Picture an owner who checks one number every morning before the office opens: deposits, divided by days. Done long enough, you can feel the answer before the laptop finishes waking up. Then, over a couple of months, that number started falling and kept falling. By July, it was $44,000.
Everything else said this should have been the best stretch of the year. The schedule was full. And the report that measures how fast the practice turns work into money, the cash conversion number, or days in AR, sat where it always sat. Steady.
It had said the same thing all year, while a quarter of the daily cash disappeared.
One of those two numbers was lying. It wasn’t the bank.
Why does days in AR look fine when cash is down?
Because the average only counts claims that already paid, and it counts every payer once. Stalled claims leave the sample, big slow payers weigh the same as small fast ones, and the number holds steady while deposits fall.
The average only counts the claims that finished
Her 23-day average was built entirely on claims that had already paid. A claim that paid in 12 days made it into the math. A claim sitting unpaid at day 55 didn’t exist yet, as far as that average was concerned. Neither did the claims that would never pay at all. Everything slow, stuck, or dying was removed from the sample before the calculation started.
Now follow that through. When an insurer stalls on her, the speed number doesn’t get worse. It gets better. The stalled claims leave the sample, the ones still paying are the easy ones, and the worse things actually get, the better the number looks. Her cash fell for a full quarter and the number never moved.
And it treats every payer the same
The second flaw was quieter. Ten insurers, one average, each counted once, as if they mattered equally. Her small payers settled fast. The two that carried the bulk of the dollars took far longer, and the simple average split the difference and called the practice fast. Weight the same number by dollars, the way you’d weight anything you actually cared about, and it moved by weeks.
The bulk of the money was sitting with the slowest insurers, and the average never mentioned it.
Every payer has a pattern. The break is the news.
Each insurer has a normal. One had paid her like clockwork for a year. Another ran slower and always had, and that was fine, because it was slow the same way every time. The number matters less than the drift. So when the clockwork insurer showed up weeks late, that was the news. Nothing flagged it. No report on the system compared an insurer’s present to its own past.
For weeks, the only alarm in the building was one morning division.
Three ways a payer breaks pattern
They slow down. Payment speed drifts outside their own range. That’s the visible break, if anyone is comparing this month to their history.
They pay light. Remits, the payment notices insurers send back on each claim, arrive right on schedule and the dollars come up short, or come in at zero. A speed report clears this insurer completely. They’re “paying” on time. They’re just keeping the money.
They go quiet. Fewer payments arrive at all. This is the break a paid-claims average can never see, because every payment that does arrive looks normal. There are simply fewer of them. Catching it takes a different question: given this insurer’s own history, how much money should have arrived this week, and how much did.
Each break hides from the detector built for the other two. The quiet version, and what a full elimination sequence looks like when cash drops, gets its own treatment. Watch speed alone and you’ll clear the insurer mailing zero-dollar notices on schedule. Watch speed and dollars-per-claim and you’ll still miss the one that quietly stopped ruling on claims, because a stalled claim never shows up in payment data at all.
The average also starts late
There is a second thing this number does not cover, and it sits before any payer touches the claim.
Days in AR is receivables divided by average daily charges. Receivables are what you have billed. So the days between the visit and the claim going out are not late in this metric. They are absent from it. A note sitting unsigned for two weeks, a chargeslip nobody created, a claim complete and never released: none of that is slow AR, because none of it is AR yet.
HFMA publishes the benchmarking standard for revenue cycle metrics, and it already has keys for this. Total charge lag days counts the days between the date of service and the date each charge posts. Late charges counts anything posted more than three days after the service date, so the standard treats day four as late. Days in discharged not final billed measures care delivered and not yet billed.
Days in final billed not submitted to payer measures claims complete and still sitting in the scrubber, the tool that checks claims for errors before they go out.
Four numbers, published, standardized, and applicable to physician organizations as well as hospitals. Almost no independent practice runs any of them.
That is the honest version of the problem. The definitions exist. What is missing is anybody producing them, because the standard reports do not assemble that data on their own.
You can approximate the first one this week. Pull a sample of visits, record the date of service and the date the charge posted, and average the gap. A day or two means your front end is clean and your payer variance is the whole story. Two or three weeks means your average was late before a single payer saw the claim.
How operations-run businesses see it
Any business that extends credit watches everyone who owes it money against their own track record. A lender doesn’t average its borrowers into one number and call the portfolio healthy. The moment something drifts from its own history, a flag goes up with the variance attached. Healthcare ended up with one flat average, built only on the claims that already paid, and everyone treats it like the truth.
Nobody chose that. It’s just what the billing system spits out, and the habit stuck.
What this means for you
Run that check on your own numbers. Pull last month’s payments, group them by insurer, and set each one’s days-to-pay against its own prior twelve months. Then take your overall conversion number and weight it by dollars instead of counting every insurer once. If either exercise moves the number more than a few days, your dashboard has been telling you a story. Just not the whole one.
The same by-insurer cut explains why a percent-of-Medicare figure can drop with no contract change.
And if the payments arriving aren’t even recorded yet, you have a different backlog to measure first. Groups running more than one location have one more blind spot on top: the blend itself.
You’ll see who’s drifting and what it’s costing before your average ever moves. Grab 30 minutes with us. Prep nothing.
Questions people ask
Why should I track days in AR by payer instead of overall?
Because an unweighted average hides a single payer breaking pattern. One insurer slowing from twelve days to thirty will barely move a blended figure while it moves your bank balance, and the average recovers on paper as other payers pay normally.
What does it mean when one payer’s days in AR suddenly rises?
Something changed at that payer, and it will apply to every claim you send them from now on. Catching it in the age distribution is weeks earlier than catching it in cash, which is the difference between a conversation and a recovery effort.
Does days in AR include unbilled work?
No. HFMA’s standard excludes any account not yet billed to the payer or patient. Everything between the visit and the claim going out, including unsigned notes and uncreated charges, sits outside the number entirely.
What upstream metrics should I run alongside days in AR?
HFMA publishes four that most practices never produce: total charge lag days, late charges posted more than three days after service, days in discharged not final billed, and days in final billed not submitted to payer.