitments live in a team tracker your company runs, that's your commitments layer. Don't build a shadow copy.
- Your real problem is volume. If you have too much work rather than badly-sorted work, three neat buckets will show you the same overload in better handwriting. That's a different problem.
- You're taking notes for one project with an end date. A single document is fine. Systems are for text that outlives its context.
The decision rule
If you can't say which of the three jobs a piece of text belongs to, the problem is your system, not the app.
Here's the audit. Open your notes, take the last twenty entries, and label each one F (fleeting), R (reference), or C (committed). How long this takes depends on how much you have accumulated. Two things will show up: entries you can't label at all, and entries that need two labels because a promise is hiding inside a reference note.
Then do one thing this week — pull every C out into a single list with dates on it. That's the bucket where being disorganized costs you something other than time.