The pressure to prove a return on AI investment is nearly universal across health systems. What isn’t universal — not yet — is the infrastructure to actually measure it.
Our 2026 CIO Research Report found that four out of five health system technology leaders say they struggle to determine or measure the ROI of their AI investments. That statistic sits alongside another: 65% of respondents rated the pressure to operationalize AI at 7 or higher on a 10-point scale. The gap between those two numbers is where most health systems are stuck.
Why ROI is harder to measure than it looks
The challenge isn’t only technical. It’s also definitional. When we asked leaders what they’re tracking, we found four distinct categories:
- Revenue generation (cited by 62%)
- Hard dollar cost savings (59%)
- Patient outcome improvement (49%)
- Staff productivity gains (49%).
These measures don’t always point in the same direction, and they don’t always resolve on the same timeline.
“If we can’t prove that it’s going to help a patient or a clinician, it’s probably a non-starter,” said Matthew Anderson, MD, CMIO at HonorHealth. “You have to find both hard and soft ROIs.”
What makes this harder is the runway question. 38% of respondents say they don’t expect to see ROI for 13 months or more. And yet, 74% say they need to see it within one year to justify the investment. Those timelines are structurally misaligned, which means many AI deployments are being evaluated against benchmarks they were never positioned to meet.
No benchmarks, no scaling
39% of respondents have no clear process for benchmarking AI performance or ROI at all. Without benchmarks, scaling a pilot beyond a limited deployment becomes nearly impossible to justify internally to the CFO, to the board, and to the clinical staff being asked to change their workflows.
The leaders making the most progress have resolved this problem by defining success criteria before any contract is signed. Morgan Jeffries, MD, Medical Director for AI at Geisinger, described the framework his team uses: “How to choose use cases: is it objective? How likely would we be to achieve it? Financial impact and short time to value carry a lot of weight.”
That kind of upfront discipline is increasingly becoming a differentiator, and a requirement. 59% of leaders in our survey said guaranteed ROI or a risk-sharing pricing model is a preference in vendor selection, with many saying they now treat partners who won’t share outcome risk as effectively disqualified.
Joseph Sanford, MD, Associate Vice Chancellor and Chief Clinical Informatics Officer at the University of Arkansas for Medical Sciences (UAMS), described the measurement framework his team applies: time savings for clinicians, reduction in after-hours charting, administrative throughput efficiency, and per-user licensing costs weighed against revenue generation and expense reduction. That specificity is what makes the ROI defensible and gives the health system the evidence it needs to scale.
The path forward
The health systems closing the gap on ROI measurement share a few traits: they define metrics at the point of vendor selection, they include clinicians in setting success benchmarks, and they treat measurement as infrastructure rather than an afterthought.
Our report details what the top-performing health systems are doing differently, and lists five concrete actions every CIO should take in 2026 to close the measurement gap. Read the full report.
