The four-thousand-note test
Every note app is pleasant at forty notes. The ones worth choosing are the ones that are still usable at four thousand.
The trouble with choosing a note app is that the thing you need to know takes three years to discover, and by the time you know it you have three years of notes inside and no reasonable way out. So the choice has to be made on properties you can check on day one that predict what happens on day one thousand.
Search is the entire product
At forty notes you navigate by memory and folders feel important. At four thousand you navigate by search and folders are decoration. So the only question that matters is whether search is fast, covers the full text of everything, and looks inside attachments and images.
You can test this before you commit. Import a large folder of text files, or paste a long document twenty times, and search for a word that appears once near the end. If there is a visible pause, that pause becomes a reason not to open the app at all.
Structure that emerges beats structure you design
A folder tree designed on day one describes the work you expected to do. Two years in it describes nothing, because the actual shape of what you collected was not knowable in advance. Tags, backlinks and saved searches all let the structure arrive later, which is the only way it arrives correctly.
Can you read it without the app?
This is the insurance question. If the underlying store is plain text or Markdown in ordinary folders, nothing can be taken from you: the company can fold and your notes are still files. If it is a proprietary database, your exit depends entirely on an export feature continuing to exist and continuing to work.
Test the export before you need it. Export everything on day two, open the result, and see whether it is something a human can read.
Two taps from locked
Capture and organisation are different jobs. The thought you have while walking has a short life, and an app that needs unlocking, opening, choosing a notebook and picking a template will lose it. Some people solve this with a second, dumber app that does nothing but capture, and move things across later. That is a perfectly good answer.
What does not predict anything
The onboarding, the theme, the icon, the number of integrations, and how the app looks in a screenshot of somebody else's setup. None of it survives contact with four thousand notes.
Last reviewed 25 September 2026 · Productivity
A collection that follows this
More guides
How to tell a subscription trap from a fair price
Four questions that separate a tool worth paying for every month from one that should have been a single purchase.
Read →What a good sleep timer actually does
The single feature that decides whether you finish an audiobook, and why most implementations get it wrong.
Read →One frame, many worlds
Nine categories in the same quiet layout. 1 week free, then $11.99 CAD per week until you cancel.