We were reviewing payment files with a billing lead recently when she flagged the strangest payer in the stack. Every metric cleared them. Payments arrived on schedule. Claims got ruled on inside the window. Days-to-pay sat comfortably in range. And the practice had collected almost nothing from them in months.
The payer was paying on time. The payments were for zero.
On time is not the same as paid
Every insurer payment arrives with a notice stating what was paid, what was denied, and what the patient owes. The industry calls these remits. A zero-dollar remit is a real notice, delivered on schedule, ruling on the claim, containing no money. To a speed report, it’s indistinguishable from a payment. The claim adjudicated. The clock stopped. The metric cleared. The dollars never came.
That’s what makes this the quietest break a payer can have. A payer that slows down shows up in days-to-pay. A payer that goes silent shows up as missing volume, if anyone’s counting. A payer that sends zero-dollar notices on schedule hides inside your best-looking numbers, because every instrument you own was built to measure time, and they’re not taking time. They’re just keeping the money.
Why is the insurance company taking money back?
Recoupments: the payer decides it overpaid earlier claims and subtracts the difference from new payments, inside the same check. Deposits shrink with no line item, and only the remit’s negative rows show it.
The second half of the trick
Alongside the zeros, money started moving the other way. Recoupments: the payer reaching back into claims it already paid, deciding it overpaid, and taking the dollars back out of new payments. On the reports, the current claims looked ruled and clean while the arriving checks kept shrinking, because they were arriving pre-shrunk.
The bank can’t warn you about this one, and it’s worth understanding why. Recoupments come out inside the same payment that settles new claims, so the deposit that lands is simply smaller, one number, no line item announcing that money left on the way in. The only place the subtraction is visible is inside the payment notice itself, as negative rows. If nobody reads the notices line by line, which is exactly what happens under a posting backlog, the takebacks are invisible by construction, which is exactly what makes them the preferred quiet exit for a payer that has decided to stop paying.
And there was a procedural catch buried in the arrangement. The zero rulings stood unless the practice rejected them, one by one. Nobody knew that was the job. So the zeros aged politely into permanence while everyone watched dashboards that said this payer was fine.
Zeros like these usually have a mechanical origin. A coverage rule changes upstream, a documentation requirement gets added, and suddenly every claim of one type rules at zero, week after week, until somebody notices and rejects the batch. The pattern repeats precisely because it’s automated. Nothing about it will stop on its own.
What the count showed
When we finally counted it the only way that works, money in against money owed, this payer was sitting on $196,000 while clearing every speed metric on the stack. Not slow. Not silent. Present, punctual, and paying nothing.
The instrument that catches it
The fix is a yield line next to the speed line. For every insurer, every week: cents actually paid per dollar adjudicated, against that insurer’s own history. Speed tells you they answered. Yield tells you what the answer was worth. A payer whose yield drops toward zero flags immediately, however punctual the paperwork, and recoupments show up as what they are, negative yield, instead of vanishing into netted deposits.
Underneath the line sits a queue: every zero-dollar ruling, oldest first, each one either rejected and refiled or accepted with a reason written down. The zeros stopped being a status and became work with names on it.
Working the queue has one rule that saves most of the labor: batch by claim type. Zeros run in families, because the rule change that produced them applies to a category, so the correction that revives one usually revives its siblings. Fixing the family beats fixing the claim, and a rejection here is a formal challenge with the correction attached, filed back through the same channel the ruling came in on. One afternoon of family-level work typically clears what months of claim-by-claim effort never touched.
What turned up, what we built, what stays
Turned up: $196,000 with a payer that cleared every metric by paying nothing on time, plus recoupments quietly shrinking every check that did arrive.
Built: the zero-pay and recoupment monitor. Yield by payer against its own baseline, recoupments broken out instead of netted, and the rejection queue worked oldest first.
Stays: any payer whose yield breaks from its own pattern, or who sends zero-dollar notices twice in a row on the same claim type, flags itself that week. Punctuality stopped counting as payment.
Your check
Pick your five biggest insurers. For each one, take last month’s dollars actually paid and divide by dollars adjudicated. Anything drifting toward zero is hiding in your speed report right now, looking like your most reliable payer.
We’ll run yield by payer on your own data and show you who’s been paying you in full, zero dollars at a time. Grab 30 minutes with us. Prep nothing.