Making the next switch cheaper

You will do this again. Not necessarily soon, and not necessarily by choice — a product changes, an employer changes, a plan you were on stops existing.

The switch you just finished was expensive because the things you cared about lived inside an account. The next one costs an afternoon if they live in a folder instead.

The four files

Everything that made this hard reduces to four artifacts, and all four are plain text.

The instruction file. Your standing instructions, written as a document rather than typed into a settings field. Built once, pasted into whatever product you are using.

The prompt library. The handful of prompts you actually use, with their triggers and placeholders, in one file.

The reference folder. The documents you attach constantly, one canonical version of each, in formats that will still open.

The context brief. A page describing your work in the terms you use, for seeding a cold account.

That is the whole kit. It fits in one folder, it is small, and it is the entire difference between a month-long migration and a morning.

Keep them in one place, with the archive

One folder, in your documents, backed up like anything else you would be annoyed to lose. The archive from this switch goes in it too, in its own subfolder, with the note recording what it contains.

The property that matters is that no part of this folder depends on a product being available. That is not a philosophical point — it is the specific thing that made this switch expensive, and it is fixed by location rather than by discipline.

The habit that keeps it true

When you improve something in the product, improve it in the file the same day.

This is the only ongoing requirement and it is where the approach usually fails. You refine an instruction in the settings field because that is where you noticed the problem, the file quietly falls a year behind, and at the next switch you paste something you fixed three months ago and cannot remember fixing.

Same day, or it does not happen. If that feels like too much friction, the alternative is a monthly reminder to reconcile the file against what is in the product, which is worse but survivable.

Write down what broke this time

While the switch is fresh, add a fifth file: two paragraphs on what went wrong.

The export that was missing attachments, the shared link nobody found until a colleague complained, the integration that was still authorised four months later, the monthly task you did not test. Whatever it was for you specifically.

Nobody remembers this in two years, and it is exactly the list that would have saved you a week. It is also the material that turns the generic order of operations into your own checklist, with your own accounts and your own failure modes in it.

Reduce what you accumulate inside a product

The other half of the work is not building artifacts but declining to build dependencies.

Three habits, all small. Send colleagues text rather than shared links, so nothing you gave someone depends on your account existing. Upload from your reference folder rather than from wherever a file happens to be, so the folder stays authoritative. And grant integrations narrowly and deliberately, so the revocation list stays short.

None of these cost anything at the time. All of them are steps you did not have to do this switch.

What moves

WHAT MOVES — the portable kit

  · Instruction file, prompt library,
    reference folder, context brief
                    → move to any product, in an
                      afternoon, permanently.

  · Notes on what broke last time
                    → move, and are the most valuable
                      page in the folder.

  · Accumulated context in the new account
                    → STAYS BEHIND at every switch. Not
                      fixable, only shortened.

  · The organisational layer, wherever you
    are
                    → STAYS BEHIND every time. Rebuild
                      only what you use.

  · Files kept only inside a product
                    → STAYS BEHIND. This is the whole
                      lesson.

  · Skipping the export before cancelling,
    next time as well
                    → IRREVERSIBLE. The rule does not
                      get easier for being familiar.

Take the export annually, not only when leaving

One reminder a year, on the account you currently use: request an export, verify it opens, store it alongside the last one, delete the older one if you like.

This changes the character of every future switch. An archive that already exists is not a deadline, which means cancelling stops being the step that can destroy something, which means the whole sequence relaxes into ordinary administration.

It takes fifteen minutes a year and it is the single highest-value habit in this subject.

What never gets cheaper

The cold start. Every switch, on every product, you arrive with a page of context where the last account had a year of it, and for a fortnight the new tool feels worse than the one you left.

The kit shortens that and does not remove it. There is no artifact for accumulated fit, no product exports it, and nothing you write down substitutes for the months of small adjustments that made the last setup feel like yours.

That is the standing cost of leaving, and it is worth knowing it is the only one that remains. Everything else — the archive, the instructions, the prompts, the documents, the sequencing — is solved work, done once, in a folder you own.