The only true measure of my mpg is when I fill up at the pump (same one every time if possible, and on the same level forecourt) and calculate my mileage for that complete tankful. What the ScanGauge shows me is all relative. The instantaneous read out tells me how I am doing, but is not a definitive figure. The same goes for my daily or trip mpg. It shows me whether I am better or worse than yesterday. It is not a million miles out, but the calculated figure is better (and the average over several tankfuls better still.
The accuracy of the ScanGauge improves with every tankful (calibration adjusted each time you fill up to steadily improve the error), but unfortunately I botched the setup at the beginning of September, and had to do a remedial setup on half a tank, to get reasonable numbers again. Because I only have to fill my tank twice a month it takes a spell before the numbers are accurate again.
The reason I botched things is I was trying to rush on a busy forecourt and stuck too many gallons into the equation! I now fill up, collect my receipt with the numbers I need, then pull into the nearest bay (supermarket car park) to do my sums and enter my data.
So all in all I think the accuracy of the data was the biggest factor, though I have now added a couple more psi to the tyres (they don't get so hot in the winter) and I have developed a modified driving style for dual carriageways, based on controlling my Manifold Absolute Pressure. Every little thing helps.
Regarding the average over several tanks. Fuelly.com, where I record my top ups, gives the average over the last 10 tankfuls, and my figure is currently 60.0 mpg (my Spritmonitor numbers, shown beside my avatar, are for my time of owning the vehicle).