Every Denial You Don’t Appeal Teaches the Algorithm Something

By Missy Harbert, Solution Advisor, Revecore 

During our HFMA national webinar last month, Cynamin Kinard, VP of Revenue Cycle at Tanner Health System, reminded attendees that payers have a cost to process denial appeals too. When a hospital starts pushing back on a category of denials, that cost climbs and the payer’s math changes. 

We see this pattern across our provider partners. When high-volume, low-dollar denials go unworked, payers keep issuing them. Once a team starts appealing them consistently, the volume often slows. Payers watch your appeal rate by denial type, your cost-to-appeal floor and how quickly your team responds, and they adjust. 

Cynamin has spent 25-plus years across nearly every corner of the revenue cycle. Her stories, along with the audience’s live poll responses, showed how closely payer models track the way hospitals respond to denials. 

Watch the HFMA webinar

The audience already feels it 

We asked attendees how much they believe payers use AI to influence denial decisions. 40% said extensively, and another 46% said moderately. Nearly nine in ten are seeing it in their own results. 

Tanner’s data shows the same trend. Cynamin compared zero-pay remits arriving within five days of the bill date in July through September 2024 versus this year, and the share rose from 9% to 22%. Payer medical record requests at Tanner have also jumped 50% to 70% in two months, which signals payers may be building cases before issuing denials. 

Payers can read your thresholds 

Payers know what it costs you to appeal. Premier’s national survey found the cost of fighting a denial rose to $57.23 per claim, and that roughly 70% of denials are ultimately overturned and paid. A claim worth less than that is low risk for the payer, and each denial type that goes unchallenged lowers the risk further. 

So, I wasn’t surprised when 42% of attendees said undetected underpayments are where the most revenue goes missing. These claims post as paid and never reach a denial work queue. Becker’s cites industry analyses suggesting providers lose 1% to 11% of net patient revenue to underpayments each year. 

Tanner spent about two years building line-item review of denials and underpayments.

Among its catches: 

• Mammogram and venipuncture denial patterns across several payers, some of which said Tanner was the first provider to report the issue. 

• A Medicare sequestration double-deduction, caught within 24 hours and escalated to the MAC. 

• Bulk short-stay reviews from one national payer, with 30-day response windows, that Tanner overturns 75% to 80% of the time. 

“Just because they’re using AI doesn’t mean it’s right,” Cynamin said. Tanner appeals every no-auth denial where a valid authorization exists, and those appeals recover significant dollars each quarter. 

Prioritize the way payers do 

Our third poll asked how organizations prioritize denials and underpayments. 38% said they use data, weighing likelihood of recovery against impact. Two years ago, I’d have bet on first-in, first-out. Payers score claims by probability and value, and providers can do the same by weighing effort, overturn rates by payer and denial type, and full financial impact, including the case mix index erosion that follows a DRG downgrade. 

Tanner puts just as much effort into prevention. Through root-cause work with its denials governance team, it reduced quarterly zero-pay remits by about 81% over two years, all before adding AI. Cynamin was candid that “escalation is not prevention.” 

Where AI and clinical judgment meet 

Tanner is now rolling out EHR-based AI for appeals and record releases, starting with ED E&M downgrades. The appeal template went into the AI workflow only after it overturned 19 of 36 test appeals by hand with one payer. 

We follow a similar order at Revecore. AI helps us review lengthy medical records and draft payer-specific appeals faster, and our RNs, certified coders, and attorneys make the final clinical and legal call. For underpayments, we review the full paid-claim population rather than stopping at a balance threshold. Cynamin said sending complex denials to Revecore “cleared out a lot of noise” and freed her team for higher-value work. 

Cynamin’s closing advice was to keep challenging what you know you should be paid for. The payer’s model is watching which denials you fight and which you let go. 

Go deeper: Download the whitepaper 

When AI Tips the Scales: How Payers Are Redefining Denials and What Hospitals Can Do About It lays out a framework for spotting AI-driven denial and underpayment tactics, prioritizing by recovery potential, and building prevention into your strategy.