By the end of a normal month, the denial queue has been worked. Appeals went out on time, the aging report looks defensible, and everyone can say what happened to the claims the team touched. Almost nobody owns the other question: did everything you performed actually become a claim?
Denials are loud. They arrive with codes, land in queues, and get counted in somebody’s report. The failures that cost more are quiet ones: the completed test that never became a charge, the charge that never became a claim, the claim that bounced at the clearinghouse and was never sent again. None of that shows up in denial reporting, because denial reporting starts with claims that exist. Finding it is reconciliation work, and for labs and specialty practices it’s often worth more than the next round of appeals.
A denial at least announces itself
A denied claim is at least a work item. It has a code, a dollar amount, and a place in a queue. You can dislike your denial rate and still respect that it’s measurable. The unbilled service produces none of that: no remit to read, no reason code to argue with, no aging line for a manager to question. Billing systems report on what entered them, and this never did.
In a lab, the miss looks like an accession with a completed test, a final report, and an insurance-bill flag, and no charge behind any of it. In a clinic, it looks like an administered drug that made the treatment record but not the claim, or units documented at one number and billed at a lower one. Ask a billing team how often that happens and the honest answer is usually that nobody knows. Not because anyone is careless, but because every report they run starts downstream of the miss.
The gaps open at the handoffs
The leak rarely happens inside one system. It happens between them. An interface between the LIS and the billing platform drops a record on a Tuesday and no one is comparing counts. An order is canceled and re-accessioned, and only one of the two crosses over. A charge gets held for a question that never gets answered. A client-bill designation is wrong, so a billable test is quietly treated as someone else’s revenue.
Then there’s the rejection trap, which deserves its own paragraph. A claim that fails clearinghouse or payer front-end edits never reaches adjudication. The payer never denies it, because as far as the payer is concerned it never really arrived. So which report does it land on? In plenty of operations, none. The claim isn’t denied, isn’t pending, isn’t paid. It’s simply absent, and the acknowledgment file that said so scrolled past unread.
Twelve months, then it isn’t revenue anymore
What turns these gaps from annoying into expensive is the clock. Medicare fee-for-service claims must be filed within one calendar year of the date of service. After that the claim is denied, and a timely-filing denial carries no appeal rights. Commercial windows are contractual and often shorter. So an unbilled service isn’t a stable asset waiting to be found. It’s inventory with an expiration date.
That’s the real difference between leakage and loss. For a while, the missed charge is money you haven’t collected yet. Once the window closes it becomes money you can count but not collect, and the only value left in it is learning which handoff failed. A reconciliation you run annually will always find some of both; run monthly, it finds the misses while they’re still billable.
A missed claim is worth more than it used to be
The other thing that’s changed is what a single miss costs. Medicare’s lab spending keeps concentrating: genetic tests took 43 percent of Part B lab-test dollars in 2024 on roughly five percent of the test volume. We walked through that curve in the molecular billing note, but the reconciliation consequence stands on its own. When the money rides on a small share of claims, misses stop averaging out.
A dropped routine panel was noise; a dropped molecular claim is a real number, and an interface defect that eats a few of them a month is a budget line. Oncology practices know the same arithmetic from the drug side, where the revenue sits in units and a handful of high-cost lines rather than in visit volume. The higher the value per claim, the less comfort there is in “we probably catch most of it.”
The same match runs in both directions
There’s one more reason to run the match deliberately: it doesn’t only find underbilling. Set the performed record next to the billed claim often enough and you’ll eventually find the reverse, four units billed where the source record supports three. That finding comes with obligations. Federal rules give a provider 60 days to report and return an overpayment after identifying it, with a lookback period of six years.
Some teams hear that and conclude it’s safer not to look. It isn’t. The rule’s knowledge standard is borrowed from the False Claims Act, and it reaches deliberate ignorance as well as actual knowledge, so “we never checked” is not the shelter it sounds like. The practical move is to decide, before you run the first match, who reviews what it finds, how corrections get made, and how a refund would be handled. Reconciliation done with a plan is a compliance asset. Done casually, it’s a pile of findings nobody is prepared to own.
Six counts worth running every month
None of this requires new software. It requires counts, run on a schedule, with a named owner for every difference. A workable monthly pass:
- Insurance-billable services performed versus charges entered, by location and by month.
- Charges entered versus claims submitted, with held or suspended charges listed by age.
- Claims submitted versus clearinghouse acknowledgments, with an owner for every rejection.
- Accepted claims with no remittance activity after a defined number of days.
- Billed units versus source-record units on a sample of high-dollar claims.
- Client-bill and insurance-bill designations, re-checked against what actually happened.
What the queue can’t tell you
The encouraging part is that nothing here depends on new data. The order file and the charge export already exist. So do the claim file, the acknowledgment, and the remittance. What usually doesn’t exist is the connective tissue: identifiers that survive the handoffs, so the same service can be followed from accession to charge to claim number to payment. Preserve those, and the matching is mechanical. Lose them, and every reconciliation becomes archaeology.
Teams that build the habit stop discovering leakage in year-end cleanups and start catching it while the filing window is still open. Some run the counts themselves. Some borrow outside capacity for the first historical pass and keep the monthly rhythm in-house. Either way, the point is the same. A denial queue tells you how well you handled the claims you had. Only the counts tell you whether you had all the claims.
Important note
This article is for general educational purposes only and is not legal, billing, coding, coverage, or reimbursement advice. Coverage, documentation, appeal, and payment requirements vary by payer, plan, jurisdiction, contract terms, and claim facts.