McKinsey and Celonis share insights on improving end-to-end healthcare performance with AI

A gathering storm or a golden age? This is the question the healthcare industry is reckoning with in the face of unexpected market change and macroeconomic forces.

Whether it’s regulatory pressures, supply chain strains, expanded tariffs, rising workforce costs, or demanding consumer expectations – healthcare delivery challenges seem to keep coming and putting organizations’ adaptability to the test.

But even storm clouds have silver linings. With these pressures compelling healthcare organizations to innovate more rapidly and embrace the technological advancements of AI, the automation age finally appears possible. That brings new drug therapies, greater productivity, and a radical rethink of the patient experience to deliver a better standard of care.

So what positive change can organizations achieve in business and patient outcomes when operational transformation meets healthcare innovation? McKinsey & Company partner Jung Paik and Celonis’ Chairman for North America Mike Kaufmann discussed the role of process intelligence (PI) in unlocking healthcare services and AI value on a recent webinar.

While the webinar focused on healthcare, this is just one example of the major industries where Celonis has achieved significant results. Here are some of the insights, real-world use cases, and actionable strategies shared on how healthcare providers can drive measurable impact.

The potential of AI in healthcare delivery

Paik set the scene by explaining how, although automation and AI adoption in healthcare has historically lagged behind other industries, according to McKinsey research healthcare is now one of the top three industries with AI agent use at scale.

There are many examples of AI being used in healthcare innovation to generate positive ROI, the first of which is research and development (R&D). McKinsey studies find healthcare organizations are using AI to reduce the cost of drug R&D by 30-50%, double the speed of deployment, and increase the probability of drug efficacy by 4-10 pp.

Then there are payer-driven and provider-driven AI improvements, including:

• 13-25% savings on admin costs
• 5-11% savings on medical costs
• 3-12% boost to revenue
• 11-17% of Net Patient Service Revenue (NPSR) contribution margin impact

Together, the projected net savings of using AI, traditional machine learning, and deep learning are up to $360 billion in healthcare spending.

But healthcare innovation with AI encompasses much more than use cases generating financial value. Automation and data infrastructure investments can also generate value for patient care by streamlining communications, simplifying scheduling, enhancing diagnostic equipment, and triaging patient needs. And AI is finally putting a patient-centered healthcare system within reach by holistically analyzing social care data, patient-generated clinical data, provider-generated data, financial data, and wellness data.

To realize this abundant potential, however, healthcare organizations need to combine AI with process intelligence (PI).

There’s no AI without PI in healthcare

Process intelligence is the missing piece AI needs to know how healthcare businesses flow.

Healthcare organizations are applying agentic AI and automation tools to improve care delivery and patient engagement. But they often rely on informing these AI applications with a subjective impression of how business processes run. Instead, they need to give AI a true, data-based, enterprise-wide view of how work flows.

The Celonis Process Intelligence Platform constructs a living digital twin of your operations from your system data, enriched with your unique business context (including KPIs, benchmarks, and enterprise architecture). This provides the end-to-end visibility you need to transform your healthcare organization, and the operational context AI needs to make better decisions and deliver better outcomes.

Kaufmann broke down the benefits of process intelligence into three value categories, which apply to healthcare and just about every other industry too.

Category 1: Direct savings

Process intelligence can surface the root cause of a problem and automatically orchestrate corrective action. For example, it can discover costly inefficiencies like duplicate payments, shipped-not-billed orders, and unnecessary credit blocks. And by connecting end-to-end systems, process intelligence can trigger action to fix these issues in real time and drop savings directly into your bottom line.

In other industries, automotive companies like Mercedes-Benz have used Celonis Process Intelligence to pinpoint bottlenecks in aftersales processes, so they can improve the customer experience by more efficiently and proactively sourcing spare parts.

Category 2: Insights

Next there’s the value of process intelligence revealing how healthcare providers’ processes actually work. These insights help uncover opportunities to improve operations, customer satisfaction, revenue, and margins.

Hospitals use process intelligence to look at emergency room wait times, for example, finding bottlenecks they can remove to improve the patient experience as well as companies’ finances. Other healthcare organizations use it in their supply chain to see orders at risk of late delivery, due to Procurement delays and availability, so the company can intervene by offering alternatives, boosting customer satisfaction.

Another example is  using AI-powered process intelligence to perform sentiment analysis on customer correspondence, finding the most common reasons for complaints and enquiries which the organization can then mitigate.

Category 3: AI-readiness

Companies’ data is typically not in a form that AI can understand and use. Process intelligence connects, curates, and cleans structured and unstructured data across disparate systems (from spreadsheets and PDFs to ERPs and CRMs), so you can deploy and monitor AI with confidence.

And once deployed, healthcare providers can use process intelligence to continuously monitor AI performance, ensuring they’re only getting the right outputs. Vinmar, a global distributor of plastics and chemicals, shows how that’s done by implementing process intelligence as an intelligence layer that gives AI the context it needs, ensures it’s deployed in the right places, and orchestrates it with all their other solutions and systems.

Best practices for healthcare organizations implementing AI with process intelligence

Kaufmann walked webinar attendees through what he learned as CEO at Cardinal Health, where he achieved millions of dollars using process intelligence insights.

His first piece of advice is that any process intelligence project should be business-led instead of being delegated to the CIO. When cost savings come from correcting duplicate invoices and credit hold issues, they don’t typically end up in IT, so the wider business reaps the rewards while IT gets charged the licence fee. COO, CFO, or CEO ownership minimizes this negative push and pull in the organization.

Another learning is that too many businesses jump into a process intelligence project without setting clear value goals (such as an improvement in patient outcomes, site effectiveness, or costs) and metrics to measure project success against. Lacking a clear business case is one reason behind the infamous MIT finding that only 5% of enterprises have seen a positive, measurable ROI from AI. Healthcare companies need to define the value they’re trying to unlock, identify the priority processes, and determine how they’ll quantify outcomes. Otherwise they can inadvertently pursue limited AI use cases like chatbots that don’t deliver impact where it matters.

Maximizing value from AI and PI initiatives also depends on senior leaders staying highly engaged, and organizing regular check-ins with the project team about the use cases they’re working on.  Without the leadership and championing of an executive sponsor, ROI can become trapped in one department rather than expanding enterprise-wide. The leader should be purposeful about digital transformation projects targeting all three of the value categories mentioned above, to avoid missing out on indirect savings and AI-readiness benefits.

As Kaufman summarized, process intelligence can make your existing processes work harder, but it can also help you to redesign processes for better outcomes. With AI and PI working together in your enterprise, you can clear the gathering storm clouds and thrive in a bright new dawn of optimized healthcare.

Catch up on the full on-demand webinar to learn more tips for successful AI adoption in healthcare, including six insights that have guided organizations through their AI journeys over the past year.