Healthcare organizations are investing heavily in artificial intelligence to improve revenue cycle management automation. AI is being deployed to predict denials, automate workflows, accelerate coding, and streamline reimbursement operations. Yet denial-related burden continues to grow.
Initial claim denial rates rose to 11.81 percent in 2024, up from 11.53 percent the year before. At the same time, Experian Health’s 2025 State of Claims report found that denial rates are increasing for many providers, while 59 percent plan to invest in denial-reduction technology within the next six months.
The trend exposes a growing disconnect. If organizations are investing more in automation than ever before, why does denial-related work continue to rise?
Part of the answer lies in how healthcare measures success. Most organizations focus on financial outcomes such as denial rates, clean claim rates, and days in accounts receivable. While useful, these metrics fail to capture the operational burden required to achieve those outcomes.
A claim may ultimately be paid, yet still requires multiple staff interventions, payer calls, documentation requests, corrections, and follow-up activity before reimbursement is received. Those activities consume labor, delay payment, and create operational friction that traditional metrics rarely reveal.
Denials Create a Hidden Operational Tax
Denials are often viewed as a reimbursement problem. In reality, they create a significant workload problem.
Every denial initiates a chain reaction of administrative effort. Staff must investigate the issue, gather documentation, contact payers, resubmit information, monitor status updates, and determine whether an appeal is worthwhile. Even when reimbursement is eventually secured, the effort required to get there represents a real operational cost.
This hidden burden can be viewed as a “Denials Tax” imposed on healthcare organizations.
The scale is substantial. According to KFF, Medicare Advantage insurers processed nearly 53 million prior authorization requests in 2024. Yet only 11.5 percent of denied requests were appealed. The American Medical Association reports that 59 percent of physicians do not appeal denials because they believe the effort is unlikely to succeed, while 52 percent cite insufficient staff resources.
These figures highlight a reality that denial rates alone cannot capture. Organizations are not simply managing denials. They are managing the work created by denials.
Why AI May Be Solving the Wrong Problem
Many AI solutions focus on improving individual tasks within the revenue cycle. They automate documentation, predict denial risk, identify coding opportunities, or prioritize work queues. These capabilities can create value. However, automating a task does not necessarily reduce the total amount of work required to get paid.
A denial prediction model may flag a claim as high risk. An RCM automation tool may accelerate documentation review. A workflow engine may route work more efficiently. Yet if the claim still requires repeated interventions after submission, the underlying burden remains.
In some cases, automation can simply accelerate flawed workflows rather than eliminate them. The challenge is not making workflows faster. It is reducing the total effort required to move claims from submission to resolution.
The Case for Touch-Level Intelligence
As healthcare organizations expand AI investments, many are beginning to recognize that financial outcomes provide only a partial picture of performance.
A growing alternative is touch-level intelligence, which measures the interactions required to resolve claims rather than focusing solely on reimbursement outcomes.
This approach helps leaders answer questions traditional metrics cannot:
- How many staff interventions are required to resolve a claim?
- Which denial categories generate the most rework?
- Which payers create the greatest administrative burden?
- How much effort produces little or no reimbursement value?
Many denial-related issues begin long before a claim reaches the payer. Eligibility verification errors, prior authorization gaps, coordination-of-benefits issues, and documentation deficiencies frequently trigger avoidable downstream work. Without visibility into where that work originates, organizations remain trapped in reactive workflows.
The Future of Revenue Cycle Intelligence
As payer requirements become more complex, healthcare organizations will need more than faster automation. They will need better visibility into the work occurring behind reimbursement outcomes.
The next evolution of revenue cycle strategy may not be defined by how quickly organizations process claims. It may be defined by how effectively they identify, measure, and eliminate unnecessary work.
AI will remain an important part of that future. But solving the denials problem requires more than automating existing processes. It requires understanding where operational friction originates, how it spreads across workflows, and how much effort it creates along the way.
Until healthcare begins measuring the work behind reimbursement, the “Denials Tax” will remain one of the industry’s most expensive and least visible operational challenges.
The author, David Henriksen, is CEO of MedEvolve.