Thermodynamically, what really matters with regard to system on/off is how much heat is being drawn from the engine, i.e. how much heat is being blown into the cabin. If no heat is coming into the cabin then you are pretty much guaranteed that the heating system isn't having an impact on the engine.
As for a percentage, I think it depends on your driving cycle and how you use the heater. If you only allow it to run after the engine is fully warmed up then I doubt you could even measure the impact. An engine has to reject some heat in order to run, and all you are really doing is directing it into the cabin rather than through the radiator or out the exhaust. Trying to blast the heat while the engine is still cold is a different story, as it would prolong the distance over which the engine is burning extra fuel to bring itself up to temp. I imagine the penalty while operating in that state over a very short driving cycle could be quite steep since the engine management system will be fighting you every step of the way.
Of course, that's going to matter a whole lot less in Southern California than it does on a cold New England winter morning.
The short answer is "it depends." The only way to get a real answer would be through experimentation using the specific vehicle in question.
As we know, a warm engine produces the best FE. Unfortunately, it takes the 'Teg FOR-EV-ER (Sandlot movie reference) to warm up in cold weather. Long-story-short, I have to tough it out the best I can to get it to up to normal operating temp by not running the heater blower -- otherwise, you don't get the engine up to "waste-heat" temp and the TC comfortable to engage. On those cold nights in the 'teens when the car's sitting at the airport, it's a teeth-chattering "feats of strength" cometition to see how long I can tolerate zero-heat until the gauge is up to NT. When my fingers go numb is a usual indicator of when I need a little heat. But after about 20 minutes of driving, I get waste-heat and can get some warmth into the cabin; HOWEVER,
If I run the heater blower too fast, the temp gauge actually drops -- so it must be used in moderation. Honda built this cooling system to operate in the Mojave at WOT/full load all day long! This comes in handy in the summer when EOC'ing.
The solution? An EBH. I'm up to NT in 5 minutes, and can use the heat almost immediately. Only if they had plugs at the airport
If a Blower is using 100 watts, and the alternator is only 50% efficient, then the Blower is really using 200 watts!.
If the ICE is only 25% efficient, then the blower is using 800 watts of raw petrol in order to generate 100 watts of electricity.
Hmmm.... A blower motor uses 100 watts , period! Regardless of the wattage source (providing it is equal to or more than the blower motor). Watts = power-to-spin at some RPM. The RPMs won't go over the design value... but lower RPMs can be obtained fron a lower power source.
The (100W) blower will not "suck out" 100 watts of power if the power source ain't got it! Petrol, power company, or bicycle-powered generator! Watts cannot be made from the source! Regardless!
Regarding the power to spin the alternator...sure it takes more torque (and more throttle) to spin the loaded alternator (blower "on")...Energy out = energy in + losses.
Are we tangled up in our definitions? Hmmm....