I forget who it was, but I think it was Ghrist that mentioned in his TDA book that expectations were to be extremely tempered before trying to apply TDA to a given problem domain.