What the provider is actually contesting

The determination separated three questions and numbered them. This takes what came back and sorts each argument to the section it contests, so a reviewer opens the appeal already knowing which questions are in issue — and which were never answered at all.

The letter it answers

This is a lexicon, not a model. Each argument is scored against the three sections by the terms it shares with them, weighted so a word that appears under every question counts for nothing. Measured against appeals it had not been tuned on, it puts about three arguments in four on the right section. That is why every argument below is shown whichever way it routed, why the evidence behind each route is on screen, and why an argument it cannot place says so instead of picking the nearest one. A misroute costs a reviewer a glance, not a decision.

The appeal

Pasted for analysis

What this does not do

It does not decide the appeal, overturn a finding or record a disposition.
It does not move an amount. The exposure figures are hypotheticals, computed from the letter's own dependency between sections.
It does not judge whether an argument is correct — only which question it is answering, and whether it brought what that question asked for.
It does not hide an argument it could not place.