The Full ROI Chain: Beyond Time Savings

Sasanka (Sy) Yellamanchali is the CEO of DeliverHealth, with a background in operations and product management and three years leading the company's push into AI documentation and coding.

When health system CFOs sit down to evaluate AI documentation and coding tools, the first question on their list is almost always the same: how much time does this save the physician per encounter?

It is a fair starting point. Physician time is one of the most expensive resources in any health system, and a tool that reduces documentation burden by 15 minutes per encounter generates real value in multiple ways. No one is disputing that.

But if 15 minutes is where most ROI conversations stop, stopping there leaves most of the value unaccounted for.

In conversations with health system leaders over the past year, I have started asking two follow-up questions after the time savings number comes up: then what? And so what? Those questions tend to land most of the time quietly, but they point to something most organizations are genuinely missing in how they evaluate these tools.

The chain, and where it breaks

A physician encounter produces a clinical note. That note is the starting condition for everything that follows in the revenue cycle. When the note is complete and accurate, the downstream processes operate with minimal friction. When it is incomplete or abbreviated, every step that follows absorbs the cost.

Complete documentation feeds first-pass coding accuracy. A thorough note contains the clinical specificity that coding engines need to assign the right codes on the first attempt. Accurate coding produces cleaner claims. Cleaner claims generate fewer denials. Fewer denials mean the payment cycle tightens, and revenue that was otherwise tied up in the system starts to move.

That chain runs from the encounter all the way to paid claim, and time savings is only the first link. An organization that measures the first link and calls it the ROI calculation has measured an input and reported it as an outcome.

Why this gap exists

There is a structural reason this gets missed, and it is not a failure of attention or intelligence. Health systems are organized in ways that make it genuinely difficult to track value across the full chain from any single vantage point.

The VP of physician operations is accountable for physician productivity. Documentation time saved shows up directly in her metrics. But the denial rate and payment velocity improvements that flow from better documentation land in revenue cycle, under a different leader with different reporting lines and different quarterly targets. These two leaders may never look at the same dashboard.

As a result, AI documentation tools often get evaluated on the productivity metric alone, by the person whose job is productivity. The downstream revenue cycle impact is invisible to the evaluator — not because it is not real, but because it belongs to someone else’s budget and someone else’s review cycle.

This fragmentation is a structural reality of healthcare administration at scale. It is not going away. But it does mean that health systems are consistently undervaluing tools that have substantial financial impact well beyond the encounter itself.

Who sees it clearly

Mid-tier physician groups and independent practices tend to see the full picture with less difficulty. Fewer organizational layers stand between the encounter and the payment. In many cases, the same person who watches physician productivity also watches denial rates and cash flow. The chain is visible end to end because the organization is small enough to see it whole.

Some of the most complete ROI conversations I have had were with leaders at group practices. They have already done the math on documentation completeness, denial reduction, and payment velocity together. They are not asking how many minutes their physicians get back. They are asking how much revenue is currently sitting uncollected because the documentation chain broke somewhere before the claim went out.

That is the question larger health systems need to be asking more consistently.

What the right evaluation looks like

A complete ROI calculation for AI documentation and coding should account for more than physician minutes. Documentation completeness feeds first-pass coding accuracy, which directly shapes denial rates. Reduced denials cut the staff cost of working rejected claims and accelerate payment timelines. Undercoding recovery, which becomes possible when documentation captures the full clinical picture of the encounter, represents additional recoverable revenue that most time-savings evaluations never surface.

When those factors are quantified together, the financial case is substantially larger than the time savings argument alone. In most health systems, the denied claim recovery opportunity and the undercoding recovery combined exceed the time savings figure by a significant margin. The number most CFOs are using to evaluate these tools is the smallest number in the actual return.

The organizations getting the full return are the ones that brought their physician operations leaders and their revenue cycle leaders into the same evaluation conversation. The tools look very different when both ends of the chain are visible at once.

The chain is becoming concurrent

The framework I just described reflects how most health systems still operate. It is also already becoming an incomplete picture.

McKinsey published research this year projecting that agentic AI could reduce cost to collect by 30 to 60 percent. Potentially a wide range; however, that number does not come from making the existing sequential process faster. It comes from the model changing entirely.

If one thinks about how they book a flight today online. In many cases, AI has already searched options, compared prices, and is ready to complete the purchase before you have opened a single airline website. The agent acted before you navigated anywhere. The same shift is beginning to happen in healthcare revenue cycle. AI is starting to act on the clinical note, initiating prior authorization, referral workflows, and coding, before a human has moved to the next screen. The note can trigger multiple things at once rather than handing off to one process at a time.

Most health systems are not there yet. But the ones building towards it now are the ones that will be positioned to capture the full return when it arrives. The organizations building for what comes after that, concurrent agent-driven workflows firing from a single encounter, will find a return that no time-savings calculation was ever going to surface.

A different starting question

Physician time savings matters. It is real, measurable, and improves physician satisfaction in ways that have long-term retention and recruitment value. None of that changes.

But it is the opening data point in a longer financial story, not the conclusion. The organizations making the most of AI documentation and coding are the ones that started their evaluation by asking what happens after the encounter and kept following the chain all the way through to cash.

At DeliverHealth, InstaNote and InstaCode are designed around the full chain. InstaNote captures the documentation foundation. It gives every downstream process something accurate to work with. InstaCode carries that into coding, denial reduction, and payment velocity. The direction we are building toward is the agentic layer, where multiple workflows initiate from the clinical note simultaneously, without human coordination for each and every transition. The ROI in that model cannot be captured by a linear evaluation framework. When agents are working multiple parts of the revenue cycle at the same time from the same encounter, the return accumulates across dimensions that most CFO dashboards were not originally built to see.

Frequently Asked Questions (FAQs)

Q: What should health systems actually measure when evaluating AI documentation ROI?
Start with time savings, but do not stop there. The more complete picture includes first-pass coding accuracy, denial rate reduction, the staff cost of working rejected claims, payment velocity, and undercoding recovery. These downstream factors typically represent a larger financial return than the physician productivity gain alone. The organizations that measure across the full chain make better purchasing decisions and capture more of the return after implementation.

Q: What is agentic AI and what does it mean for revenue cycle?
Agentic AI refers to systems that execute complex processes autonomously rather than surfacing recommendations for a human to review and act on. In revenue cycle, that means AI initiating prior authorization, referral workflows, and coding simultaneously from the same clinical encounter, without waiting for a human to move the process forward at each step. McKinsey projects this shift could reduce cost to collect by 30 to 60 percent — a number that is not achievable by making the existing sequential process faster.

Q: How do InstaNote and InstaCode work together across the revenue cycle?
InstaNote captures clinical documentation at the point of care, giving every downstream process something complete and accurate to work from. InstaCode takes that into autonomous coding, denial reduction, and payment workflows. The value of the two together is that the documentation foundation InstaNote builds directly improves the coding accuracy and denial outcomes InstaCode drives. The return is in how they connect across the full chain, from the clinical encounter to the paid claim.