Do you or your organisations evaluate the impact of training (typically Kirkpatrick's Level 4)?

If you or your organisation don't, what factors prevent evaluating for impact?

If you or your organisation do evaluate impact...
* When do you do it?
* Do you evaluate every programme? Every course? Or just a few? Why is this?
* Which methods or models do you use and why?
* Do you try to isolate the effects of training from other factors? Why/why not? How?
* What measures do you use?
* Who asks for these evaluations?
* Who else is interested in these evaluations, even if they don't ask for them?

If you could wave a magic wand so that evaluating the impact of training was as perfect as you need it to be, what would this perfection look and feel like for you? Stakeholders? Learners? Customers?
Martin Schmalenbach