i think it depends on the application in which the learning technique is used. For example in the case of task scheduling the minimum Makespan is a good measure. But as a general method you can use some data sets (if its available) to test the techniques and then investigate the outcomes based on the related measures in that field.
Since there is only 1 other answer, I will make a suggestion. I don't know of any 'accepted' measure of outcomes and do ot thnk therre can be an accepted measure because the measure of any outcome is a dichotomy, ie you successfully completed an outcome or not. The way I have used them in the past was to describe outcomes at various levels and then match the student performance with a particular level. My current use of outcomes however is to set tasks for student to perform, then judge the learnings implied by their performance of the task. By collecting evience of these learnings, I can plan the next learning for each student at an appropriate level. Possibly not what you were after.