The goal behind creating most heuristics is to find good solutions to a problem. What is a good solution? How can ``goodness'' be measured?
Normally, solution quality is measured in terms of percentage above the optimal solution (for minimization), though for some applications, an absolute difference is the desired measure. For instance, a good solution for the traveling salesman problem might be defined to be one that is within 10% of the optimal solution value. Some heuristics are better measured as the difference between its value and the optimal one. A heuristic to minimize the number of machines used to meet a certain production goal might use 2 machines more than optimal. In fact, a heuristic might always use no more than 2 machines more than optimal, no matter what optimal is. Here difference is a better measure than relative value.
The goal in most heuristics is to get good solutions for some set of problems that will arrive in the future. Since these problems are unknown (though aspects of them may be known), deciding between heuristics is a difficult thing to do. Even deciding what is meant by a good solution is hard to do. Here are some examples of what might be meant by good solutions: