For instance, you tell the vendor that your entire site upgrade is contingent on having access to betas, previews, or some limited number of licenses for X number of months so that you may port and test your internal applications. This is hardly an uncommon problem when upgrading, and it's a good bet for a software vendor to allow small samples for a short period in order to enable larger purchases.
It also doesn't hurt to remind vendors of what you could be doing instead of putting up with them. Tell Microsoft you'll virtualize all of Windows on Mac OS X. Tell them you've been impressed lately by OpenOffice 3. It doesn't even matter if you have a real intention to switch, they just have to believe that you will.
I know this is what the newspapers say, but it's not clear to me that it's true. Saving implies that someone out there is paying interest for the privilege of using your money, and that in your own opinion, whatever they're doing is a better use than whatever you'd be spending it on if you had to spend it. Saying you should spend instead of save to help the economy sounds like saying you should deliberately lose money (if you thought it would make more money, you'd be doing it already, right?) to help other people make it... that is, you should be a charity.
It seems "only human" for people to cut back and save when the economy is bad, but unfortunately that is also the wrong thing to do in order for the economy to recover, and the wrong way to reboot a business. Now is exactly the time for companies to be investing in their futures, and I.T. is one of those major spending areas. With smart spending today, they could be extremely well positioned in a year or two.
A related problem is that I've yet to witness an I.T. group that operates incrementally, or in a parallel testing fashion; it seems that everything has to become a snowball in order to happen at all.
There are ways to make upgrades perfectly safe, and achievable over a longer period of time when resources are scarce. For example, machines could be upgraded a few at a time in isolation (ideally with a production-like parallel test environment), starting with the more experienced users. And, the culture could evolve to commit to incremental improvements in business applications on a regular basis, so that change is expected instead of being some rare event that scares everyone and threatens to tear the company apart at the seams.