How's that ? You mean by obsessively careful attention to sealing with mastic etc ? Even so, the insulation on flexduct ain't that great, plus it seems to be compomised a lot by being compressed in bends and by hanging materials. I think the only way to really minimize the losses is with the encapsulated crawlspace approach - so the temperature on the outside of the ductwork is much higher - and that seems to be one of the most compelling reasons (for closed crawlspace).
As BG said, with short R-8 flex, thick FG blankets over the trunks, and careful airsealing (a PITA) duct losses (conduction and leakage) can be <=5%. Not a deal breaker.
You're empirically measuring your COP ?!? Cool-o. How ? I'd love to do that.
running geek flag up pole.....
If you know your BTU/h load versus outdoor temp, and log your HP run hours, you can estimate its BTU output per hour at different temps. If you measure its power consumption (e.g. with a home energy monitor), you can compute your COP from (kBTU/kWh)/3.414. Quite tedious, but if you are logging the data anyways, not that hard to sit down and bang out. You need to use a second heater (e.g. resistive aux) with known output to get BTU loads. Switching to aux (emergency mode) for a few selected days to get data costs money.
In practice, I do it the other way.... The module COP, BTU and kW versus outdoor temp are tabulated by the mfr, (using an ASHRAE standard test regimen). You can also check the kW in the table and for the blower while you are at it. The table does NOT include blower power in the kW or those BTUs, as far as I can tell. On a spreadsheet, you can predict HP run hours versus outdoor temp from the tabulated BTU numbers, and the daily run hour log matches the model when defrost is not active. Data is noisy, you need to throw out very windy days, etc, but the model matches.
During typical 'dry' defrost conditions, the unit runs ~18% more than the model prediction, suggesting COP is reduced by 1-1/1.18 = 15%. Under 'heavy frost' or wet snow conditions, it has to run close to 50% more, suggesting a 33% reduction in COP. Before I rejiggered the defrost control, the COP hit was closer to 40% under dry conditions! At 30°F and dry, I would have had COP=1.5 as installed. Now it is 2.1-2.2 or so.