How to mine your 835 files for six-figure savings that actually stick

Hi and happy Tuesday,

A lot of practices we pitch Auxee to say the same thing:

“We already have a claim scrubber.”

They’re right. Nearly everyone does. Automation at the front end is now a table-stakes requirement. However, here’s the problem:

Once a claim is submitted, the scrubber’s job ends. Then, your real revenue risk begins.

The blind spot most teams miss

Let’s start with a stat: denials often run between 10 and 15 percent. That kind of volume adds up to significant lost revenue. One behavioral health operator we spoke to manages dozens of facilities and already had claim scrubbers in place, following all the standard best practices.

Then we asked:

“What happens when a claim gets denied?”

There was a long pause. In that silence, the real issue came into focus: revenue quietly slipping away.

Here’s what we typically see:

• Denials go unresolved due to weak triage

• Rework happens manually, with no learning loop

• 835 files go unanalyzed, so patterns stay hidden

Where the answers actually live: Inside your 835

We get it. The 835 remittance file is… ugly, dense, and easy to ignore.

But buried in that mess we can see signals that indicate:

• Recurring denial types by payer

• Modifier and documentation patterns

• Small-dollar adjustments silently slashing revenue

A major U.S. health system started paying attention to its 835 data and made small shifts with big results:

• Denials dropped from 27% to 6.5%

• “Discharged not billed” accounts shrank by $15M/month

• Overdue billing hit historic lows (top-tier Epic performance)

The lesson: If you’re not looking at the 835 regularly, you’re likely missing the most actionable part of the revenue cycle.

What top-performing teams are doing differently

They’ve moved past just cleaning claims up front. Instead, they’re building active feedback loops:

• Ingesting 835s daily to surface denial trends

• Using AI to auto-correct and resubmit (when allowed)

• Flagging costly denials before they pile up

• Prioritizing appeals by dollar impact, not filing order

If your team is still reviewing 835s manually, or worse, not at all, you might be flying blind.

Here are four signs:
❌ Denial trends only surface after the deadline
❌ You can’t trace top revenue leaks back to specific codes or payers
❌ Denial rates are flat despite high-effort scrubbing
❌ Rework keeps showing up downstream

Scrubbing alone isn’t enough. The feedback loop has to start where the payer speaks — the 835.

Clean claims are just the beginning. True performance comes from tracking how payers respond and closing the loop at denial resolution. That’s where most teams are still losing (unless they're using Auxee 😉).

Start mining your 835s with this AI prompt

Here’s a plug-and-play prompt to analyze your 835 files using ChatGPT. Copy the prompt below, attach a cleansed 835 file and watch the magic happen.

⚠️ CAUTION: You must scrub your 835 files from any Protected Health Information (PHI) for compliance. NEVER feed full data into any public AI. For a list of 18 items that must be deleted or de-identified, refer to this list.

At Auxee, we can help you deploy a secure internal AI chatbot for your practice at a fraction of the cost charged by OpenAI or similar services. Your team will be able to use ChatGPT without ever worrying about HIPAA violations. Reply to this email for a quote.

Next week, I’ll show you all the different ways AI can be used in healthcare practices, from the second a patient walks through the doors all the way to collecting your hard earned dollars.

See you next Tuesday,
Dino Gane-Palmer