I think the root cause of all this is the desire for a super-simple, one-number "score" for everything - not just EVs. Folks want - nay, DEMAND - simplification of quantification. Too bad - because some things don't yield to the oversimplificaiton. This is even true of regular gasoline-fuled, ICE-powered vehicles. One number does not tell the whole story. The payloads, driving styles, fuel details, tire types, etc, etc. all are factors. We can neglect some things and get to a "sweet spot" of "minimally complex" description of "reasonable accuracy" - but a description that meets the demanded accuracy is almost always more than just one simple number. Because it's NOT simple. And the average person really needs to come to grips with that, or remain uninformed. Those are really the two choices. Even the federal government can't simplify something beyond it's minimum level of complexity, and get a correct answer...
Bert is on the right track with postulating a minimum set of numbers. I would rather see the "extra" numbers on a car window sticker myself. The "old" way of two numbers - city and highway MPG - at least paid some service to this notion that one number does NOT give enough granularity.
An interesting way of looking at it. The driving cycle has always been the key. Before the latest changes in the cycle, one could look at the EPA figures and estimate, in general, that they would do better. Only a truly atrocious driving cycle wouldn't beat the EPA figures. Their cycle was pretty harsh, rarely getting up to highway speeds, and even the highway cycle featured a lot of city (start/stop) driving. With the latest changes, however, about the only way to best the highway figure on new vehicles is to get on the highway, as soon as the vehicle shifts into highest gear, and hold the vehicle at that speed, steady-state, with few to no stops. The latest figures were meant to be more realistic, but I think did more damage than good.
Perhaps this is like all the ideas for creating a new calendar: they all have shortcomings.
The problem with stating miles per KWh or MPG is that the number depends entirely on a driving cycle, so it ends up not being very descriptive. And therefore, not showing up the wasteful vehicles for what they really are.
My preference would be, HP or KW for speed x, and HP or KW for speed y, and then a measure of acceleration efficiency.
For speeds, make one a city speed, perhaps some value between 30 and 50 mph, and the other a freeway speed, perhaps 80 mph. This tells you a lot about the tire rolling resistance, drivetrain efficiency, and aerodynamics.
The third number, to appropriately penalize overweight vehicles, might be joules to accelerate from 0 to 60 mph, or similar.
On the other hand, the advantage of any start/stop feature, or regenerative braking, would not be apparent. Reason being, any advantages of this type of feature depends ENTIRELY on an assumed city driving cycle. My niece, for example, was "disappointed" with the efficiency of her new diesel car, mainly because she does a lot of freeway driving. It came as a surprise! Had she been given the three numbers I suggest above, the surprise would have been lessened considerably.
As to the MPGe figure, it's mostly bogus. Why? Because the specific energy of gasoline is about 33.1 KWh/gallon, however that's the chemical energy content. Depending on what type of power generating system you have where you recharge the EV, the actual comparison with internal combution engines may be far less than that number implies. It could even be less good than a gasolene car, if the electric generating plants use coal, or about the same if natural gas (going by weight and not volume).