Models compare, they do not predict
A simulation of an unbuilt building tells you which option is better. It does not tell you what the meter will say.
Energy models are routinely presented as forecasts. They are not, and treating them as such is how a project ends up defending a number rather than improving a building.
Hold every assumption constant and change one thing. The difference is trustworthy. The absolute value is not.
Occupancy, operation, weather, and the gap between specified and installed construction each introduce more variance than most of the design decisions being tested. What survives all of that is the ranking: option B used less than option A under identical assumptions, and by roughly this margin.
This is not a limitation to apologise for. Comparative modelling answers the question a design team actually has, which is what to do next. It only becomes a problem when the output is promoted into a performance guarantee somewhere between the report and the brochure.
The honest form of words is narrow and worth using: under these assumptions, this option reduced this demand by this much relative to this baseline.