Long term, I think the most important metric should be increase in the value of Steem tokens, but it's not always going to be easy to match that to work that is done, of course.

So I agree that other metrics will be useful for measuring proposals, and there's not going to be one or even a few metrics that will be useful for the wide variety of potential proposals. Marketing proposals could be the easiest to measure, IF they can show successful adoption by new users (especially if the users are retained over time).

The impact of new infrastructural features gets more difficult to measure, and I doubt any single metric is going to work for such things. Personally, my original vote is going to be based on how useful I think such a feature is (coupled with the price asked/etc) and my continued vote is going to depend on how well the task is being executed over time.