Process ownership: The Overlooked Driver of AI Success
Image: Depositphotos
While artificial intelligence (AI) has the potential to be transformative, the track record to date is disappointing. Although billions have been invested in AI, recent research reveals that only 1 percent of companies surveyed consider themselves to be “mature” – i.e. to have fully integrated AI into workflows and thereby produce better business outcomes.[i] The same research report found that the biggest barrier to scaling AI is not employees—but leaders.
Simply stated, leaders are not leading in the right way. They launch too many proof-of-concept projects. In many cases the scope of AI projects is far too small. Far too often, such projects target solving small problems inside of departmental boundaries. No wonder that AI projects are difficult to scale and the return on investment (ROI) from AI is low. It has been estimated that only 22% of companies have advanced beyond the proof-of-concept stage to generate value, and only 4% are creating substantial value.[ii]
A traditional mindset by leaders, where performance is typically viewed in the context of departmental activity as opposed to value creating cross-functional workflows is a large part of the problem. Another contributing factor is that AI vendors need budget, so they cater to the whims of department heads. Accordingly, it’s not surprising that most AI projects are intended to automate narrow tasks or procedures inside departments – instead of the organization’s end-to-end value creating processes, as the latter requires cross departmental collaboration and focused change management. Data is another significant stumbling block. AI demands a great deal of good data and curating unstructured content, improving data quality, as well as integrating diverse sources is an ongoing challenge. This is exacerbated when department heads persist in hoarding data. In one Amazon Web Services (AWS) survey, 93% of respondents agreed that data strategy is critical to getting value from generative AI, but 57% had made no changes to their data thus far. [iii]Accordingly, relatively few companies can string together multiple AI projects to improve an enterprise-wide process.
Enter the process owner.
While the concept of a process owner is not new, it’s potential to drive AI implementation success has largely been overlooked. Three decades ago, it was considered a cornerstone of business process re-engineering, process management and Six Sigma in the 1990’s. The principal thought leaders in Six Sigma, reengineering and continuous improvement have all extolled the virtual of process ownership. In “The Six Sigma Way,” the authors wrote “Perhaps the most essential step in the transformation to process management is the designation of process owners.” Davenport and Short argued in 1990 that high impact processes should have owners. The late Dr. Geary Rummler, one of the earliest and arguably the most articulate proponent of process ownership, wrote that a process owner is “someone with clout” who takes action to improve the performance of an entire cross-functional process. The late Dr. Michael Hammer, considered to be the foremost vocal advocate of reengineering, wrote that “every process in a process enterprise requires a process owner – a manager responsible for ensuring that the entire process keeps flourishing.”
The role of the process owner is to monitor, improve and manage an end-to-end process by fostering collaboration – as opposed to control. It’s a difficult role – one that requires forming strong partnerships and at times can be like herding cats.
The process owner must also have a solid grasp of technology to drive AI implementation success. This includes an understanding of process mining to define and manage processes based on data from event logs and the effective use of robotic process automation (RPA). In terms of AI – this encompasses the use of generative AI, predictive AI and even agentic AI. In addition, the process owner needs to have an appreciation for the importance of business rules. In many companies, outdated rules and policies have an adverse impact on performance and drive non-value-added work.
Process ownership can significantly drive AI implementation success by ensuring accountability, clarity, and continuous improvement. It can contribute to improved performance in the following ways:
1. Align AI deployment with long-term strategy, preventing isolated implementations.
2. Ensure that AI implementation goals align with business objectives
3. Act as a bridge between technical teams and business stakeholders.
4. Establish standardized workflows for AI adoption.
5. Pay close attention to regulatory and ethical considerations.
6. Facilitate training, support, and communication to drive adoption.
7. Act as champions for AI, helping teams adapt to changes smoothly.
8. Track key performance indicators (KPIs) and business impact.
9. Drive iterative improvements based on feedback and evolving needs.
10. Ensure AI solutions deliver measurable value (e.g., cost reduction, efficiency).
Let’s consider the potential of the process owner in using AI to drive success in a new product development (NPD) process. The NPD process has not changed much in most organizations for decades with fewer than 30% of new product projects becoming commercial successes. While excellence in new product development is important to most organizations, the success rate of bringing new products to market has remained stubbornly low (e.g. one out of seven). Despite the evolution of enabling technology relatively few companies (only about 13% of firms surveyed) are using AI in NPD.[iv]
New product development in the pharmaceutical sector is particularly interesting. This sector is slow to change, highly regulated, and leadership is particularly conservative. Within the pharmaceutical sector, the oncology clinical trial process is exceptionally challenging. Bringing a new oncology medication to market still costs over a billion dollars and takes over ten years. Further, fewer than 4% of cancer drugs make it to market – which is well below the performance of other therapeutic areas.[v] While the benefits of AI in compressing the time to develop new medications have been praised, projects have targeted small parts of the process and have yet to demonstrate significant results. Similarly, while there’s much lip service dedicated to the patient experience, clinical trials are still quite burdensome on patients.
