Payer Revenue Cycle Automation Starts Here | Skan AI Process Intelligence

U.S. health plans collectively face $87 billion in improper payments and a 19 percent in-network claim denial rate. Process intelligence is the first step to fixing both: it establishes the operational ground truth that every downstream revenue cycle improvement, automation, and AI deployment depends on. Most AI investments stall because technology is deployed onto processes that have never been accurately mapped.

What is revenue cycle automation in health insurance?

Revenue cycle automation is the use of technology to reduce manual work across the claims lifecycle, from prior authorization through adjudication, coding, payment, and denial resolution.

For health plans, the revenue cycle touches every team and system that handles a claim: eligibility verification, utilization management, prior authorization, claims adjudication, coding review, payment integrity, and appeals. Each step involves multiple applications, multiple staff roles, and dozens of process variations.

Automation produces consistent results when it runs on consistent, well-understood processes. Without that foundation, health plans automate the wrong version of a workflow, at scale.

Why is the payer revenue cycle still broken?

The payer revenue cycle is still broken because most health plans lack visibility into how work happens across their operations. They have system logs, but not operational ground truth.

The 2025 CAQH Index found that U.S. healthcare avoided $258 billion in administrative costs through electronic transactions in 2024, yet more than $20 billion in additional savings remain from manual workflows that have not yet shifted to electronic. For prior authorization alone, CAQH estimates the industry could save nearly $494 million annually by moving from manual to fully electronic.

Three compounding problems drive the gap between what is possible and what is being achieved:

• Process variation: Staff complete the same tasks in dozens of different ways depending on the system, team, and claim type. No two processors work identically, and most payers have no map of that variation.
• Application complexity: Claims processors toggle between multiple legacy systems, generating invisible time losses that never appear in a workflow log and cannot be measured by traditional event-log tools.
• Context-free automation: AI tools deployed without an operational baseline inherit the inefficiencies they were meant to fix, automating the most common workflow variants rather than the best ones.

Traditional process mining tools rely on back-end event logs, capturing roughly 15 to 20 percent of actual work activity. The rest of it: desktop interactions, manual lookups, workarounds, and informal coordination, is invisible to the systems meant to improve it.

Gartner projects that more than 40 percent of agentic AI projects will be canceled by 2027 due to escalating costs, unclear business value, or inadequate risk controls (Gartner, June 2025). Health plans accelerating AI investment without an operational foundation are running directly into that risk.

What does the CMS prior authorization mandate require from health plans?

The CMS prior authorization mandate (CMS-0057-F) requires impacted health plans to issue standard prior auth decisions within 7 calendar days and expedited decisions within 72 hours, effective January 1, 2026, with FHIR-based electronic APIs fully operational by January 1, 2027.

The rule applies to Medicare Advantage organizations, Medicaid and CHIP managed care plans, and QHP issuers on federally facilitated exchanges. It is the most significant structural change to prior authorization in a generation.

Medicare Advantage plans made nearly 53 million prior authorization determinations in 2024 (KFF). The AMA reports that 94 percent of physicians say prior authorization delays patient access to necessary care, with physicians and staff spending an average of 13 hours per week managing the process.

The mandate does not just compress timelines. It requires payers to document specific denial reasons, report PA volume and turnaround metrics publicly, and connect to a standardized electronic infrastructure. Meeting those requirements means knowing, in precise operational detail, how prior authorization workflows execute in practice across every team and system today.

Agentic Process Intelligence provides that baseline. Skan AI observes every step of the prior authorization process as it happens, not as it appears in policy documentation. That makes it possible to standardize the workflow, identify where delays and errors are introduced, and build Agent Operating Procedures (AOPs) that scale best practices across the entire prior auth operation.

Health plans that have established process intelligence as their AI foundation report consistent improvement across claims cycle time, prior authorization throughput, and payment integrity outcomes. The starting point is always the same: see how work gets done.

What are the most common types of claims denials?

The most common types of claims denials fall into five categories: registration and eligibility errors, administrative failures, prior authorization issues, coding errors, and medical necessity challenges.

KFF's 2024 analysis of ACA marketplace plans found that insurers denied 19 percent of in-network claims. Administrative reasons accounted for 25 percent of those denials, prior authorization or referral issues for 9 percent, and medical necessity for 5 percent. Eligibility errors and coding-related denials make up a significant share of the remainder, varying by payer and line of business.

The rework cost per denied claim ranges from $25 to $57 for commercial claims. For prior authorization and medical necessity denials, costs can be significantly higher. Premier estimates that providers collectively spend $19.7 to $25.7 billion annually adjudicating denied claims. A majority of denied claims are never reworked or resubmitted, compounding the revenue impact.

The most telling data point: KFF found that 80.7 percent of Medicare Advantage prior authorization denials that were appealed in 2024 were overturned. That figure has exceeded 80 percent every year since 2019. A large share of denials are not clinically justified. They are process failures, not coverage decisions. They are preventable.

