Response to Intervention, when done correctly, begins by addressing all students receiving excellent instruction in the classroom (Level 1). Even with excellent instruction, there are a number of students who may need either more time or more intensity of instruction to achieve grade level.
That is what the level 2, small group, interventions are for-- additional time and attention. I do level 2 math interventions, but my school does not use an artificial 20% cut-off. I only work with students who are not progressing at level 1, which in our school, is probably 5% of the population. In my experience, the level 2 kids often are ones who have behavior that keeps them from focusing their attention in large groups, have low intelligence (not PC to point out- but some kids just take a LOT of repetition to learn), or will eventually be diagnosed with a learning disability. If level 2 interventions don't work, then kids often are tested for learning disabilities and go on to level 3 interventions, one-on-one. [Level 3 is typically going to be classification for special education, as I understand it (but please correct me if I'm wrong) - cj]
It is actually a fantastic model that can save districts a lot of money by preventing kids from being diagnosed with disabilities that they do not have. The problem comes in when you don't have excellent level 1 programs and/or you arbitrarily decide that a certain percentage of students needs to be at level 2 regardless of whether or not they could learn in a whole group situation with excellent instruction.
It is too bad that RTI done poorly is probably going to doom this as another failed waste of money, when done well it would be exactly the opposite and really does help kids who are struggling for legitimate reasons.