The process owner for the oncology clinical trials process would look at the end-to-end process from both the company’s point of view and that of the patient.
The process owner would use process mining software to capture information from enterprise transaction systems and provides detailed — and data-driven — information about how the NPD process is performing. Process mining examines event logs, and these logs make visible how work is happening, including who did it, how long it takes, and how it departs from the average. Then, process analytics create key performance indicators for the process, which enables a company to focus on the priority steps to improve. This technique replaces previous subjective approaches to understanding business processes.
This would lead to recognizing multiple opportunities to deploying technology to improve the end-to-end oncology clinical trial process; including, but not limited to:
· Using AI to improve clinical trial design
· Using AI to create more patient friendly informed consent forms (ICF’s)
· Deploying AI to reduce the burdensome nature of patient screening
· Employing RPA to reduce the administrative loadburden during the conduct of the trial
· Using AI to analyze and report trial progress
· Using process mining to gain a new and valuable lens into the entire process
The process owner for the oncology clinical trial process would focus on a big picture view and measure performance from both the company’s point of view as well as from the patient’s point of view. By so doing, it would become clear that the opportunities to apply AI in improving the oncology clinical trial process need to be linked. This demands an understanding of where to apply robotic process automation (RPA) and the use of both predictive AI and generative AI. This involves forging partnerships with department heads in early development, product development, diagnostics and information systems. It also requires paying close attention to the patient experience and accomplishing all of this through collaboration – as opposed to control – as work does flow across multiple departments.
The effective implementation of process ownership will not come easily. While powerful in theory, the practical application has eluded many organizations. Recall Rummler’s assertion that an effective process owner is a senior manager with clout, so assigning middle managers to the role of process owner does not work – yet far too many companies have done just that.
The impetus for process ownership must be top down. The incumbent must be a senior manager and have the respect of department heads for the role to succeed. This is particularly important with respect to the oncology clinical trial process due to the highly regulated nature of the industry, the conservative nature of leadership and the adverse impact of outdated policies.
The scope of responsibility is important and should be carefully determined – neither too small – nor too large. If the scope of responsibility of process owners is too wide – then the political challenges involved may be so great that it may not succeed. For example, the scope of process ownership for “order to delivery” is superior to “order to cash” as the former is what matters to customers and the scope is more manageable.
It is equally important to avoid having the scope of responsibility of process owners be too small and defined within department boundaries. That would produce overlap and redundancy between what departmental management and process owners do.
In theory, every process in a process enterprise should have a process owner. In practice, that can be very challenging. Appointing just one or two process owners for customer touching processes such as “order to delivery” or “inquiry to resolution” or “idea to launch” may be a better way to get started, build momentum, and take advantage of the potential for process ownership to improve AI implementation success. A customer focused, process-based view emphasizes value creation and also provides a better governance structure for AI than a vertical, department-based perspective.
Although deploying process ownership is far from easy, the payback can be significant. Leveraging process ownership can prevent fragmented AI efforts, improve operational efficiency, and maximize AI’s potential impact.
[i] Mayer, Hannah, Lareina Yee, Michael Chui, and Roger Roberts. "Superagency in the Workplace: Empowering People to Unlock AI’s Full Potential." McKinsey & Company, January 28, 2025. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work.
[ii] Boston Consulting Group. Where’s the Value in AI? November 2024. https://www.bcg.com/publications/2024/wheres-value-in-ai.
[iii] Davenport, Thomas H., and Randy Bean. "Five Key Trends in Data Science." MIT Sloan Management Review, January 9, 2024.
[iv] Cooper, Robert G. "The Artificial Intelligence Revolution in New-Product Development." IEEE Engineering Management Review, Vol. 52, No. 1, February 2024
[v] Berezow, Alex. "Clinical Trial Success Rates by Phase and Therapeutic Area." American Council on Science and Health, June 11, 2020. https://www.acsh.org/news/2020/06/11/clinical-trial-success-rates-phase-and-therapeutic-area-14857.
Andrew Spanyi is President of Spanyi International. He is a member of the Board of Advisors at the Association of Business Process Professionals and has been an instructor at the BPM Institute. He is the Editor for Cognitive World and a member of the Think Tank on enterprise AI.