- Auxee AI for Healthcare
- Posts
- Finding the Dollars Hiding in Paid Claims
Finding the Dollars Hiding in Paid Claims
A simple weekly check that uncovers underpayments, recovers lost dollars, and prevents repeat shortfalls
Dino Gane-Palmer
August 26, 2025
Hi and happy Tuesday,
During a check-in with a mid-sized health system last week, their billing lead shared something troubling:
“We’re not seeing more denials, but our collections feel light.”
A quick review of recent payment
s told the story. Several claims had been marked as “paid,” yet the amounts were lower than the expected payment amount.
No alerts. No denials. Just small shortfalls adding up.
These quiet gaps—payment variances—can slip past even experienced teams.
A straightforward weekly process
If you have your paid claims in a spreadsheet or CSV like this:
(AI will add variance column)
You can use your favorite AI chatbot and run the following prompts and have the problem identified in minutes, just remember to work from de-identified data.
Compare paid vs. expected allowed
Prompt: “Using this CSV, add a column showing the difference between paid_amount and expected_amount. Flag any difference greater than 2% or $5.”
Group and rank by payer and code
Prompt: “From this table, group flagged rows by payer and CPT/HCPCS code. Count the claims in each group and total the variance dollars. Rank from highest to lowest dollars.”
Identify patterns
Prompt: “Highlight any payer–code combinations that appear more than twice. Suggest possible reasons for these variances based on the data provided.”
Create a recovery worklist
Prompt: “List the top five payer-code combinations by total variance dollars. For each, provide the claim IDs and variance amounts in a table.”
Track for prevention
Prompt: “Summarize the top three recurring issues from this data. Output in bullet points I can add to my payer reference guide.”
But why does this matter?
Short answer: because more revenue is lost, and it occurs at different scales. For example:
In 2023, hospitals lost $130B to Medicare and Medicaid underpayments - Medicare only paid 83 cents on the dollar for care.
Skipping variance checks can cost 1%–11% of revenue, even at the low end, which is equivalent to multiple salaries a year.
Medicare pays interest on clean claims over 30 days old, but few track it.
On 8,000 monthly claims at $180 each, a 1% shortfall is $14K a month.
Is AI the answer here?
AI isn’t the fix; discipline is. However, AI makes that discipline easier to apply consistently, every week, without the grind. Here are 5 ways AI can help:
Catch underpayments hiding in “paid” claims. The ones your team doesn’t have time to hunt.
Group repeat variances, so you’re not solving the same problem repeatedly.
Turn spreadsheets into clear worklists, what to fix, where to follow up, and who owes what.
Spot patterns early so you can update payer guides and stop future shortfalls cold.
Make underpayment checks part of your routine, not a one-off project or fire drill.
But, from our experience, there are much higher ROI use cases for AI in RCM. Check out these past issues for examples:
Next week, I’ll cover 5 prompts that will change your RCM.
See you next Tuesday,
Dino Gane-Palmer
That's it for today!Before you go we’d love to know what you thought of today's newsletter to help us improve the experience for you. |
 Dino Gane-Palmer | About the Author Dino Gane-Palmer is the founder of Auxee and CEO of PreScouter, an Inc. 5000–recognized innovation consultancy that helps Fortune 500 companies and global organizations capitalize on new markets and emerging technologies. He launched PreScouter while earning his MBA at Kellogg and later founded Auxee to help teams use AI to tackle complex, research-heavy workflows. His work has supported decisions at some of the world’s leading healthcare, manufacturing, and consumer brands. Dino is also the author of the best-selling book Do More With Less: The AI Playbook, a practical guide to applying AI where it matters most. |