Payment posting is the most under-discussed drag on the RCM P&L. Every 835 that lands is either auto-posted in seconds or manually keyed for minutes. This is the deep dive on what the 835 actually contains, what auto-posts cleanly, where the exceptions live, and the workflow that frees 1.5 to 2.5 full-time billers at 5,000 claims per month.
1. The manual-posting drag most operations ignore
Payment posting is the most under-discussed line item on the RCM P&L. Nobody wants to talk about it because it is boring. Boring is exactly why it eats so much biller time.
Every 835 that lands in your operation is either auto-posted or manually keyed. The math is not subtle: at 5,000 claims per month with a 3-minute average manual touch, that is 250 hours of biller time going to a task that has essentially zero decision-making content.
2. What the 835 actually contains
The 835 is the X12 HIPAA transaction set for electronic remittance. Every line carries the same structural elements:
- Claim identifier. Payer's internal claim number plus your billing system's claim reference.
- Service line details. CPT, DOS, billed amount, allowed amount, paid amount.
- Adjustment codes. CARC (Claim Adjustment Reason Code) explaining why the payment differs from the billed amount. RARC (Remittance Advice Remark Code) providing detail.
- Provider level adjustments. Recoupments, refunds, forwarding balances.
- Payment method and EFT trace. Links the 835 to the actual bank deposit.
Everything you need to auto-post, reconcile, and route denials is in that data. The complexity lives in the edge cases: unmatched claims, partial payments, secondary payer coordination, and unusual CARC/RARC combinations.
3. What auto-posts cleanly
The clean-post cohort, roughly 85 to 95 percent of volume in a healthy operation:
- Fully paid claims. Payer paid at or above the contracted rate. Post payment, record adjustment, close account.
- Contractual write-offs. CO-45 adjustment at the expected contract rate. Post adjustment, no dispute needed.
- Straightforward denials. CO-16 (missing info) with clear RARC, CO-197 (auth missing) with clean flag. Post the denial to the account and route the claim to the appropriate workflow.
- Coordination of benefits. CO-22 primary/secondary. Post primary payment, forward to secondary.
4. Where exceptions live
The 5 to 15 percent that requires human judgment:
- Unmatched claims. The payer's claim reference does not resolve to a claim in your PM. Usually a submission ID mismatch, needs manual investigation.
- Partial payments below the contract rate. CO-45 at less than the contracted rate is either a silent rate change, drift, or carve-out misapplication. Requires reconciliation before posting.
- Unusual CARC combinations. Denials with multiple CARCs (CO-50 plus CO-97) require reading the RARC narrative to decide the appeal path.
- Recoupments and refunds. Payer took back a previous payment. Requires reconciliation with the earlier claim and often a dispute.
- Manual EOBs. Payers still sending paper EOBs require OCR conversion before auto-posting can run.
5. A four-step posting workflow
Step 1: Ingest every 835 within 24 hours of receipt
Automated ingest, no manual download step. Every payer either pushes to your clearinghouse or delivers via SFTP; the workflow should pick up new 835s hourly.
Step 2: Auto-post the clean cohort
Match, post, adjust, close. Log the transaction. For denials that auto-post, route the underlying claim to the denial team queue by CARC.
Step 3: Route exceptions to human queue with context
Exception queue is prioritized by dollar amount, aging, and expected recoverability. Reviewer sees the 835 line, the underlying claim, and the specific reason it was routed (unmatched, partial payment, unusual CARC, recoupment).
Step 4: Reconcile deposits to posted payments daily
Sum of posted payments must reconcile to the EFT deposit total. Discrepancies get investigated same-day. Most posting errors surface here.
6. Software category map
7. Five metrics to instrument
- Auto-post rate. Percent of 835 line items posted without human touch. Target 85-95%. Under 75% means either exception routing is too aggressive or the underlying data is noisier than average.
- Exception queue depth. Number of unresolved posting exceptions. Should stay flat week over week; a rising queue is a capacity signal.
- Time to post. Hours from 835 receipt to fully posted. Target under 24 hours.
- Reconciliation variance. Difference between posted payments and EFT deposits. Should be zero every day; anything else is a posting error to investigate same-day.
- Denial routing accuracy. Percent of denied claims correctly routed to the right workflow bucket (CO-16 to correction, CO-50 to clinical appeal, etc). Target 95%+.
8. Where to start Monday morning
- Audit your current auto-post rate. Pull the last 30 days of ERAs and calculate how many line items required human touch. Anything under 80 percent is opportunity.
- Categorize the exceptions. Bucket the human-touch cases by reason (unmatched, partial payment, unusual CARC, recoupment). Top three account for most of the volume.
- Fix the top-bucket root cause. Unmatched claims are almost always a submission-ID configuration issue. Partial payments are almost always contract reconciliation. Recoupments need a specific workflow.
- Instrument the four metrics above. Report weekly for a quarter. If auto-post rate climbs and exception queue holds flat, the fix worked.
Or have us build the AI biller that runs this workflow for you.
Every workflow in this pillar is exactly what our AI Biller does. Delivered in 30 days. Lives inside your PM system. Money back if it does not outperform your current process on three metrics you pick.
See the AI Biller →Frequently asked questions
Want to talk through this for your organization specifically?
A 30-minute call. Bring your top pain points and the numbers you already track. We will tell you what a first-cycle assessment would surface, what an AI biller build would look like for your setup, and whether we are the right partner. No slides, no pitch.
Book a 30-minute consult →