The office suite goes last¶

A plain-language companion to the draft working paper "The Open Source Workstation: A Migration Guide for Applications, the Desktop, and the Managed Linux Estate" (v0.8, July 2026). The guide carries the product assessments, the reference architecture, the break-even model, the change-management chapters, and the source-by-source evidence; this text carries the ideas.

Read the full draft working paper (PDF)

The short version. Migrating the end-user workstation to open source is two decisions, and only the second one is hard. Replacing the applications while staying on the operating system you already run costs no licence money, needs no fleet-management change, installs alongside the incumbents, and is reversible at any point. Moving to Linux underneath is a multi-year programme whose difficulty sits almost entirely in the management layer rather than in the applications, which reached parity years ago. The guide's most useful finding inverts the usual plan: the Swiss study's own prioritisation method, applied to its own recommendations, puts the office suite and the mail client last and the small stateless tools first. And no source in the corpus supports a claim of savings inside three years.

Two stages, one of them cheap¶

Stage 1 is open-source applications on Windows or macOS. Nothing has to be switched off, because the new applications install in parallel. There is no operating-system change, no new management stack, no licence cost, and no point of no return. An organisation that completes Stage 1 and stops has moved its documents into an open format, made its applications portable, and removed a set of licence dependencies. That is a legitimate end state, and for many organisations it is the right one.

Stage 2 is the Linux desktop. The guide treats it as the destination while insisting the two stages are separable, which is where it departs from its Swiss source: that study excluded the client operating system by design. The work in Stage 2 is provisioning, policy, packaging, inventory, client identity, encryption, printing, remote support, and hardware. The applications are not the problem.

Start with the boring tools¶

Most migration plans open with the office suite, which maximises disruption at the moment the programme has the least credibility to absorb it.

Run the source study's own prioritisation method on its own recommendations and the order comes out backwards from that. Media player, password manager, archiver, PDF viewer, image viewer, desktop search: these carry no shared formats, install in parallel, cost nothing, and break nothing when they arrive. Doing them first builds the packaging, distribution, and support machinery that everything later depends on, so that by the time the office suite moves the organisation has done this a dozen times and the service desk knows how.

The German and French sources agree on the same logic one layer down: the operating system changes last, after every layer above it has already moved.

What it costs, without guessing your contract¶

The guide refuses to assume what you pay today. It derives instead the incumbent per-seat cost at which migration breaks even:

Euros per seat per year. Above the figure, migration is cheaper over that horizon; below it, the incumbent is.

Two features are specific to the workstation. There is no subscription line to remove, because none of the recommended applications charges one, so the entire cost is effort and disruption. And productivity loss dominates at scale: at 25,000 seats, one lost user-day per person outweighs the technical effort several times over. Every change-management recommendation in the guide is, read economically, an attack on that one term.

Indicatively, Stage 1 runs 12 to 24 months and can start immediately. Stage 2 adds 24 to 48 months and should not begin until Stage 1 is well advanced and the management layer is proven. The two documented field cases ran four to seven years end to end, at the pace of hardware renewal.

The five things that actually decide it¶

The line-of-business application portfolio, which most organisations do not fully know because departments bought much of it without telling anyone. This is the workstream that overruns.

Design, pre-press, and engineering populations. The field evidence rates professional graphics maturity at 1 out of 5 and says migrating expert users is not currently possible. Plan the exception rather than arguing with it.

Regulated workflows needing signatures in proprietary formats, which remains unsolved.

Whether the IT department is willing. The only observational evidence in the corpus, from the French study, finds internal resistance stronger than user resistance. This is usually the binding constraint and rarely the one in the plan.

The hardware and peripheral estate, smartcards included, which is a common late blocker.

What no source supports¶

Savings in the first three years. The most rigorous source in the corpus concludes long-run cost parity with savings only after several years, and the French study contains no cost figures at all, recording that cost reduction was noticed after the fact rather than being the motive. The case for migrating rests on exit cost, exposure to price escalation, and optionality. Presented as a discount, it invites a comparison it will lose.

The one thing to do this quarter¶

Install the incumbent's fonts wherever the office suite runs, which is the cheapest quality improvement available anywhere in the guide, and deploy the top of the priority list. That costs nothing, risks nothing, and builds the capability the rest depends on.

The small print¶

This is a draft working paper, a practitioner reference rather than an economics paper. Product statuses are stated as of a given date, and entries the guide marks as moving need re-checking before any procurement decision. The break-even figures rest on stated assumptions about effort, disruption, and horizon, and are meant to be re-run with your own numbers. The companion volume covers the back end: directory, storage, mail, and real-time communication.