Revenue-cycle management is the work of getting a healthcare provider paid for care it has already delivered, and the failure mode is well measured. Change Healthcare's analysis, reported through MGMA, found that eighty-six percent of claim denials are potentially avoidable, and that forty-eight percent of avoidable denials are never recovered. Put those two figures together and roughly half of the denied revenue a provider could have collected is lost permanently, not because the care was disputed on its merits but because a claim went out with a fixable error and then sat unworked. That is what makes revenue-cycle management a clean subject to walk step by step. The benchmarks are public, the cost of each mistake is priced, and almost none of the lost money turns on clinical judgment.
Why this is one instance of a general pattern
A revenue-cycle department runs on people who move data between systems. A patient's coverage lives in a payer's eligibility system. The encounter lives in an electronic health record. The claim is assembled in a practice-management or billing system and transmitted to a clearinghouse, which routes it to the payer, which sends back a remittance that a person reads and keys into yet another screen. The same shape, people carrying structured data across applications that do not connect to each other, appears in a staffing firm reconciling timesheets to invoices and in any back office where the long tail of document variability defeats a rigid integration. Healthcare is one well-instrumented example of a problem that exists wherever the system of record and the system of payment are different systems.
What a denial costs, and how often one happens
The price of getting it wrong is documented. Change Healthcare's denials research, reported through MGMA, puts the average cost to rework a claim at $25.20. That figure covers the staff time to research why a payer rejected the claim, correct it, and resubmit. The frequency is documented too, and it is rising. Kodiak Solutions' benchmarking, drawn from more than 2,100 hospitals and 300,000 physicians on its revenue-cycle analytics platform, found initial denials rose 2.4 percent in 2024, to 11.81 percent of claims. Experian Health's 2025 State of Claims, a survey of 250 revenue-cycle leaders, found 41 percent of respondents seeing at least one in ten claims denied, and 54 percent saying denials are increasing. MGMA's polling puts the typical first-submission denial rate around eight percent and reports that sixty percent of practices saw denials climb in 2024. So a provider pays about $25.20 to rework each of a growing share of its claims, and on the Change Healthcare numbers, half of the avoidable rework that does get attempted still ends in revenue written off.
Where the time actually goes
If you sit behind a biller, the work does not divide evenly across the claim lifecycle. It concentrates at two points, and neither is the part that requires a credentialed coder's judgment. The first is the front end, before service: confirming that the patient's coverage is active, that the plan covers the procedure, that any prior authorization exists. Eligibility errors are a leading avoidable denial reason, and they are knowable before a claim is ever built. The second is the back end, after a denial returns: opening the remittance, reading the payer's reason code, deciding whether the claim was wrong or the payer was, and then either correcting and resubmitting or assembling an appeal. Between those two points sits a large volume of clean claims that need only to be checked against payer-specific rules and sent. The genuinely hard work, an appeal that turns on medical necessity or a coding call that depends on the clinical note, is the smallest slice by count.
The redesigned workflow
The redesign moves the mechanical work to an agent and reserves the judgment for a person. Before service, the agent verifies eligibility against the payer in real time and flags coverage gaps while the patient is still reachable, when an error costs a phone call rather than a denied claim. Before submission, it checks coding and documentation against that payer's published edits and the encounter record, holding claims that would predictably bounce. After a remittance returns, it categorizes each denial by reason code, separating the ones it can correct and resubmit on its own from the ones a human must judge. A person owns the clinical and payer-policy decisions: whether a documented service meets a coverage definition, whether an appeal is worth filing, when a payer's rule is being applied wrongly and should be challenged. The agent assembles the supporting detail, the original claim, the remittance, the relevant policy language, so that when the work reaches a person it is a decision rather than a research task. This is the aligned engagement shape, where the agent absorbs the volume of mechanical claims and the expensive human attention concentrates where it changes an outcome.
Why a denial decays the longer it sits
The recoverable share of a denied claim is not fixed. It falls with time, for reasons that are structural rather than a matter of staff diligence. Payers impose timely-filing and appeal deadlines, often measured in a small number of months from the date of service or the date of denial, after which the claim cannot be recovered at all regardless of merit. The supporting record decays in parallel. The staff who can reconstruct what happened move on, the patient becomes harder to reach for a corrected detail, the encounter grows harder to verify. A denial worked the week it returns is a correction against a fresh record and a live deadline; the same denial worked three months later may be a write-off by rule. This is the mechanism behind the forty-eight percent that Change Healthcare found is never recovered. Avoidable denials are lost because the rework queue is finite and human, the high-value appeals get attention first, and the rest age past the point of recovery. An agent that works the queue continuously, at the rate denials arrive rather than the rate staff can clear them, attacks exactly the share that time would otherwise erase.
A reasonable counter
A reasonable counter is that denial rates are driven by payer behavior, not by provider workflow, so automating the back office treats a symptom while payers keep moving the goalposts. There is real force to this. Kodiak's 2024 increase came from denials over medical necessity and requests for more information, even as authorization-related denials fell, a pattern that reflects tightening payer adjudication. And no amount of clean submission stops a payer from denying claims it has decided to deny. But the Change Healthcare finding cuts the other way on the part that is actionable. If eighty-six percent of denials are potentially avoidable and nearly half of those are never recovered, the binding constraint for most providers is not payer intransigence on the contested minority. It is the avoidable majority that goes out wrong and the recoverable revenue that ages out of an overloaded queue, which is precisely the work that does not require negotiating with a payer to fix. That is also why these denial figures, like the problems the standard delivery models leave unaddressed, have held across thousands of practices for years: the avoidable share is structural, and it is sitting in the workflow.