I'm using data from TeslaMate, would this not be as accurate since this would be to be all data coming from the car itself?
View attachment 690058
You'd have to read in the docs about how the "used" energy is calculated, and how the "added" was calculated.
Since it appears to have access only to the API, for the added energy, the app would need to know the scaling of 0.955 to apply to the "kWh added" value. It all depends on how the code is written.
In order to calculate efficiency, TeslaMate would also need information from the wall (I assume it looks at voltage and current over time during a charging session, which would be fine, integral of V*A over time is energy from the wall (at the vehicle plug and neglects other wiring losses)).
Anyway, it should take the kWh added from the API, multiply it by 0.955, and divide by the wall energy from the integral, and that would be approximately correct. That's just the way it works (can verify with SMT).
"kWh added" is potentially different than "kWh used" - but it depends on their definitions. (Clearly is in this case. There's kWh used from the wall, and kWh used from the battery, two totally different quantities.)
Nothing wrong with Teslamate, as far as I can tell - looks like a good open-source tool. But the API values for kWh added do not match the energy added to the battery. (Again, the added energy is just rated miles added multiplied by the vehicle charging constant.)
Max possible efficiency is around 88-89%, per the EPA docs (which uses max charge rate for the vehicle based on recent info that came to light in some of the documents). And this will align with observations too.
Your max values coming in at 93% or so (apparently) are consistent with this, too. Make the 4.5% adjustment, making those 88.5%, and then everything lines up.