Are we interested in MPG or HP/torque improvements here?
If HP/Torque, it might be my memory failing or just a long week, but under WOT (like in dyno testing) won't the stock ECU be modulating fuel to hit its A/F target, so if you are burning more efficiently and therefore consuming more oxygen (no change in airflow right?) it throttles back the fuel, thereby reducing the potential for any real HP improvement? Not saying there wouldn't be something, but it might be at data noise levels, I would think. I mean, we're talking about a step improvement in atomization, not orders of magnitude, right?. I'm starting to doubt the benefit of asking for dyno anymore...
Then there's closed loop, where you are running off tables in the ECU. You could watch via ODB2 the STFT (short term fuel trims) pulling your LTFT into shape after changing the injectors, to then know that SOMETHING is happening. You couldn't correlate a % change to any sort of "improvement", especially since you would be swapping older injectors, possibly with reduced flow, for new full flow ones (as pointed out by
@2001LC).
So perhaps a better measurement of an efficiency (and therefore MPG) improvement would be before & after emissions testing? I would expect that 'more efficient combustion from the 12 holes vs 4' would be revealed through reduced NOx and VOCs. This would be scientific on pretty reliable equipment available all over at relatively low cost.
Thoughts? Or has it just been a really long week for me... ?