It is not possible for an un-falsified theory to be a failed theory. Instead, among theories there is only more interesting and less interesting. It is up to each researcher to individually decide which theory they think is the most interesting, on the basis of what is the most promising. There is no scientific basis for doing that, because by definition we are talking about trying to guess which theories
future evidence will support.
Some people argue that since so much of the supersymmetric theory parameter space has been ruled out, that this reflects negatively on supersymmetry as a whole. However you must understand that this is exactly the kind of non-empirical argument that some stringers get so much flak for deploying. Nobody knows what the prior distribution should be over parameter space and it is not possible to do something like applying a uniform distribution, because if you do something as simple as algebraically re-arrange the parameters the uniform distribution over the new parameters will not be equal to the uniform distribution over the old parameters.
I can give you a concrete example to illustrate the fallacy of trying to measure the "volume" of the ruled-out parameter space. Suppose there is a grain of rice somewhere on a chessboard, and you are trying to find it by asking a person who can see the board whether the grain is in zone 1, zone 2, and so on. You count all the way up to zone 63, and the person who can see the board says "no" to every query. You conclude that if the grain is not on 63/64ths of the board it is unlikely to be on the board at all. Later, you happen to walk in to the room with the chessboard, only to discover that the man in the room had drawn a weird grid that mapped zones 1 through 63 to different parts of square A1, and mapped 64 to the entire rest of the board.