Commonly used section of no traffic flat road (I'm too urban unfortunately) ideally 10kms to 10 miles long - the length doesn't matter much so shorter could be used as well - just long enough to do sereral sets of P&Gs.
Do your best P&G technique - reset the current mpg on the SG when you are at the end of a pulse to start the test - then observe what the current average is to end of the last pulse cycle where your speed was the same as the starting pulse. Note both that MPG and the average MPH from one or more consistent 2 way runs.
Then use those average speeds and compare to best steady speed by foot control or cruise control. Personally, I think cruise is way better than me on the flats - it's using decimal point level info without any SG lag and using a drive by wire gas pedal. Anyway, see what the MPGs are at those average speeds and that would a good way to find out what the technique is worth for a given vehicle at a given average speed.
----- ----- ----- ----- -----
I think at 2 tons and with 5th being fairly lazy and usable down to 31mph (shift in requires 39mph - 2000rpm @ 60mph) the pulsing wastes energy because the engine can match the load at low powers easier than accelerating it - that or my pulse technique is all wrong. Either way though, it would certainly be nice to know.
I've found that vehicles with throttle-by-wire can more-effectively modulate the throttle for FE on level ground with cruise control than my big foot can. Several runs of the same route with both techniques have proven a significant difference in the final FE trip tally.
On the traditional cable-throttle cc's, a limiter can help with hills. I've installed a governor of sorts on the cruise control that limits the throttle activation to about 20% TPS -- you can scrub-off speed on hills, drive with load, and not inadvertently try to keep up with traffic. I don't use it unless I'm exhausted and prefer not to concentrate on the SG feedback.