A good theory should explain the observations or results of an experiment or phenomena, being understandable to a lay person while also being reasonable enough to allow for further testings. A good theory should also be frugal in their nature so others can test it, and they should also be predictive.
They should encourage further testing and expansions of the hypothesis. Good theories mean that others should be able to test them and, if possible, disprove them. This doesn't mean that theories should be disprovable, but that they should be designed so that they are neither impossible to be proved or disproved. In this way, theories should be made to facilitate further research and insights, not discourage them.
Theories should be able to predict what will happen from a given experiment. This is what gives them a better standing because their basis is not on pure speculation but informed hypothesizing. Good theories focus on the effects, not the causes of a phenomena. They are also never regarded as statements of fact, but instead of likelihood. Theories aren't regarded as facts because they are frequently revised and rethought. To say a theory is a fact, is to take away the notion that they could ever be further tested and reformed, a practice highly important and regarded by the scientific community.