That's interesting. It demonstrates that regenerative braking really works. The energy you expend going uphill, you mostly get back going downhill. The energy you expend speeding up, you mostly get back slowing down.
His tests were a round trip, so start and end altitude are the same.
And he kept a fixed speed on a freeway, so there wasn't much acceleration energy expenditure or energy loss into friction brakes.
You don't get drag or rolling resistance back, so that apparently dominates.
Those don't vary too much with load.
Looks like rolling resistance decreases with diameter [1]. So, is it from the increased drag from higher stance? Would lowering the car the same work better?
Here he talks about towing, and he demonstrates loading the truck to max capacity makes nearly no difference: https://youtu.be/UmKf8smvGsA
He also covered an attempted Cannonball run where they stuffed two extra battery packs into a Rivian R1T: https://youtu.be/yfgkh4Fgw98
Real differences makers are smaller wheels and aerodynamics