McKinsey estimates healthcare revenue cycle costs the industry more than $140 billion annually. Sitting inside that number is a specific, structural loss that is almost entirely preventable: revenue that gets denied, never appealed, and written off -- not because the denial was valid, but because no one had the time or the data to fight it. As many as 60 percent of denials are never appealed. At a community hospital with $200 million in net patient revenue and a 20 percent denial rate, that could represent $14 million or more in uncontested write-offs every year.
$140B+
Annual industry RCM cost
60%
Denials never appealed
$14M+
Potential annual write-offs at $200M NPR
Payers have invested heavily in making sure that doesn't happen. Their adjudication systems are sophisticated, automated, and designed around a simple economic insight: a denial that is never appealed is free revenue. They just need to make the appeal process slow, expensive, and uncertain enough that most providers conclude the effort isn't worth it. For smaller and rural hospitals operating on thin margins, that calculus often comes out in the payer's favor. The hospital writes off the claim and moves on.
That dynamic is not going to fix itself. The only response that changes the economics is a systematic one -- intelligence applied at every stage of the revenue cycle, before the denial happens when possible, and with the specific CMS documentation needed to fight it when it happens anyway.
Seven places to intervene -- and why all seven matter
There is no single intervention point that solves the problem. Organizations that focus only on denial management improve their overturn rate but don't reduce their denial volume. Organizations that focus only on clean claim submission improve their first-pass rate but miss the upstream documentation and prior authorization issues that keep denial rates elevated. The HFMA revenue cycle framework maps seven distinct stages where intelligence makes a material difference, and all seven are connected.
Front end -- pre-service (Stages 1-2)
Stage 1: Registration and eligibility. Surface payer constraints before care begins. The intelligence layer checks eligibility transactions, flags authorization requirements and deadlines, and identifies coverage limitations and network risk -- including CMS coverage checks against plan rules.
Stage 2: Prior authorization. Submit correctly, on time, with justification. The system cross-references the payer's auth matrix with clinical indication documentation, surfaces historical denial rates for missing auth, and flags cases where a Medicare Advantage plan requires authorization for a service CMS does not require -- with appeal rights surfaced immediately.
Most denials don't originate in billing. They originate in registration and prior authorization. An eligibility check that doesn't surface a benefit limit at registration becomes a denial six weeks later that is genuinely hard to fight.
The most important output of the intelligence layer at stage 2 isn't just "auth required." It's the specific clinical documentation the plan needs to approve the auth -- drawn from that plan's actual criteria. And when a Medicare Advantage plan is requiring authorization for a service that CMS does not require, that's a specific, actionable fact that belongs in the auth submission and potentially in an appeal.
Middle -- point of service (Stages 3-5)
Stage 3: Clinical documentation and coding. Prevent documentation-related denials at the source. The system flags payer-specific denial risk with percentages, identifies the specific code to add for medical necessity, and alerts on undercoding with comorbidity review.
Stage 4: Utilization review. Justify admission status and continued stay. The system identifies inpatient status risk with specific criteria gaps, compares length of stay against payer DRG averages, and flags cases where the plan is applying stricter standards than CMS -- documenting for appeal before the denial arrives.
Stage 5: Discharge planning. Align post-acute disposition with payer criteria before discharge. The system identifies post-acute denial risk, flags required clinical assessments before referral, and confirms when CMS qualifying criteria are met with the documentation path for appeal.
A physician documenting a diagnosis doesn't know that this specific payer denies that service at a 34 percent rate when paired with that diagnosis code. A case manager planning a discharge doesn't know that this payer requires a physical therapy assessment before it will approve the SNF stay.
When an MA plan applies more restrictive criteria than the CMS standard -- and they frequently do -- the concurrent review record needs to document against the CMS standard so the appeal argument is built before the denial arrives. It's a matter of surfacing the right prompt at the right moment in the clinical documentation process.
Back end -- post-service (Stages 6-7)
Stage 6: Pre-bill review and claim submission. Final risk assessment before the claim goes out. A claim risk score -- based on this payer's actual behavior with this claim configuration, not industry averages -- lets the revenue integrity team make a data-driven decision about whether to hold the claim and fix something, or submit and prepare the appeal documentation. The system also runs a CMS conflict check and identifies appeal arguments pre-submission.
Stage 7: Denial management, appeals, and learning. Fight winnable appeals -- and every outcome feeds the model. The system classifies denials by appeal winnability, drafts appeals with CMS citations, and propagates pattern updates back to stages 1 through 6.
The learning loop: every claim outcome -- paid, denied, overturned -- flows from the clearinghouse back into the intelligence layer, updating denial predictions across all seven stages for every future similar claim.
The organization that has been running this system for twelve months knows its payers' denial behavior better than it did at the start. It files more appeals, wins more of them, and prevents more denials at the front end because the model has learned what actually gets paid.
The CMS thread that runs through all of it
CMS publishes National and Local Coverage Determinations that define exactly what is covered, for which diagnoses, under which conditions. Medicare Advantage plans are legally required to cover all services that traditional Medicare covers -- they cannot apply internal criteria more restrictive than CMS standards as grounds for denial.
In practice, they do it constantly. They count on the fact that most hospitals won't identify the specific CMS rule the denial violates, won't cite it in the appeal, and won't have documentation structured to support the CMS standard. An intelligence layer that tracks NCD and LCD updates and surfaces the specific citation when an MA plan's denial contradicts a CMS standard changes that calculation entirely. The No Surprises Act creates additional appeal rights. The three-day qualifying stay rule protects SNF coverage. These aren't obscure regulations -- they're public law. Using them systematically is an operational capability, not a legal strategy.
What this means for the CFO and COO
At $200 million net patient revenue, a 1.5 point improvement in cost to collect is $3 million annually. The investment is defensible because of sequencing -- you don't have to solve all seven stages simultaneously. Start with a Revenue Integrity Audit, find the two or three intervention points where the ROI is highest and the data is ready, generate the return inside a budget cycle, and use that proof point to fund the next phase.
The CFO question is always "how fast do we get paid back, and how do we know?" The answer has to be specific: a denial prediction model targeting your highest-volume payer, generating measurable improvement in first-pass rate within 90 days. Not a platform deployment. A targeted intervention with a measurable outcome.