Don't you do experiments to test hypotheses? An infinite number of experiments are possible, so you have to pick one. A novel prediction made by a theory is a good place to start.

And there lies the issue that string theory came up against: it has not produced any novel and feasibly testable predictions. But to be fair, most of the speculative fundamental physical theories suffer from the same issue. The origin form of the following experiment, proposed by an opponent (though not an aggressive one) of string theory, requires an interferometer larger than the solar system, arranged with precision accuracy that would make Rosetta reaching the comet look like a cakewalk: https://en.wikipedia.org/wiki/Free-orbit_experiment_with_las...