The average health system spends 3–5 cents to collect every dollar of patient revenue. For many, it's closer to 8 cents. Multiply that across a $90.6 billion US revenue cycle management market and you're looking at billions of dollars burned on the process of getting paid — not on patient care, not on growth, just on chasing money that was already earned.
Now McKinsey and HFMA are putting a number on what AI can do about it: 30–60% reduction in cost to collect. And new HFMA survey data from May 2026 shows the healthcare industry is moving on this faster than anyone predicted.
The HFMA Data: 27% at Scale, 53% Piloting
The HFMA Revenue Cycle of the Future report, published May 2026, surveyed revenue cycle leaders across health systems of all sizes. The adoption numbers tell a clear story: AI in healthcare RCM has crossed the inflection point from curiosity to deployment.
- 27% of health systems now deploy AI at scale across multiple revenue cycle functions
- 53% are running active pilots — testing AI in at least one RCM workflow
- Only 20% haven't started — and that number is shrinking every quarter
That means 80% of health systems are actively using or testing AI in revenue cycle management. If your organization is in the remaining 20%, you're not cautious — you're behind.
Why Cost to Collect Is the Metric That Matters
Revenue cycle leaders track dozens of KPIs — days in AR, clean claim rate, denial rate, net collection percentage. But cost to collect is the metric that captures overall operational efficiency because it measures the total expense of the revenue cycle machine itself.
Here's why AI has such an outsized impact on cost to collect specifically:
1. Labor is 60–70% of RCM cost. The revenue cycle runs on people — billers, coders, authorization specialists, follow-up staff, payment posters. AI doesn't eliminate these roles, but it dramatically reduces the hours spent on repetitive, rules-based work. When an AI agent handles eligibility verification at scheduling instead of a human making calls, the cost per verification drops by 80–90%.
2. Rework is the hidden cost multiplier. Every denied claim that requires manual appeal, every eligibility error caught post-service, every prior auth that wasn't obtained upfront — these create rework loops that multiply cost to collect. AI attacks rework at the source by preventing errors before they generate downstream costs.
3. Speed compounds savings. A claim that resolves in 7 days instead of 45 doesn't just improve cash flow — it reduces the total staff touches, system queries, and follow-up actions required to close it. AI-driven claim submission and payer follow-up compress cycle times, which compresses cost per claim.
The $90.6 Billion Market — Projected to Hit $308 Billion by 2030
The US revenue cycle management market sits at $90.6 billion today. HFMA projects it will reach $308 billion by 2030. That growth reflects two simultaneous forces: increasing healthcare volume and increasing complexity of getting paid.
Every new payer rule, every additional prior authorization requirement, every coding update (270 new CPT codes in 2026 alone) adds friction — and friction adds cost. Without AI, cost to collect will rise in lockstep with market complexity. With AI, organizations can absorb increasing complexity without proportional cost increases.
"It's like becoming a blacksmith when the Model-T is coming out of factories."
That's Dr. Gerard Brogan, Chief Revenue Officer at Northwell Health, describing the coding workforce talent crisis. The quote captures a broader truth about the entire revenue cycle: manual processes built for a simpler era cannot scale to meet today's complexity. AI isn't optional — it's the only way the economics work.
Where AI Delivers the Biggest Cost-to-Collect Reduction
The HFMA report identifies mid-cycle operations as the area where AI will have the greatest impact, with 26.3% of leaders expecting the most change there. Mid-cycle includes claims management, coding, clinical documentation, and charge capture — the dense operational core where errors are expensive and volume is relentless.
But cost-to-collect reduction isn't confined to mid-cycle. AI compresses costs across the entire revenue cycle:
Front-End: Eligibility and Prior Authorization
Front-end errors are the most expensive errors in the revenue cycle because they cascade downstream. A missed eligibility issue at registration becomes a denied claim 30 days later, which becomes an appeal that takes another 45 days. AI-powered prior authorization and real-time eligibility verification eliminate these cascading failures at the point of origin.
Mid-Cycle: Coding, Claims, and Documentation
AI coding assistants are already demonstrating 85–95% accuracy on routine encounters, with human coders reviewing exceptions rather than processing every chart. For claims, AI scrubbing catches errors that would become denials — each prevented denial saves $25–50 in appeal costs and weeks of staff time.
Back-End: Denials, Appeals, and Collections
AI denial management remains the highest-ROI entry point for most organizations. Payers are using AI to deny claims at scale — Droidal reported a 75% reduction in claim rejection handling time with their Claims Processing AI Agent in May 2026. Health systems need matching firepower on the appeal side.
Revenue Intelligence: Underpayment Detection and Contract Compliance
The Quadax/SlicedHealth partnership announced May 26, 2026 illustrates the newest frontier: AI-driven revenue intelligence. By combining AI contract modeling with payment analysis, organizations can automatically detect underpayments, model contract scenarios, and ensure every dollar owed is actually collected. This layer of intelligence turns passive payment posting into active revenue recovery.
| Revenue Cycle Function | Manual Cost per Transaction | AI-Assisted Cost | Reduction |
|---|---|---|---|
| Eligibility verification | $6–12 | $0.50–2 | 75–92% |
| Prior authorization | $20–45 | $3–8 | 80–85% |
| Claim submission + scrubbing | $4–8 | $1–2 | 60–75% |
| Denial appeal | $25–50 | $5–12 | 70–80% |
| Patient collections outreach | $8–15 | $1–3 | 75–87% |
The Workforce Readiness Gap: Only 7.4% Say "Very Prepared"
The HFMA survey surfaces a critical challenge alongside the adoption data: only 7.4% of organizations say their workforce is "very prepared" for AI-driven transformation. This isn't a technology problem — it's a change management problem.
Organizations achieving the largest cost-to-collect reductions share a common approach:
- Start with denial management. The ROI is immediate, the workflow is well-defined, and staff see the benefit within weeks — not months.
- Redefine roles, don't eliminate them. AI handles the repetitive pattern-matching. Staff focus on complex payer negotiations, patient financial counseling, and exception management — work that's more valuable and more engaging.
- Measure cost per transaction, not just headcount. The goal isn't fewer people. The goal is lower cost per claim, per authorization, per collection — with the same or fewer staff handling significantly more volume.
The HFMA Conference Inflection Point
The HFMA Annual Conference runs June 7–10, 2026 in National Harbor, MD. Every major AI vendor, every health system CRO, and every revenue cycle leader will be there. The conversation has shifted from "should we use AI?" to "how fast can we deploy?"
If your organization is attending without a concrete AI deployment plan, you're going to spend four days watching competitors describe the results you don't have yet.
How BAM AI Delivers Cost-to-Collect Reduction
BAM AI deploys AI agents across the full revenue cycle — not point solutions, not chatbots, not dashboards. Autonomous agents that execute the work:
- Eligibility verification — real-time, at scheduling, every patient, every visit
- Prior authorization — automated submission, status tracking, and approval management
- Claim submission — scrubbed, coded, and submitted with 98%+ clean claim rates
- Denial management — pattern detection, auto-generated appeals, recovery tracking
- Payer follow-up — automated AR management that cuts days in AR by 40–60%
The result: cost-to-collect reduction that compounds month over month as AI agents learn your payer mix, catch patterns specific to your practice, and eliminate the rework loops that drain your revenue cycle.