Part 1 of a three-part series
Becky Carlson, Head of RCM at Joyful Health, spends her days inside denial data across specialties, untangling why claims stall once something’s already gone sideways. Eric Kaufmann, Head of Strategy and Operations at Manatee, a pediatric behavioral health organization, came to revenue cycle management through operations rather than billing, and now oversees clinical, finance, and revenue cycle work there.
They sit on opposite sides of the same problem. Speaking at the SUD Business Summit, they agreed on the answer to that problem faster than either expected: the financial health of the revenue cycle is shaped long before a claim reaches a billing queue.
For Carlson, that answer starts even earlier than most people assume. "When a patient picks up the phone and schedules an appointment with your practice, they are participating in a revenue cycle activity," she said.
It's a small reframe with a big consequence: it moves the starting line of the revenue cycle much earlier than where most teams draw it.
"We need to stop viewing revenue cycle management as a reactive process focused solely on appealing denied claims," Carlson said. "We should shift our attention further upstream, starting with patient intake."
Where revenue actually leaks
The revenue cycle extends well beyond billing and collections. Pinpointing exactly where revenue breaks down is no simple task. “It’s frustrating, but the revenue breakdown could happen anywhere, or everywhere,” Carlson said.
A date of birth entered with one digit wrong. A patient scheduled with a provider outside their network. Benefits that were never verified. Providers delivering excellent care who were never formally trained to document and bill for it, so the time and work they put in never makes it onto the claim. And at the far end, aged A/R claims that were simply never touched.
Each issue looks different operationally. Financially, they can all eventually show up as the same thing: revenue that is delayed, denied, or never collected.
Kaufmann came into his role expecting the largest opportunity to sit at the claims stage. His experience at Manatee changed that assumption.
"I assumed right away that the most value was at the claims end stage. But in reality, we found the most issues, and a disproportionate amount of value, in improving the upfront patient intake and insurance eligibility processes, versus simply working claims harder."
That distinction matters. A denial may appear in a billing queue. The activity that created it may have happened weeks earlier, somewhere else in the organization.
Why revenue cycle problems stay hidden
Although these problems originate upstream, organizations often don't notice them until they've already begun affecting cash flow. Two factors explain the delay.
The first is timing. Many of the metrics used to evaluate revenue cycle performance reflect outcomes that have already occurred. "A lot of the key metrics we use to assess revenue cycle performance are lagging indicators," said Carlson. "We look at collection rate and denial rate. But payer adjudication timelines are 14 to 45 days, and sometimes as long as 60 days. If you're using the payer's adjudication of a claim as your signal, whether it was successfully collected or denied, you're fixing problems that started two months ago."
Net collection rate, days in A/R, aging A/R, and denial rate all matter. But by the time those numbers move, the underlying operational behavior may have been occurring for weeks. Kaufmann sees the same issue play out in executive reporting: "You can get fooled into overly fixating on those metrics instead of the root causes," he said.
The second factor is structural. Kaufmann described a workflow that crosses care coordinators, billing and administrative staff, clinical teams, and often the patient—each holding one piece of the picture, none holding all of it. "You need someone who can run point and connect the dots from A to Z, because everyone is working in their different silos," he added.
Carlson sees the same gap surface as a question no one can answer cleanly. Ask an organization who owns the net collection rate, she said, and the response is usually some version of: "Everybody, nobody, I don't know."
Without clear ownership, downstream teams can spend enormous effort resolving the visible symptom while the activity creating it continues upstream. Carlson calls it the "appeals factory": teams repeatedly working denials without creating the feedback loop needed to reduce the volume coming back.
What a single denial code can hide
The issue becomes concrete with CO-16, the missing-information denial that Carlson uses to illustrate how misleading a single code can be. On paper, CO-16 looks simple: one code, one meaning. In practice, it only means that the payer has found something wrong, without saying what.
"One of our favorite examples is CO-16, which, for those who are not RCM nerds, is a missing-information denial. One thing I find really interesting is that when you break down CO-16 denials and their full volume, they can represent many distinct root causes," Carlson explained.
For one Joyful customer, Carlson's team built a separate project for each root cause sitting underneath that single code. They ended up with 17. Rather than treating each one as a missing-information denial to correct and move past, they traced what was actually driving it. "Some were caused by a provider taxonomy code not being appropriately listed in the payer enrollment system. Some were caused by the billing address being wrong on the insurance contract," she said.
Those two examples belong to different teams. One is a credentialing and enrollment fix. The other is a contract data fix. Grouped under a single denial code in a single work queue, they look like the same problem, and they get worked as though a corrected resubmission will resolve them. It resolves neither.
That is the difference between categorizing a denial and understanding it. The payer response gives you enough to know something broke. Finding out what broke, and who owns it, is a separate piece of work.
Two assumptions worth examining
Beyond the operational fixes, two assumptions about how RCM works are worth revisiting.
The first concerns technology. "AI is not a magic bullet in RCM. That's a misconception I would love for folks to leave behind," Carlson said. Her framing is people, process, and technology, in that order. Are the people enabled with what they need? Are the processes working, and is there a handoff breaking somewhere? Only once those two hold up does technology actually help. Skip ahead, and all it does is speed up the chaos already underneath it.
For Kaufmann, the assumption worth challenging was that revenue cycle problems belong entirely to the revenue cycle team. "RCM expertise is extremely important to doing this well, effectively, and correctly," he said. "But you can partner with operations or finance to make improvements and almost offload some of the effort and complexity."
What’s next in the series
Part 2: Why the industry keeps reading signals that are already two months old, and what to track instead if you want warning before a denial, not after.
Part 3: What AI genuinely does well in revenue cycle work, where human judgment still belongs in the driver's seat, and where that line actually sits.
Start with the root cause. If your team is working denials by code rather than by driver, our Root-Cause Matrix maps the operational owner behind each of the most common denial categories. Download it here.
