Part of this is Speedometer calibration.

Fuel consumption is calculated in lab conditions, using an externally arbitrated distance and the fuel consumed is also measured under lab conditions.

Speedos will never be 100% accurate so manufacturers make sure that the error is conservative. Someone being charged with speeding when their speedo was exactly on the speed limit would be bad press. Most new cars are travelling at about 90-95km/h in 100km/h zones.

Petrol expands and contracts as it changes temperature. This is a good reason to only fill your tank on cold days, in the morning. (Although the cost difference is not worth considering, the difference in volume of fuel is measurable.

Finally, the fuel pump flow that is measured in your EFI system and the diameter of your tyres are also affected by temperature.

These three factors affect what your fuel gauge tells you. You may be travelling 100km and using 5.6l, but your ECU may be calculating and displaying a higher result.

Is that really surprising? The ratings are in ideal conditions when people do not drive like that.

I remember an offhand comment Jeremy Clarkson made on Top Gear where you look at the city driving number, the worst L/100K, as that is what you should expect when driving on a highway when it is suppose to be most efficient.

I get the claimed figures on both my car and van so it's not like hitting them is impossible, but it is highly dependent on how and where you drive. Many people drive with the smoothness and ability to look ahead of a myopic orangutan (e.g. accelerate hard, immediately jump to brakes, back to acceleration, repeat) but even with a bit of care there's only so much you can do if your driving involves a lot of short trips and stop start traffic.