Yeah, I do agree.
I'm using only API but not as an "absolute" measurement, but like a benchmark.
That being said, using the same kit for a period of time and compare each result to see the changes.
Heck, for API result, I can't in my life tell the diff between 5-10-20, or 40-80...