I am thinking of complementing what @Mark_Bailey and @statman said with an experiment or two.
Starting with picking a common task which you and your colleagues need to do quite frequently, as they suggested.
Let me use fitting and analyzing linear regression models as an example.
Step 1. List the steps that you need to accomplish the task by using JMP. And list the steps that you need to accomplish the same tasking using another software of your choice.
Step 2. Start timing how long each approach would take for you and your colleagues to finish.
Step 3. Change the data set, go to Step 2.
Repeat Step 2 and Step 3. To see how much difference between them. If there is difference, then in the end, use what @SDF1 suggested to convert time into money. You may have to try some different tasks to be fair, I guess
The above experiment is for some repetitive tasks. Now try some spontaneous tasks. For this experiment, you have to find two individuals who are experts of the two software that you are comparing. So one is a JMP expert, the other individual is specialized in the other software of your choice. Now pick some non-trivial tasks. For example, some kind of predictive modeling. The assessment includes at least two parts: (1) timing (2) accuracy. What you are going to observe will be the achieved accuracy under the same time constraints. Or you may observe the different time consumption to achieve the same accuracy. Seems like that the second measurement might be easier to convert the difference into cash. Probably have to try several different tasks to be fair as well, I guess.
These experiments will reveal things that you need to consider more or less similar to what @ron_horne pointed out.
