Knowledge base › Tools

We have twelve comparison articles about which tool beats which. Not one about what comes next: how you carry your last two years across without losing them, and without the migration itself killing your habit. That is what this is — export, format loss, and the realisation that most of your old journal is not worth moving at all.

The real cost of switching is not the data

Most people fear losing data. In practice that is rarely what goes — the data can usually be exported in some form. What actually gets lost is the habit.

The mechanism is precise. Daily writing rests on a settled sequence of movements: this icon in this place, this screen, this button. On switching day that sequence becomes invalid, and until the new one settles — typically ten to fourteen days — every entry requires a separate decision. That is the window in which most switches die: not because the new tool is worse, but because the old automatism is gone and the new one is not there yet.

Export: what to take before you delete anything

Whichever tool you are coming from, the first step is the same: find the export option in Settings and download the full archive before you start moving anything. This is not part of the migration — it is the backup.

It is worth knowing what the formats mean in practice:

1. JSON. The most complete: usually every field, dates, tags, sometimes location. Uncomfortable to read, but the easiest to transform.
2. Markdown or plain text. Human-readable and importable anywhere. In exchange it typically loses the structured fields — the mood value, the tags, the relations.
3. CSV. Good for measurable data (date, score, tag), awkward for long text.
4. PDF. Excellent for archiving, useless for further work. If that is all you have, your data is now a picture, not text.

Attached images are a separate question. Many exports put images in their own folder and leave a reference in the text — if you do not keep the folder next to the text, the references are empty. This is the most common quiet loss in migrations.

What you ALWAYS lose — and why it does not matter

No migration is lossless. Three things typically do not survive, and it is worth accepting that up front, because otherwise the hunt for a perfect migration blocks the whole thing.

1. Formatting. Bold, lists, headings — every system stores these differently. The text comes across; its appearance rarely does.
2. Some of the metadata. The exact writing time, the device, the location, the mood value: these often find no new home in the destination.
3. Internal links. If entries pointed at each other in the old system, those links are typically dead after the move.

What always survives is the part that matters: the date and the text. And honestly: if you read an entry back in three years, you will not miss its formatting.

The 80/20 rule: most of your old journal is noise

This is the most uncomfortable and most useful claim in this article. If you read back two years of entries, you will find that most of it is a daily chronicle: what happened, who you spoke to, what you ate. That had its use at the time — but read back, it says nothing.

The valuable part is usually in three places. First: the turning points — those twenty or thirty entries where a decision was made or something changed. Second: the recurring complaints — the same sentence in six different months. Third: the early descriptions of things you now take for granted — these show how much you have changed, and they are the most common reason it is worth reading back at all.

You are not carrying two years across. You are carrying the thirty entries that made the two thousand worth writing.

From which follows the practical rule: do not migrate everything. Migrate the whole thing as an archive — one file, searchable, set aside — and lift out of it the twenty or thirty entries that actually matter. Those go into the new system as real entries.

Two strategies — pick one

The clean page. You start today in the new system, and the old material rests in one file as an archive. Upside: you start immediately, the habit does not break, and the old material does not dilute the new system. Downside: for the first weeks there is nothing to read back. This is the right choice if your old material is more chronicle than processing — which is the common case.

The seeded import. You bring those twenty or thirty selected entries into the new system with their original dates. Upside: you have a past from day one, and the read-back works immediately. Downside: half an hour to an hour of manual work. This is the right choice if you have at least a year of material with real turning points in it.

What we do not recommend is a full machine import of everything. Lifting two thousand entries into a new system does not give you a past, it gives you noise: search fills with results you never wanted to see again, and the new system feels crowded from day one.

The 45-minute migration protocol

It fits into one sitting. It does not need a free weekend — which is exactly why it gets done.

1. Minutes 0–5: full export. Settings → Export in the old tool. Choose the most complete format. Download it and put it in its own folder together with the images.
2. Minutes 5–10: backup. Copy the folder to a second location as well — the cloud and a physical device. This is the step that makes everything else safe to do.
3. Minutes 10–25: the selection. Skim the export and mark the twenty or thirty entries that are turning points, repetitions or early state descriptions. Do not read it all — you remember the turning points, look for those.
4. Minutes 25–40: entry. Put the selected entries into the new system with their ORIGINAL dates. The date is what matters, not the formatting.
5. Minutes 40–45: the first new entry. Write one for today, in the new place. That closes the migration and starts the new habit.

