What the adjustment section actually is
When a carrier applies a retroactive change, it doesn't edit history — it appends. A tier correction shows up as a pair: a credit reversing what was billed, and a rebill at the correct amount. One real-world change, two lines. A change corrected twice (it happens) is four lines. On a mid-size group with normal churn, the adjustment section runs dozens of lines a month, and after open enrollment it can be longer than the current-charges section.
| Line | Member | Description | Amount |
|---|---|---|---|
| ADJ | Member A | Credit — July, EE+SP | −$687.59 |
| ADJ | Member A | Rebill — July, EE | $320.09 |
| ADJ | Member B | Credit — July, EE | −$320.09 |
| ADJ | Member B | Rebill — July, EE | $320.09 |
| ADJ | Member C | Credit — June, FAM | −$864.61 |
| Adjustment subtotal | −$1,232.11 | ||
Three members, five lines, three completely different situations:
- Member A is a real correction — the pair nets to −$367.50, a genuine tier change credit. Verify the new tier is right and the months covered are complete.
- Member B is pure noise — a system rebill artifact. Same amount out, same amount in, net zero, no information content. These pairs exist because carrier systems reissue lines during internal reprocessing.
- Member C is a credit with no rebill — a retroactive termination. Is one month of credit the right amount? If the term was effective in May, it isn't — and this line is the only clue you'll ever get.
The trap: netting instead of reading
The rational shortcut — sum the section, sanity-check the net — fails in both directions at once. Zero-net pairs inflate the apparent activity, training analysts that the section is mostly meaningless. Meanwhile the net figure conceals composition: −$1,232.11 "in your favor" reads as good news, so nobody asks whether it should have been −$2,000.
There's a subtle matching rule buried here, too: a pair with the same member, plan, and month but different amounts is not noise — it's a rate or tier change and deserves scrutiny. Only exact-offset pairs are safely discardable. Get that filter backwards and you either drown in false flags or silently delete real corrections. This exact filtering problem — which offsetting pairs are removable and which represent genuine changes — is one of the trickiest pieces of logic in any billing reconciliation, human or automated.
Reading the section like an auditor
- Discard exact-offset pairs first (same member, plan, period, equal and opposite amounts). Typically eliminates a third to half of the lines.
- Match every surviving credit to its cause — a term, a tier change, a rate correction in your census data. A credit you can't explain is a carrier-side change you weren't told about.
- Check completeness, not just correctness. For each cause, are all affected months present? A May term with one month of credit is a partial fix wearing a full fix's clothes — the three-amount problem from the retro-window guide.
- Look for what's absent. Cross the census's recent changes against the adjustment section. Every change without a corresponding adjustment is an open item.
That procedure is correct, complete — and roughly a full day per month by hand for a mid-size group. It's the single most automatable task in billing operations, because every step is a mechanical match between two datasets you already have.