In general, I think it is important to assess success and failure regularly in the context of short-term manageable goals, and build the larger vision (and benchmarks for success and failure for that vision) based on the practical results of these short-term goals. Each interation - whether a success or failure - provides new information to the process, and helps provide better direction for future goal-setting.

If there is a particular aspect of the project that is a failure consistently, it becomes important to rethink the vision of the project such it incorporates the truth of that repeated failure.

One specific problem which I myself consistently encounter is that over-commitment to a goal leads to excessive self-sacrifice. While I'm generally happy to suffer on the crosses of my ideals, there does come a point when enough is enough...