A large orthopedic surgery and sports medicine group with 85+ physicians across 20 locations — performing a high volume of surgical procedures, joint replacements, and in-office treatments. The group processes approximately 30,000 claims per month, heavily weighted toward commercial payers and Medicare, with denial follow-up managed by a 12-person AR team. Orthopedic claims carry above-average denial complexity due to frequent prior authorization requirements, medical necessity challenges, and bundling disputes on surgical procedures.
The group’s denial rate wasn’t the problem — at roughly 11%, it tracked close to industry benchmarks. The problem was what happened after a denial came in.
The 12-person AR team was manually triaging denials from ERA and EOB files, classifying them by hand, and deciding which to pursue based on individual judgment. Senior staff knew the payer-specific nuances — which denials were worth appealing, what documentation each payer required, which timely filing windows were tight. But that knowledge lived in their heads. When junior staff worked the same denial types, resolution rates dropped and turnaround times doubled.
A deeper analysis revealed the real damage: the team was only working about 65% of all denials. The remaining 35% — mostly lower-dollar claims that didn’t make the manual triage cut — were aging past filing deadlines and getting written off. Not because they couldn’t be overturned, but because the team didn’t have the bandwidth to get to them.
The CFO commissioned a write-off analysis. The finding: roughly $2.4M per year in denials were being written off without a single follow-up attempt. Of those, an outside review estimated that more than half were recoverable — straightforward coding corrections, missing documentation that existed in the EMR, and eligibility issues that could be resolved with a single call.
The group had tried adding two contract AR specialists. It helped with volume for a few months, but the contractors didn’t know the payer-specific patterns. Their overturn rate was less than half of the senior staff’s — and at $30–$40 per hour, the cost-per-recovered-dollar was hard to justify.
NDS was brought in to solve a specific problem: make the existing AR team capable of working all denials — not just the ones senior staff had time to get to.
The deployment started with the client’s ERA and EOB data. NDS ingested six months of historical remittance files and mapped every denial to its CARC/RARC codes, payer, dollar amount, procedure type, and outcome. This produced a detailed picture of which denials the team had been winning, which they’d been losing, and which they’d never touched at all.
The system then classified every incoming denial automatically — by type, root cause, payer, dollar value, and complexity — and routed them into prioritized work queues. Prioritization rules were configured with the client’s AR leadership: high-dollar denials first, then claims approaching filing deadlines, then recoverable lower-dollar denials that had historically been written off.
The capability that changed the team’s day-to-day workflow was the AI-powered follow-up guidance. A custom LLM — trained on denial resolution data, payer-specific guidelines, and corrective actions — provided each AR staff member with step-by-step resolution instructions for every denial in their queue. For a coding denial from a specific commercial payer, the guidance would tell the staff member exactly what documentation to pull, which corrected code to submit, and whether the payer required a formal appeal or accepted a corrected claim resubmission. For an authorization denial, it would specify the form, the fax number, and the supporting documentation the payer’s policy required.
This meant junior staff — and even the contract specialists the group brought back — could work denials at the same level as the most experienced AR staff. The institutional knowledge that had been trapped in two or three people’s heads was now embedded in the system and available to everyone on the team.
As denials were resolved, the system captured every outcome and fed it back into the model through reinforcement learning. Guidance for denial types the team was successfully overturning became more precise. Denial types with low overturn rates were flagged for root cause investigation and fed back to the coding and front-end teams to prevent upstream.
A real-time denial dashboard gave management visibility into the entire denial inventory for the first time — not just the denials the team was working, but every denial that came in, its status, its age, and its resolution. The CFO described it as the first time he could see the full picture of what was being left on the table.
Pilot on the client’s highest-volume commercial payer — representing roughly 30% of total denials. One early issue: the AI’s initial guidance for bundling denials referenced a payer policy that the payer had updated mid-year but hadn’t published externally. The client’s senior AR specialist flagged the discrepancy, NDS updated the payer intelligence layer, and the correction propagated across all bundling denials for that payer within 24 hours. This became the template for ongoing payer rule updates.
Expanded to all commercial payers and Medicare. Custom queue configurations added for authorization-related denials, which the client’s team handled differently from coding denials. The two contract AR specialists were brought back and onboarded in under a week — the AI guidance eliminated the ramp-up time that had been the bottleneck in previous staffing attempts.
Full production across all payers and denial types, with ongoing reinforcement learning from resolution outcomes. Root cause reporting activated — the client used upstream denial pattern data to make two specific changes to their front-end eligibility verification workflow that reduced authorization-related denials by 18% within three months.
in annual recovered revenue from denials that were previously written off without follow-up — representing 75% of the $2.4M the CFO’s analysis had identified as recoverable.
of denials now worked — up from 65%. Every denial is classified, queued, and tracked through resolution. The 35% that had been falling through the cracks no longer exists.
increase in denials resolved per FTE per day. Junior staff and contract specialists now resolve denials at a rate and accuracy comparable to the most experienced AR staff — guided by the same payer-specific intelligence.
reduction in average days to denial resolution. Faster classification, prioritized queues, and step-by-step guidance eliminated the research time that had been the primary bottleneck.
Unexpected outcome: The root cause analytics surfaced a pattern of eligibility-related denials from a single payer that had been increasing for months but was invisible in the manual workflow. The pattern traced to a change in the payer’s COB requirements that the front-end team hadn’t caught. Once corrected, eligibility denials from that payer dropped 62% — preventing an estimated $340K in annual denials that would have entered the AR pipeline.
What the Client Said
"We didn’t have a denial rate problem. We had a denial follow-up problem. The denials were coming in at the same rate as everyone else — but we were only getting to two-thirds of them. NDS didn’t change the number of denials we receive. It changed the number we resolve. And the fact that my newest AR person can now work a complex payer denial as effectively as my 15-year veteran — that’s what made the CFO sign off on the second year."
Discover the way forward through more case studies.
Share a sample of your remittance data and your organization’s business rules, and we’ll build a proof of concept showing how NDS classifies, prioritizes, and guides resolution — tailored to your specific payers and denial patterns. No setup cost. No commitment. Just proof on your actual denials.
© 2026 NDS InfoServ. All rights reserved.