Where are you coming from? Four typical starting points

The details of the migration depend on what you had. Four cases cover the overwhelming majority.

From paper. There is no export here, and that is liberating: you do not have to digitise everything. The method that works is to keep the notebooks as a physical archive and type up only the twenty or thirty pages that are turning points. An hour of typing — and in exchange, what matters becomes searchable.

From a notes app. The most common case. The text is usually easy to get out, but the entries often live in ONE long note with no dates. Then half the migration consists of assigning dates to sections — which is exactly the work that argues for moving only the essentials.

From a journal app. Here there is a proper export and structured fields too (mood, tags). The trap: those fields are precisely what does not survive, so if your mood history matters, save it separately as CSV so it at least survives as a chart.

From Notion or a similar workspace. The most complex case, because there the journal is typically a database with views and relations. The relations never survive. The best strategy here is the clean page: the Notion page stays as the archive and you start today in the new system.

The date problem

There is one thing not to economise on during a migration: the original date. If your old entries arrive with today’s date, you lose the one thing journal data adds over plain text — its place in time.

In practice that means: before you enter anything, check whether the destination system allows backdating an entry. If it does not, you have two poor options: enter with today’s date (and the order is a lie), or write the date into the first line of the text (searchable, but not sortable). Neither is a tragedy — but it is better to know in advance than after five hundred entries.

If the destination does support backdating, the seeded import is clearly the better choice: thirty entries with original dates give you a usable past immediately, and the first read-back already happens in the new place.

What to do with the archive long term

The archive is not junk, it is insurance — but only if you can find it. Three rules are enough.

1. One file, not five hundred. Concatenate the export into a single text or PDF file, sorted by date. You will never search across five hundred separate files.
2. A meaningful name and two locations. “journal-archive-2019-2026.txt”, in the cloud and on a physical device. The value of an archive is that it still exists in ten years.
3. Open it once a year. Not to read: to check that it opens. An unverified backup is the most common form of a backup that does not exist.

What to do with the old app

Do not cancel it the same day. Give yourself ninety days of overlap, for three reasons. First: missing things usually surface in the first weeks, and you can still reach back. Second: if the new system does not work out after all, stepping back should cost nothing. Third: after ninety days you know whether you really switched — before that you only hope so.

After ninety days, though, do close it: the subscription you keep paying for years “just in case” is the most common hidden cost of changing tools.

Four mistakes almost everyone makes

1. Deleting before exporting. Once the account is closed the data is typically unrecoverable. Export is always the first step.
2. Hunting for the perfect migration. A week goes into researching formats while you write nothing. The loss is not the formatting — it is that week.
3. Moving everything. The new system is noisy from day one and search becomes useless.
4. Leaving the images behind. The text arrives, the image folder stays — and the references are empty. Always move images together with the text.

If you are coming from several systems at once — journal, habit tracker, goals in separate apps — then this article is only half the work: the strategy of consolidation is covered in its own article.

Frequently asked questions

Can I move all my old entries into another journal app?

The text and the dates usually yes; formatting and structured fields (mood values, tags, internal links) mostly not. The first step is always a full export from the old tool — including images, because those often go into a separate folder and the references are empty without them.

Do I have to move everything?

No, and usually it is not worth it. Most of two years of material is daily chronicle that says nothing when read back. Archive all of it in one searchable file, and move only the twenty or thirty entries that are turning points, recurring patterns or early state descriptions.

What is the biggest risk when switching?

Not data loss but the break in the habit. Daily writing rests on a settled sequence of movements that becomes invalid on switching day; until the new one settles — typically ten to fourteen days — every entry takes a separate decision. Most switches die in that window.

When should I cancel the old subscription?

After ninety days of overlap. That is when missing things surface, when stepping back is still painless, and by the end of it you know whether you really switched. After ninety days do close it, though — the old subscription kept “just in case” is the most common hidden cost of changing tools.

Which export format should I choose?

JSON if it is offered: it usually contains every field and is the easiest to transform. Markdown or plain text is human-readable but loses the structured fields. PDF is good for archiving and useless for further work — from then on your data is a picture, not text.

How long does the whole thing take?

Forty-five minutes in one sitting, if you do not try to make it perfect: five minutes to export, five to back up, fifteen to select, fifteen to enter, and five for the first new entry. That last step matters most — it is what starts the new habit.