Leaving OneDrive: move the cold half first

The list of safe European file platforms was longer four years ago than it is today. In November 2023 ownCloud GmbH of Nuremberg and DRACOON GmbH of Regensburg went to the same Californian buyer within a few weeks, along with a British provider.

Nothing about the software changed that day. Everything about the parent companies' legal position did.

Concluding from this that alternatives are pointless is the wrong lesson. The right one: operating it yourself, in an open format, is worth more than the flag on the website — because it survives a change of ownership.

The estate has two halves

In every company the file estate splits into two parts that pose entirely different migration problems.

The cold half is archives, closed projects, contract folders, image libraries, accounting years. Nobody co-edits there, nothing changes, and all that matters is that folder structure and permissions survive. This half moves over a weekend. In most companies it is by far the larger share of the volume.

The hot half is the folders where three people work in the same spreadsheet at once. That is not a storage problem but a co-editing problem — and it therefore gets solved later, together with the question of which office tools you will use.

Do both in one step and you no longer have a migration project; you have two at once, and you will fail at the second.

What is on the table

Nextcloud (Stuttgart) is the obvious destination: open source, self-hostable, with calendar, contacts and document editing bolted on. The catch never appears in the proposal: from now on you operate a server. Without a named owner for updates, that becomes a security problem in eighteen months — and the sovereignty gain turns into a risk.

Luckycloud (Berlin) offers hosted Seafile with end-to-end encryption; Filen (Germany) is consistently encrypted and a young product with a small team.

Tresorit (Hungary/Switzerland, owned for some years by the Swiss Post) is mature and expensive, with the best permission model in this group.

kDrive by Infomaniak and pCloud (both Switzerland) sit outside the EU, but under an adequacy decision and without a CLOUD Act equivalent.

The difference between these options rarely lies in the feature list. It lies in whether you want to run a server or pay an invoice.

Where it actually gets stuck

Path lengths and special characters. Grown file shares contain folder names with slashes, colons and three hundred characters. That surfaces during the move, not before. A test run with the largest folder settles it in an hour.

Share links. Everything you have shared externally points nowhere after the move. Often that includes quotes still open with customers. Pull a list of active shares before cut-over and replace the important ones deliberately.

The sync folder on every machine. If two hundred devices pull a fresh copy at the same time, your line is gone. Roll it out in waves.

Permissions nobody understands any more. Every grown estate has folders with individual grants for people who left long ago. The migration is your only opportunity in five years to clean that up. Take it — but budget the time.

What is better afterwards

The estate sits in one place you can name. Permissions are tidy. And you have a defensible answer to the question of how long a full extraction takes — because you have just done one.

That number is worth more than any assurance in a contract. It is the core of every Exit strategy, and it can only be established by doing, not by reading.

The honest part

Search across the whole estate is weaker with every alternative than with the large providers. The mobile applications are rougher. And browser-based co-editing stays slower.

Say so in advance. Leave it out and you lose credibility at the first incident — at which point the cold half that moved without friction no longer counts for anything.