The problem is that there would have to be many tests: Low speed, high speed, sedate, aggressive, each with heating, a/c variants, tire size varients, a further set of topographical variants, and road surface variants. You'd end up with 50 to 100 or more tests and there would still be people who complained that they didn't achieve the results. It would probably cost more than the car design to implement and by the time they were all finished the model year would be over.
It's very unfortunate that they used mpg (l/100km) as the number to report rather than an index because it leads people to think that the number is what they should get rather than that the number is just a basis for comparison with other cars. Had they used an index it would be obvious to everyone that a car with a 100 index would get about twice as much as a car with a 50 index.
Here in Texas I get this:
2004 Prius MPG from the logbook. (Complete years only):
2003-2004 -- 50.8 mpg 17,628 miles
2005 -- 52.6 mpg 14,688 miles
2006 -- 56.3 mpg 16,174 miles
2007 -- 57.3 mpg 18,384 miles
2008 -- 59.9 mpg 21,755 miles
2009 -- 61.4 mpg 16,177 miles
2010 -- 65.2 mpg 12,134 miles
2011 -- 66.9 mpg 11,272 miles
Now not everybody gets this kind of MPG, but I could complain the test numbers are too low.