As I understand it, the advertised and claimed fuel efficiency improvement (by oil companies) attributing to adoption of low viscosity engine oil from laboratory dyno testing conditions is over a denominator of substantially lower absolute # in useful work done...
disregarding 65-70% fuel energy losses in the forms of heat, noise, air drag, tire rolling resistance etc in typical vehicular operations ,
hence magnifying 'global fuel energy saving' in percentage terms ....
skewing away from 'truth' in real terms.
JMHO.