It keeps all the figures, and you could compare anything to anything else yes, but why would you? I don't think you're using the figures correctly, or even trying to.
What figures aren't I using correctly? If I take a 2011 A3 2.0TDI (non bluemotion or whatever Audi called it) that has got an official mpg combined of 55mpg and was tested with the NEDL method, i'd easily achieve 50mpg with everyday driving, the test gives results within 10% of reality. I then take a 2013 model of the 8V platform, it has stop start included, and is subject to the same test as the 2011 model. It achieves 67mpg combined on the test, but gives 51mpg with everyday driving. Audi boast of 20% gains (without disclaimers), you are 24% from reality. The dealership made a big point of how much fuel you will be saving - will you be disappointed by reality? It really is a poor test. If a GCSE Science student devised such a poorly thought out comparison test they would likely fail.
Imagine buying a 30cm ruler that was only 22.8cm long, or your 1L bottle of Fairy claimed to wash 25% more dishes than the old formulation, but only because it classes a 75% cleaned dish as being called clean.