Healthcare organizations have made genuine progress with AI. But when it comes to scaling, many leaders need better ways to prove AI's return on investment.
A new report, From Early AI Wins to Real Adoption: What Healthcare Leaders Are Learning, found that 54% of healthcare executives expect measurable ROI within 6 to 12 months. That's a tight window — and it may require organizations to look beyond efficiency gains alone to understand AI's full impact.
To explore where that impact may be emerging, Healthcare Dive spoke with Rizwan Pasha, M.D., Chief Medical Information Officer at Microsoft, about how organizations can connect clinical improvements to financial impact.
Documentation is the starting point, but its value goes beyond efficiency
Documentation is the top priority for healthcare leaders, with 65% naming it as one of the most urgent challenges to address with AI. But look beneath the surface, and another priority comes into focus: coding accuracy and improved reimbursement.
This is not a coincidence. Documentation and coding are intrinsically linked. More complete documentation can support more accurate coding, fewer denied claims and faster reimbursement.
The data reinforces that connection. In the report, 48% of healthcare leaders say coding accuracy and improved reimbursement justify scaling AI. And 54% report fewer claim denials when using AI tools.
For organizations that connect those outcomes, AI becomes not just an operational investment, but a clear financial opportunity.
Pasha sees this play out directly in practice:
"AI is also really good at looking at the evidence behind each of the diagnoses and suggesting more appropriate diagnoses or matching them with health care codes. That becomes a significant impact on the organization, in terms of patient care, time saved, physician burnout and capturing revenue that would have normally been missed through incomplete or inaccurate coding.”
What's at stake in every encounter
In healthcare, a wrong diagnosis or documentation gap carries real consequences for patients, clinicians and organizations alike. The way it shows up — and the revenue it puts at risk — looks different depending on the care setting.
"In an ED or inpatient visit, a lot is happening at once and spanning many hours. When a physician or nurse cares for a complicated patient, they'll often document afterward, sometimes hours later or at the end of a shift. It's very hard to recall everything, and naturally, things are missed," noted Pasha. “When a diagnosis isn't recalled and documented, it can't be coded, and if it can't be coded, it can't be billed.”
Ambulatory care poses a similar but unique challenge. A single visit might cover several issues, increasing the risk that relevant diagnoses, procedures or time-based elements are not fully documented and therefore not reflected in coding and reimbursement.
"In ambulatory, billing is based on time and any procedures performed, like an EKG, each of which should be a separate billable event. So, if the visit is captured more thoroughly and accurately, that directly leads to revenue capture," said Pasha.
Another area where the revenue impact becomes especially concrete is in managed care and Medicare Advantage populations. In these models, organizations are compensated based on how sick their patients actually are, and that complexity has to be documented accurately to be reflected in payment. When it isn't, organizations are effectively paid as if their patients are healthier than they are, leaving significant reimbursement on the table.
Across care settings, documentation gaps can quietly translate into significant lost revenue over time — creating additional strain for organizations already balancing tight margins, staffing pressures and growing demands on patient care.
From documentation wins to revenue gains
For organizations asking whether the revenue impact is measurable, Pasha says many are beginning to track indicators such as additional patients seen, more complete coding and specificity in documentation.
"Organizations are capturing the number of additional patients they're seeing, the additional Evaluation and Management (EM) codes they can bill for and the specificity of the diagnoses being captured," he said. "We're seeing better documentation that's more specific and matched to the patient's actual visit — increased complexity levels, improved revenue and better capture of HCCs."
Cooper University Health Care is a good example of how this trajectory plays out. A case study about their experience with Microsoft Dragon Copilot notes that they began with clear goals around documentation and clinician burnout. Clinicians are now saving more than four minutes per patient on documentation time, and notes are more comprehensive. With that foundation in place, Cooper is now looking to scale and capture ROI in areas like improved revenue cycle management.
These types of time savings can add up quickly across high-volume clinical workflows. Pasha noted:
"Four minutes may not sound dramatic on its own, but when those savings are multiplied across dozens of patient visits a day, they add up quickly. Over time, that can translate into meaningful reductions in documentation burden and more time focused on patient care."
Where healthcare AI value is becoming clearer
As healthcare organizations move from early experimentation to broader AI adoption, many are looking for outcomes they can measure and defend. For some, documentation improvement provides a clearer path —not just because it reduces burden for clinicians, but because it leads to financial impact.
For a broader view of what healthcare leaders are learning about adoption, ROI and scale, download the full report, From Early AI Wins to Real Adoption: What Healthcare Leaders Are Learning.
