While I acknowledge that students can often tell who the good, bad and average teachers are, I don't think it's feasible (or even a good idea!) to base teacher pay on a poll of students. Many students may well give a reasonable and accurate review of their teachers, but some may be less than reasonable or honest. Teachers who bribe their kids are going to fare the best often.
And so if that's out, you're back to test scores. I'm a believer in tests and data analysis! But, again, there are myriad factors which can affect not only overall scores, but also rates of improvement/decline that are out of control of the teachers. The problem is there's no shortage in students sitting in classrooms they do not want to be in. Teaching is one of the only -if not the only- professions where the job rating of professionals is based on the performance of children who often resent being there in the first place.
All I said is that anyone claiming the evaluation of teachers is *easy* is revealing their ignorance. It may be possible, but it is not easy or simple.