Workflow visibility maps the steps that precede a denial: where information is entered, where documentation is assembled, where authorizations are missed. Fixing the process eliminates the denial before it happens. This is the shift from reactive to proactive revenue cycle management.

How does AI improve medical coding and billing accuracy?

AI improves medical coding accuracy by identifying the specific points in the coding workflow where errors are most likely to be introduced, and by enabling consistent application of coding rules across a distributed team.

ICD-10-CM contains more than 70,000 diagnosis codes. Coding accuracy depends on both the clinical documentation available and the process through which coders access, interpret, and assign those codes. Independent academic research consistently finds that large language models perform far below vendor claims when used as autonomous coders. Their strongest use case is as a verification and auditing layer, not for autonomous code assignment.

What operational process observation adds is something AI coding tools cannot provide on their own: a map of how coders work. It identifies which teams have higher denial rates tied to coding, which application workflows create conditions for error, and which documentation steps are consistently skipped. That operational context is what makes AI coding improvements produce lasting results rather than isolated gains.

What is payment integrity in health insurance?

Payment integrity is the set of processes health plans use to ensure that every claim is paid correctly, once, to the right provider, at the right amount. Most health plans approach it reactively, through post-pay audits and recovery programs. The shift from reactive to proactive requires seeing how claims are processed before they are paid.

CMS reported $31.7 billion in Medicare fee-for-service improper payments in FY2024, at an improper payment rate of 7.66 percent. Insufficient or missing documentation drove approximately 68 percent of those improper payments. For Medicaid, improper payments totaled $31.1 billion at a 5.09 percent rate. Across all CMS programs, total improper payments reached approximately $87 billion in FY2024. CMS itself clarifies that improper payments include documentation gaps and administrative errors, and are not exclusively fraud.

McKinsey estimates that fully deployed AI could reduce a significant share of health plan administrative costs, with the highest gains coming from shifting reactive post-pay audit to proactive prevention built into the claims workflow (McKinsey, 2025). Continuous observation of adjudication decisions creates audit-ready documentation of every process step, supporting compliance reporting without a separate audit layer.

How does process intelligence compare to traditional process mining?

Process intelligence captures 100 percent of desktop activity across every system, while traditional process mining tools see only 15 to 20 percent of actual work through back-end event logs. The table below shows where the two approaches diverge across the five dimensions that matter most for revenue cycle automation.

How does process intelligence fix payer revenue cycle problems at the root?

Payer revenue cycle problems are fixed at the root by establishing the operational ground truth that every downstream improvement, automation, and AI deployment depends on.

Agentic Process Intelligence is the capability to observe, understand, and continuously improve how work happens across a health plan's workforce. Unlike traditional process mining, which relies on system event logs, Skan AI's observation-first methodology captures every click, application switch, pause, and workflow step across every system, including legacy claims platforms, modern SaaS tools, and mainframe environments, without requiring back-end integrations.

Implementation follows three phases:

Phase 1: Baseline. Skan AI is deployed across targeted roles in claims, prior authorization, and member services to capture how work gets done. The output is a Digital Twin of Operations: a continuous, data-driven view of every process variant, time allocation per step, and system usage pattern across the workforce. This answers the foundational question every health plan operations leader needs answered first.

Phase 2: Optimize. The operational baseline drives targeted improvement. Skan AI narrows focus to the workflows with the highest impact on first-pass resolution rates, prior auth cycle times, and denial prevention. It identifies which process variants perform best and what separates high-performing teams from lower-performing ones. Improvements are validated across multiple cycles, with measurable results before any technology investment is made.

Phase 3: Digitize. Optimized processes become the foundation for agentic AI. Process-native agents are trained on observed human work patterns, not assumed workflows. Agent Operating Procedures (AOPs) define how agents navigate each step of the claims or prior auth workflow. Continuous monitoring ensures both human and agent performance stays on track, and the Digital Twin of Operations captures any operational drift before it affects outcomes.

This sequence is not optional. Organizations that skip the observation and optimization phases and deploy AI agents in healthcare payer operations directly inherit the inefficiencies of their existing processes. The operational ground truth established through continuous observation is what makes revenue cycle automation produce measurably better outcomes rather than automate existing mistakes at scale.

Health plans that establish this foundation now build a process knowledge base that is structurally difficult for later-moving competitors to replicate. Every optimization cycle compounds the advantage.

What results have healthcare payers achieved with process intelligence?

Healthcare payers deploying continuous process observation as the foundation of their AI strategy are identifying $10 million to $28 million in annual savings and 30 to 40 percent cycle time reductions across claims, prior authorization, and member services operations.

McKinsey estimates that health plans that fully integrate AI and automation into their operations could achieve $150 million to $300 million in administrative cost savings for every $10 billion in revenue (McKinsey, 2025). The pattern in the highest-performing engagements is consistent: operational visibility precedes every AI investment that holds up at scale.

Health plans that have established a process intelligence foundation before AI deployment are identifying $10 million to $28 million in annual savings across claims, prior authorization, and member services. See how Skan AI builds that foundation for health plans.