The interoperator and repeatability of Threshold Audiometry is ~5+db. On a log scale, remember.
Reproducibility can be dependent on features that testing of the Neuropsychiatric Testing panels better elicit.
The data is a cloud , not a point or line, so one test is going to give only an incomplete picture.
Speech in noise is even more variable, and often parameters are not consistent.
So , I got several tests and charted them.
It costs the audiology guys $500-1000 to get somebody in the door (citation somewhere). They have capital, personnel, and time invested. Nobody agrees on a fair margin, so the market makes one up. I'm not a looky-loo, but we ain't playing the same game. It's somewhere in the space occupied by the makeup counter at Sephora, a mattress store and a new car lot. Pretends to be the Stereo section of Best Buy
And still this is only part of the map.
Give me a bid for the work up front, and it's a free market. Give me two pages of 2-pica mouse print, and try and burn me on the way out, and it's a rumble in the jungle.