At OSH in the late 70's, era of the Vari-eze, we had dinner with the late RT Jones, America's premier aerodynamicist.  I asked him, "Are the Vari-ezes fast with small HP because they are canards?"  "No," he said, "It is because they are slick."   Experience has shown that our tractor arrangement airplanes can be as efficient as the best exotic pushers and canards and such due to secondary effects such as prop/fuselage effects and such.  ATtention to detail in drag reduction is the key.  Fanaticism helps.   Let's set aside efficiency defined as fuel burned per ton mile (huge supertankers win hands down) or airplane fuel burned per seat mile (747 and A380's win) and focus on simple airplane miles per gallon.  For that you need the minimum  airplane drag and maximum efficiency out of the engine.    Total drag (induced plus skin friction parasitic plus cooling drag) can be measured by a full power blast at sea level.  We use sea level and full power because at this corner of the envelope you get the most accurate data in an area where accuracy counts for a lot.  Note that TAS calibration is a must if you want truthful numbers, or you must make four way GPS runs under absolutely stable conditions.   Caution: small errors make big changes in drag area estimates.  You can do it at altitude, but you have to make corrections for density altitude, air density, power corrections, and so forth.   Do-able, but errors multiply.   Here is how it works using the attached chart.  Multiply your full throttle sea level power by .85 for the propeller efficiency to get the net horsepower (along the top of the chart attached), then go down the diagonal line until you intersect your top speed in  MPH (along the bottom).  At that intersection go to the left or right to get the total equivalent "flat plate drag area."  Lower is better.     Some reference figures: stock Lancair IV as originally designed and tested (non- pressurized, 2900 lb gross wt, turbocharged with intercoolers) was measured at 2.12 square feet.  The cleanest 235/320 class, about 1.7 (from memory, may be wrong).  ES/Cirrus/Columbia about 3, while C210/Bonanza class about 4.  C172/182 about 6 (very rough).    I think the Legacy  works out to about 1.8-2.0  but reports on performance vary widely.  As I said, little errors in power estimation, speed calibration and such make big errors in the flat plate area number.  In my own case, non-pressurized, no turbos (and thus no intercoolers with their associated cooling drag), extreme attention to reduced cooling drag, cowl flaps closed, fresh paint with no nicks or  dust, no bugs, gap seals, the works, I got down to maybe 1.85.  Age, nicks on leading edge, foam strips that have worked loose here and there and such and it has creeped up with age to maybe 1.9, possibly a bit more.      GOOD data takes a lot of effort to get.  Carefully calibrated instruments, three test runs with careful measurement and corrections (don't forget to correct power if your manifold pressure is higher than normal due ram pressure during that high speed sea level blast) and you should get close.  It's all good fun.  Give it a try.