I tend to look at, and think in terms of, IT in larger organizations. Take a business with several thousand desktops and it's relatively hard to make a case for desktop Linux because the fundamental problem is the client-server architecture, not the specifics of the desktop OS.
Thus the righter answer for bigger businesses is usually to get rid of the problem by getting rid of the desktop OS -i.e. switching to the smart display architecture. Consider, however, a business with fewer than perhaps a hundred desktops in which one person constitutes the entire IT department. Running a couple of small business servers for this kind of service is trivial, it's desktop support that's the time killer - and the process of getting spending approved that's the change killer.
For that kind of environment switching desktop preferences to Linux can offer a big payoff both in terms of time savings for Mr. Everything IT and in terms of checks written for the business owner.
But the big problem on getting desktop Linux used is that public opinion is against you: because everyone knows that everyone uses Microsoft Windows and Microsoft Office, any deviation from that, especially one undertaken to save money, is seen and resented as the imposition of a second best solution. The result, if you're in this situation, is that the only desktops you can get changed for longer than it takes the boss to order an exception are those belonging to people who rank well below you in the unofficial but actual organizational heirarchy.
And, guess what? there's a catch-22 on this: the worse your problems have been in keeping up with Windows desktop support, the lower your organizational credibility and status will be, and therefore the harder it will be for you to improve desktop support and reduce your workload by swapping in Linux.
Try arguing with public opinion on a factual basis and you're dead: remember, you might get their attention for a few minutes at a meeting, but they already know, with certainty, that Wintel is the one socially blessed way and the PC press and its fellow travelers reinforce that certainty everyday, all day.
You can build a cost case with the one or two top people, but they're subject to the same presures and their commitment to cost control through change won't usually outlast the first whine by the first fellow employee who out ranks you in their eyes.
So what can you do?
There are some first steps that should be done before you go after the desktops: swapping in SAMBA on Linux for file and print services, moving web services to LAMP, and migrating any business applications you can to Linux and open source equivelents.
Next you should, as far as possible, change your own desktop to Linux - and if that means using terminal Server, or even hiding a PC somewhere and using VNC to give yourself desktop Wintel access, that's ok because the main thing is to be seen leading the change.
Beyond that, a strategy I'm growing fonder of is to look ahead to Microsoft Vista and the changes that's going to impose on your infrastructure to presell Linux as a downstream Plan B - which you'll find reasons to invoke as costs and compliance requirements become clear.
The most powerful option, however, is to skip the whole business of arguing cost and support complexity reductions to focus, instead, on "better."
That's an easy argument to make if you're proposing a change to smart displays because you can offer bigger screens, silent operation, improved reliability, access to a wider range of software, and the virtual elimination of PC style security concerns, along with new capital costs and balancing downstream support cost reductions.
It's a much harder argument to make for desktop Linux because you're still leaving a computer on everybody's desk and so find yourself saying that Linux is better and oh yes just about everything your users have been taught to believe about Windows is a lie.
That's a tough sell, and one you can skip by tying the change to a significent new capital investment in Systems. That's counter-intuitive, but works in part because no one believes better can be cheaper, in part because it provides a non technical focus for a key part of your decision jury to focus on, and in part because it will, later on, act as a barrier to recidivism.
The way to do that is to focus on a major new application your business can benefit from and then build your budget around it.
It takes time to do this kind of thing right: time to find the application and convince yourself it's what your group needs, time to build consensus among some key users, time to lay the groundwork for the proposal with your bosses, and time to put the whole thing together. With Windows NT 5.3 (aka Vista) due early next year the stellar clockwork may be aligning just right: start soonish, and your proposal will be on the table right about the time early cost and trouble reports on Vista start drifting into management's attention.
There's only one problem: there may not be a supported application to fit your group's needs -and that's something I'll look at in later blogs. For now, if you're in this kind of position: spend some time hunting up an applicable application or two, because if it's there, you may have the leverage you need to get that Linux desktop transition going.