← All posts
3 September 2025

What Stays in the Old App

Technical migrations succeed all the time. The files move. The knowledge usually doesn't.

Around 2015, Evernote had 150 million registered users. That number circulated for years like proof of concept — this is what a successful knowledge tool looks like. What circulated less: most of those users have since left. Downloads fell from 9.6 million in 2017 to 1.7 million in 2023. The exodus was quiet and distributed — a few thousand people at a time, moving to Notion, Obsidian, Bear, Apple Notes, Markdown folders, Roam, Logseq, and then sometimes moving again.

Nearly all those migrations technically succeeded. The files moved. Something else didn't.

Why we keep starting over

There's a well-documented phenomenon called the Fresh Start Effect, named in a 2014 paper by Katy Milkman, Hengchen Dai, and Jason Riis. The basic finding: people are significantly more likely to pursue goals after temporal landmarks — new years, birthdays, Mondays, the start of a new job. These landmarks create a mental separation between past-you and present-you. Old failures feel less personal. You feel free to begin again.

This is why gym membership spikes in January. It's also, I think, why we switch note-taking tools.

A new tool is a temporal landmark. It looks different, behaves slightly differently, promises a better system. You tell yourself: this time I'll actually maintain it. The old app's graveyard of half-finished notes belongs to past-you. Present-you is going to do this right.

The Fresh Start Effect is real and often useful. In most domains, starting over doesn't cost you much. A new gym routine builds on the strength you already have. A new diet doesn't erase the vegetables you ate last year. But in personal knowledge management, the history is the product. When you walk away from the old app, you're not leaving behind your failures. You're leaving behind your knowledge.

What the export doesn't include

Evernote's export format is called ENEX — technically an XML file, structured, inspectable, ostensibly portable. When you move notes to a new app via ENEX, a few things happen that the export guide doesn't mention prominently.

Internal note links break. These are the hyperlinks between your own notes, the connections that gave your knowledge base its texture. They depend on Evernote's internal ID system. Exported, they point nowhere. Your notebook structure disappears; ENEX merges everything into a flat list. Tag hierarchies collapse. And version history is gone entirely, which means every draft, every revision, every idea you recorded and then revised.

Attachments are unreliable. When importing into Apple Notes, users have reported seeing placeholder text where PDFs used to be. Large notebooks fail to import at all.

So what arrives in the new app? Text. The conclusions you wrote down. Not the connections between them, not the history of how you got there, not the structure you built up over years of use. Just the text.

Then something else happens, harder to document: you stop looking at the old notes. The imported content sits in the new app, organized slightly wrong, links broken, in a structure that doesn't quite fit. It's technically there. You never open it.

The thing you actually built

The problem is thinking of a notes app as a container. Containers can be swapped. You move the contents, throw away the box.

But a knowledge base — used seriously over years — is not really the container. It's the accumulated work of learning how to retrieve things. It's the specific tag that past-you chose so present-you could find it. It's a habit of opening a particular notebook before a particular kind of meeting. It's the mental model of where things live, calibrated to a specific tool over months of use.

None of that exports.

William Jones at the University of Washington spent decades studying what he called the re-finding problem — the difficulty of locating information you know you saved. His core observation: people are often good at capturing and terrible at retrieval. The system you built to help yourself retrieve isn't in your notes. It's in your head, tuned to a specific interface.

Switch tools, and you reset the retrieval system. The notes are there. You just don't know where they are anymore.

The format is the insurance policy

I'm not arguing you should never switch tools. Sometimes a tool genuinely gets worse. Sometimes your needs change enough that a different structure makes sense.

But the question worth asking before you switch: will my knowledge survive this?

The answer depends almost entirely on the format your notes live in. Plain Markdown files in a folder on disk survive migrations in a way ENEX files don't. SQLite databases are inspectable, portable, and readable with tools that will exist in twenty years. Proprietary binary formats are not.

This isn't just about avoiding vendor lock-in. It's about the specific observation that when you switch tools, you're not starting over — you're continuing with worse context. If your notes are in a format any future tool can read, you at least have the text. If they're in a format that requires a converter, you have the text minus the links, minus the structure, minus the history.

The difference between those two outcomes isn't which app you chose. It's whether the app respected your files enough to store them honestly.


Asgeir Albretsen is the founder of Harbor.

What Stays in the Old App: Harbor Blog | Harbor