The ROI of AI Voice Agents for Oncology Billing

Oncology billing carries some of the largest dollar amounts in medicine, and that is exactly what makes it dangerous. A single infusion claim can run into the tens of thousands. When one gets denied and nobody works it before the appeal window closes, the practice does not lose pocket change. It loses a number that shows up on the P&L, and it loses it quietly.

The story in most oncology groups is not that the billing team lacks skill. It is that the team is buried. Staff handle whatever is loudest that day, and the structured follow-up work slides. Meanwhile the aging report climbs and high-value claims drift toward their deadlines.

Why oncology denials outrun the billing team

The revenue cycle is a stack of deadlines, and oncology stacks them higher than almost any specialty. Prior authorization on drugs and imaging. Timely-filing windows that differ by payer. Appeal deadlines that start the day a denial posts. Miss any one of them and a five-figure claim is gone, no matter how medically appropriate the regimen was.

Denials are not rare edge cases. An analysis of in-network claims in ACA marketplace plans found insurers denied a meaningful share of submitted claims, and only a tiny fraction of those denials were ever appealed (KFF). For an oncology practice, an un-worked denial on a drug claim is earned revenue erased because nobody had an open hour to make the call.

The follow-up is structured and repetitive, which is why it slips. Checking claim status on a payer portal, capturing a denial reason, requesting a corrected remittance, calling a patient about an aged balance. None of it needs clinical training. All of it needs someone with a phone and time on the clock.

What an AI voice agent actually does in the revenue cycle

Pretty Good AI builds voice agents that handle the administrative calls in your billing workflow, integrated with athenahealth. The agent is a front-office layer. It does not decide what care a patient needs, and it does not make coding calls that require a certified coder. It works the phone-and-portal grind that keeps money moving.

On the payer side, the agent places outbound status calls on pended and denied claims, captures the denial reason and any reference number, and writes it back into athenaOne so a biller can act. On the patient side, it calls on aged balances, explains what a statement covers, and routes anyone who wants a payment plan or has a billing question to the right staff member. Before a scheduled infusion, it confirms eligibility and benefits so the practice is not finding a coverage gap after the chair is booked.

The goal is not to remove people from revenue cycle work. Denial management and appeals are where experienced billers earn their keep, and coding accuracy is central to getting paid correctly (AAPC). The goal is to stop spending that expensive judgment on hold music.

The revenue math for some of the highest-value claims in medicine

Revenue cycle leaders track the cost to collect and the share of net revenue lost to preventable write-offs, and both improve when follow-up happens on time instead of eventually. In oncology the stakes are higher than in most specialties because the claim values are enormous. Recovering even one or two denied drug claims a month can pay for the automation several times over.

Run the numbers for your own group. Take your average denied-claim value, multiply by the denials that currently age past their appeal window, and you have the annual figure sitting on the table. In oncology that number is often startling, and the reason it exists is a shortage of hours, not a shortage of skill.

Keeping it administrative and keeping it clean

Everything the agent does is logistics: status calls, balance calls, benefit confirmation, and routing. When a call surfaces anything that needs a clinician or a certified coder, the agent hands it off with the context already captured. That keeps the practice on the right side of the line. The AI moves information and money, and people make the decisions that require training and licensure.

Key Takeaways

• In oncology, an un-worked denial can cost five figures because drug and infusion claims are among the highest-value claims an outpatient practice files.
• Follow-up slides because it is repetitive, not because the billing team lacks skill.
• An AI voice agent places payer status calls, works patient balances, and confirms benefits before scheduled infusions, writing everything back to athenaOne.
• Keep billers and certified coders on appeals and coding judgment; put the phone-and-portal grind on automation.
• Calculate your exposure: average denied-claim value times denials aging past appeal deadlines equals annual revenue at risk.

Oncology billing does not fail because your team is not good enough. It fails because there are more deadlines than there are hours, and the claims at risk are the biggest ones you file. Give the repetitive follow-up to a voice agent and let your billers spend their day on the appeals and coding work that actually needs them.

Sources

• https://www.kff.org/private-insurance/issue-brief/claims-denials-and-appeals-in-aca-marketplace-plans/
• https://www.aapc.com/resources/what-is-medical-billing