Hospitals Are Investing in AI — How Can They Evaluate ROI?

Will artificial intelligence (AI) make or break hospital finances? Excitement around the promise of AI has attracted billions of investment dollars, but it remains difficult for hospitals to predict what value they can expect to see from these technologies in the future. 

A recent McKinsey survey found that while about half of health system leaders anticipate a return on investment (ROI) from AI, only 17 percent were currently able to measure a positive return. With so many hospitals operating on negative or razor-thin margins, they don’t have the flexibility to gamble their resources on tools that may or may not support long-term growth. 

Historically, hospitals have experienced innovation passively rather than driving it actively — seen in cases like mandated EHR implementation and consumer-driven digital front doors. To avoid history repeating itself, healthcare leaders need to seize the opportunities AI can provide, including operational efficiency and profitability. 

Hospitals’ AI ROI challenge

New technologies have historically had a mixed record of delivering on their promise – evoking healthy skepticism from hospital executives. The traditional healthcare technology business model has improved software companies’ bottom lines, but health systems — who rely on these technologies for everything from radiology imaging to patient claims billing — haven’t always seen the same benefits. As they continue to operate in a competitive market with median margins dipping below one percent, productive and accountable AI investments are increasingly vital to their economic survival and ability to serve patients. AI has to work for health systems, but the decision-makers need confidence that they’re making the right choices.

The combination of tight margins and a struggle for even the most well–resourced hospitals to stay afloat can force leaders into hasty AI investments. Anxious, some grab onto the first solution that seems like it could boost efficiencies, often without sufficient ROI planning. Others wait for a magic box to appear and solve their problems. Both can threaten hospitals’ future solvency and capacity to deliver care, and neither is a strategy.

To make AI work for their hospital, leaders need a plan — a clear approach to measuring value, selecting tools, and scaling what works.

Begin with the benefits

Thinking about ROI should start with appreciating all of the potential benefits. Many people think exclusively of AI-assisted automation and cost savings, but that’s only part of the picture: AI can also help teams do better work — enhancing accuracy, expanding scope, and enabling smarter decisions. 

Take clinical documentation integrity (CDI) as an example. Even the most experienced CDI teams may only be routinely catching the most common 40 percent of missed codes — like sepsis and respiratory failure. But what about the long tail — the other 60 percent? AI can help uncover less frequent but high-impact diagnosis codes, significantly improving capture and revenue without replacing the human team. Bottom line: AI isn’t just about doing the same work for less. It’s also about doing better work — smarter, faster, and more comprehensively.

And even if you have big ambitions, narrow your scope to start with a small win. Hospitals should start using AI with a relatively small number of people – with one department, say – then expand once they’re comfortable with how to select and operationalize AI. When those responsible for AI strategy start small, then demonstrate positive ROI, they’ll earn political capital that can be leveraged for further investment. 

Align on success metrics with analytics teams 

Many hospitals lack the internal analytics capabilities to measure ROI on their own, so they rely on AI vendors to do so. As the former medical director for transformation at a major hospital charged with overseeing applied AI and in my current role leading a clinical AI company, I’ve seen this challenge from both the hospital and vendor side of the table. Before a hospital selects a vendor, they should align with their analytics teams and vendors on how to attribute value and what success looks like. 

This is a crucial step: You don’t want poorly-conceived metrics to work against your AI investment. If a vendor reports that efficiency has improved by 80 percent by measuring only cost, but you still have 100 people doing a task that could be done by 20, you may feel you haven’t actually improved efficiency. Define how value will be attributed upfront and get as specific as you can. Make sure your vendor is on board — and held accountable.

Help AI vendors help you

Vendors want to understand the North Star metric they’re optimizing for, but they may need help getting there. This is a common problem for hospitals. To solve it, hospitals and vendors should walk through the steps they’d take to solve their challenges together. If a vendor can’t explain clearly how their product will deliver value in your context, that’s a red flag. And if they can, give them the data and context they need to succeed. ROI is a shared responsibility.

Build the right team for AI success

AI success doesn’t just depend on technology. It depends on the people choosing, piloting, and championing it. An AI task force of just a few highly capable leaders, backed by a strong internal analytics team, can help a hospital system make smart bets by working closely with technology partners and internal stakeholders to evaluate and validate AI tools. 

You don’t need a huge committee. Just a few strong, curious, analytically-minded people can make a big difference.

Learn by experience to plan for the future

AI adoption in hospitals is on an exponential curve as confidence in its performance grows. Capabilities that once seemed futuristic — like human-level understanding of clinical documentation — are now commercially viable with powerful large language models. If hospitals don’t start learning from AI now, they not only risk falling behind. They also risk missing out on the upside AI can provide. 

AI is already proving its value in the healthcare back office. Take revenue cycle management: AI can now perform second-level reviews of every patient chart before billing — an application that boosts efficiency while generating a 5:1 return on investment. That’s not future potential. That’s real performance, today.

AI doesn’t need to be a black box — and hospitals don’t need to invest based on blind faith. With the right structure, questions, and metrics in place, healthcare leaders can cut through the hype and make decisions that actually drive value.

In a financial landscape where every investment counts, AI can’t just be promising — it has to be productive. And for forward-looking hospitals, it already is.

Picture: Warchi, Getty Images

This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.