Most R&D or experimentation can be done on a "no licence cost" basis. The one cost that NEVER seems to be allowed is time. The nearest I have seen is when a project manager allows a little time for a team to get up to speed with a new technology.

One of the thoughts must be given weight when selecting software to use for experimentation, R&D or a pilot is what is the plan if it is successful? Stick with that technology? Migration to a different technology? etc.

It is not necessarily a cheap way of doing it when the management shout "Ship it!!!" and the support team refuse to support it without additional training as it is nothing they have ever supported before.

NOTE: This is not an anti-OSS rant but a piece of cautionary advice.