Therein lies the problem. What should it be based on? How many students pass? Standardized test scores? How about teachers that are good but get a job at a school whose students are generally poorer-performing vs teachers that aren't as good but work at a school with a higher caliber of students?
Oh - I completely understand that it is a difficult question. Many of the evaluation options thrown out by people involve more standardize testing (which will favor students, and in turn teachers in better socioeconomic classes and with less English language learners).
I'm not an educator myself, but I think an honest review for tenure purposes would have to consist of a comparison of student results from year-to-year, not just comparisons between students district/state/nation-wide. Then you could possibly see how a teacher has had a negative or positive (or net-zero) effect on student progress in their subject area.
But again, this is my own layman's guess.