How about doing a binary recursive analysis?
Ie. measure at 1MHz, at 9MHz, at 5MHz (left, right, center);
at 3 MHz and 7 MHz (new middle points);
at 2,4,6,8 MHz (next new middle points),
at 1.5, 2.5, 3.5, 4.5, 5.5, 6.5, 7.5, 8.5 MHz (if needed, see below),
etc. -- then you get a successively finer picture, and if you detect a flat curve over 5 (numerically) consecutive measurement points you could also skip the ones inbetween.