Revenue cycle has reached an interesting point in its automation journey. The question is no longer whether technology can do meaningful work. Increasingly, it can. The harder question is whether automating more individual tasks is actually making the revenue cycle work better as a whole.
AI and automation are already reshaping patient access, prior authorization, coding, billing and payment. The conversation among health system leaders is shifting accordingly. Recent discussions with executives from Mayo Clinic and Jefferson Health, for example, focused less on individual technologies and more on redesigning the enterprise operating model around how work actually moves from scheduling through payment.
Yet significant friction remains. HFMA reports that 11.65% of healthcare claims were denied on first pass in 2025. Revenue cycle teams still contend with manual handoffs, disconnected systems, payer complexity and exceptions that don’t follow a predictable path.
That isn’t an argument against automation. It’s an argument for looking at what happens between the automated steps.
A faster task doesn’t necessarily create a better workflow
Point automation solves real problems. If technology eliminates repetitive eligibility checks, accelerates claim-status work or reduces manual document processing, there is tangible value.
But revenue cycle doesn’t operate as a collection of independent transactions.
An automated eligibility response may still require someone to research conflicting information. A prior authorization may move efficiently until missing documentation or a payer-specific requirement sends it down a different path. A clean claim may still encounter a downstream denial.
The failure point often isn’t the automated task itself. It’s the handoff to whatever happens next.
When technology improves one transaction but the work continues through disconnected systems, queues and teams with poorly defined exception paths, an organization can optimize a task without materially improving the overall operating result.
Standardize the work before scaling the automation
This is why process design matters as much as technology selection.
Before asking whether another task can be automated, revenue cycle leaders should understand how that work is supposed to move. What information is required? Who owns the next step? What constitutes an exception? Where does that exception go?
Automation performs best when those answers are clear.
That doesn’t mean forcing every account through an identical path. Payer requirements vary. Clinical circumstances differ. Some work requires judgment that cannot, or should not, be reduced to a rule.
The goal is to standardize what can be standardized while deliberately designing for what cannot.
One rural Midwest hospital offers an example. Facing outdated workflows and limited resources, the organization combined EHR integration with standardized procedures for billing, cash and denials management as part of a broader revenue cycle transformation. Unbilled claims declined 30%, while billing time decreased from 16 days to 12.
The lesson isn’t that one particular technology produced those results. It’s that improving the way work moved across the revenue cycle created measurable operating change.
Measure the outcome, not the automation
That same principle should shape how transformation is evaluated.
The percentage of transactions automated can tell leaders something about technology adoption. It doesn’t tell them whether the revenue cycle is performing better.
The measures that matter are already familiar: turnaround time, staff capacity, cost to collect, denial rates, collections, quality and exception volume.
Those measures shift the conversation from “How much have we automated?” to a more useful question:
“Is the way we operate producing better results?”
The next stage of revenue cycle transformation isn’t simply about automating more work. It’s about designing how work moves across technology and people, including what happens when the standard path breaks down.
Because the real opportunity may not be the next automated step.
It may be what happens between them.
