Twenty real documents decide your office suite

In most organisations the decision about the future office suite is made the wrong way: with a feature list, a price comparison, and a trial installation in which somebody creates an empty document and confirms that bold text works.

That answers none of the question that matters. The question is not "can this program do word processing?" but "do our files survive the switch?"

The test

Take twenty documents from the real estate. Not twenty test files — twenty that are actually in use:

The quote with the company layout and the footer containing the legal form. The costing sheet with nested formulas and sheet protection. The form that goes to a public authority. The presentation that runs on a stranger's laptop at a customer site. The mail merge with a data source. And the one file everybody knows is complicated.

Open each of them in LibreOffice, in Collabora and in ONLYOFFICE. Save, close, reopen — the round trip is the actual test, not the first open.

Record only three states per file: unchanged, cosmetically shifted, unusable.

This is what the grid looks like. The rows below are an example, not a measurement — your values only appear with your files.

After half a day you have a basis for decision that no comparison site can supply, because it was produced with your files.

What comes out of it

ONLYOFFICE stays closest to the original for Word and Excel formats. If your correspondence leaves the building and is opened in Microsoft Office at the other end, that is the decisive point. The vendor, Ascensio System SIA, is registered in Latvia; its development history and roots reach back to Russia. For a trades business that is a non-issue; for a public body or an operator of critical infrastructure it is a point to check.

LibreOffice wins at everything meant to run locally and without a server. It is also the only tool in this group carried by a foundation, owned by nobody who could sell it — which, after the acquisitions of recent years, is no longer a marginal argument.

Collabora brings browser-based co-editing and fits into Nextcloud. It is noticeably slower than Microsoft 365 once three people work in the same spreadsheet.

CryptPad is the choice for content the server should not be able to read: end-to-end encrypted, in exchange for fewer features.

And Docs, from the collaboration between the French and German public-sector digitalisation bodies, is the first tool in this group where two states build the foundation together rather than each on their own.

What else the test finds

It finds the macros. Nearly every organisation has between five and twenty spreadsheets with built-in automation — the quote calculator, the commission run, the holiday planner. They often come from a person who has left, are documented nowhere, and appear on no project list.

If they surface during the round trip, you have a problem you can plan for. If they surface at month end after cut-over, you have a different one.

Sort the findings: whatever is rarely used and blocks nobody gets dropped. What remains — usually two to five genuinely business-critical files — gets rebuilt, as a script or as a function in the new system.

Reading the result correctly

If three of twenty documents come out of the round trip unusable, that is not an argument against switching. It is three templates to rebuild — a quantifiable effort you know before the decision instead of after.

If twelve of twenty are unusable, it is an argument for treating document editing as the last stage and moving everything else first. That is a result too. Open standards help more here in the long run than any format conversion: what once sits in an open format survives the next switch without this test.