How much does tutoring improve SAT scores?

My friend Catherine, after reviewing various meta-analyses, says this:

Students get about 30 points on average (across reading/math), with 8 points of that going to reading & the rest to math.
I don't think anyone has ever looked at the high-end tutors (there are families in NYC spending $25K to $40K for a year of SAT prep for one child).
Now that I'm in the thick of SAT prep, I believe the meta-analyses.
I don't see how anyone can coach the reading section, which is fantastically sophisticated, much beyond the 8-point average gain researchers find.
And 20 points on math sounds right to me, too, though we're trying for more.

30 points applied to the 1600-point SAT math and reading sections is confirmed in this report.

Test-prep programs generally include three elements: a review of test content, practice on test questions, and orientation to the format of the test. In 2009, in cooperation with NACAC, Mr. Briggs reviewed three national data sets and found the average effect of commercial coaching is positive, but slight. Test-score bumps were more in the neighborhood of 30 points (on a 1,600-point scale at the time), far from what some in the industry claim. He does point out that there may be specific programs that are more effective than others, but evidence to support that is weak.

My unscientific extrapolated estimate for the 2400-scale SAT would be that tutoring provides an average 45-points bump.