Of course, efficiency will vary, but standardized testing is an essential benchmark for comparison. So the Gravity pulling a trailer up a mountain in the dead of winter will get worse mileage than one on a flat road driven by grandma, but it helps to have some guidelines. My issue with range numbers is there is no real standardization, and depending on where the car comes from determines what method is used. China uses one that is exceptionally optimistic compared to most EPA-published numbers, and Tesla uses a different process for their numbers, resulting in overly generous numbers. I would be happy if the industry picked one method and published numbers based on that. You buy an ICE car and can be reasonably confident in getting close to the published figures.
If you buy a Tesla you can expect worse range as Elon miles aren't going to happen, my Rivian gives me about the stated miles for given tire size and BMW seems to sandbag their numbers and I get better than stated range out of my i4 M50.