To help me, I have everyone run a speed test quarterly. The test is simple and comes in three parts;
- Open AutoCAD and tell me how long it takes to get to cross hairs from double clicking the launch icon.
- Open a medium sized file I placed on the network and tell me how long it took.
- Open a large file I placed on the network and tell me how long it took.
There is an element of trust involved in having users submit times themselves. Another problem I have is getting results at all. This can be circumvented by running the test yourself off hours. In fact that might help keep it more fair by the removing varying network traffic from the equation. I have also seen lisps out there to gather data like open times, although I have not run any.
How ever you do it, knowing the facts vs. someone's perception is a much better place to be in.