Why Marathon Petroleum Started with Maintenance, Not AI

Most transformation initiatives begin with technology. Marathon Petroleum took a different approach. 

Rather than looking for opportunities to deploy AI, the company focused on understanding where value was being lost across the business. That search led to reliability, maintenance and the challenge of creating consistent processes across a diverse operational footprint. 

In this interview, Daniel Byrne, former Senior Director of Digital Transformation at Marathon Petroleum, discusses why successful transformation starts with business problems, not technology and how a focus on maintenance standardization, employee experience and process redesign helped create the foundations for AI at scale.

Lily Mae Pacey: Can you tell us about your role and how it connects to Marathon Petroleum's AI and digital transformation strategy? 

Daniel Byrne: I'm part of the Digital Transformation Group. We sit between the business and IT. Our role is to look across the business and identify opportunities that can deliver significant value and drive meaningful transformation. 

We are not using AI as a solution in search of a problem. We aren't even using technology in search of a problem. What we're looking at is what is meaningful to Marathon and what would be transformational and drive significant value. We start to redesign that, start to reimagine that, and then we start looking at technology. 

As we identify pain points and opportunities, we assess where technologies such as AI can add value. From there, each opportunity is evaluated through a cost-benefit lens. AI is one of several technologies we explored as part of a broader effort to solve business challenges and create value. 

We are not using AI as a solution in search of a problem. We aren't even using technology in search of a problem.

As Marathon Petroleum assessed opportunities across the business, maintenance emerged as one of the most significant value creation areas. The journey didn't begin with maintenance. It began with reliability.  

Lily Mae Pacey: Why did maintenance emerge as a major transformation opportunity? 

Daniel Byrne: Reliability was a big piece of it and, contributing to that, maintenance is one of our biggest areas of spend. When we looked at maintenance, we found that we didn't have a common process shared across all three parts of the business. 

So, when we started looking at how to unlock value, the question became: how do we build a system that can meet the various needs of the business, but also scale and work across all our operations? Through that work, we've identified significant value through a combination of improving reliability, reducing lost opportunities and managing costs across all of that. 

Discover how Marathon Petroleum is advancing AI adoption, workforce evolution and operational excellence to build a more agile enterprise: How Marathon Petroleum Is Advancing AI, Workforce Evolution and Operational Excellence

Lily Mae Pacey: What was the biggest surprise during that transformation work? 

Daniel Byrne: It's not actually the performance of the work itself. It's not the execution or the 'swinging of wrenches', so to speak. It's in the planning, the scheduling and the materials management. What we found was that a lot of the value sits where those areas intersect, particularly around materials management. 

Lily Mae Pacey: Why does standardization need to come before AI? 

Daniel Byrne: When we looked at maintenance, we found that we didn't have a common process shared across all three parts of the business. You can look at a refinery, which is much more sophisticated and has dedicated maintenance teams, and then compare that to other parts of the business where sites may be unmanned, have only two or three people, or rely on a regional workforce. 

Explore another perspective from Operational Excellence in Oil & Gas on why leadership plays a bigger role in transformation success than age or generational labels: The Generational Divide Is a Leadership Problem, Not Age

So, when we started looking at how to unlock value, the question became: how do we build a system that can meet those different needs while still scaling and working across the entire organization? That's what we were working to build, a system that could scale and work across the organization. 
 
Lily Mae Pacey: Many AI initiatives fail to deliver value. What guardrails have you put in place to avoid that?  

Daniel Byrne: We're starting with the process, the experience and the journey in mind. What we're doing is looking at where, in that process or experience, whether it's with a customer, a user or an employee, technology can help us.  

That's where we employ that crawl, walk, run approach. Let's build something basic and see how people use it. How do they react? Is it creating value? Are we seeing the metrics? It allows you to experiment, and it's okay to fail because you haven't spent a lot of money. Like many industrial organizations, Marathon Petroleum views data quality as one of its biggest challenges. But according to Daniel, the answer isn't waiting for perfect data. 

Lily Mae Pacey: Data quality is often cited as a barrier to AI. How are you approaching that challenge?  

Daniel Byrne: Data is probably one of our biggest challenges. Throughout my career, even before AI really started to catch on, data was usually one of our biggest challenges. We can get the data, but we still need people in the business to tell us whether it's correct and what it means. A lot of times we say, 'We'd love AI to do this.' Then we realise we don't have all the data. So, the next question becomes: how do we get that data? Well, let's build a process that automatically captures it. Once we do that, we can start training AI on it. 

We're looking at the non-value-added tasks and asking: how do we automate those for our employees? 

Beyond technology and data, Marathon Petroleum recognizes that successful transformation ultimately depends on people. 
 
Lily Mae Pacey: How are you driving adoption and building trust among frontline employees? 
 
Daniel Byrne: It's things like hunting down parts, searching for old maintenance plans or manually building work plans. Trying to build a work plan or track down an old one is very tedious work. What we're focused on, and I've mentioned this before, is the employee experience and the employee journey. We're looking at the non-value-added tasks and asking: how do we automate those for our employees? How do we put the most important information in front of them so they can use their knowledge and expertise and really lean into it, instead of getting bogged down in administrative work? 

 Reflecting on his experience at Marathon Petroleum, Byrne emphasized that organizations should focus on business value before selecting technology. 

Interested in learning more? Daniel will be sharing additional insights on reliability, maintenance transformation and the operational foundations required for AI at the upcoming Operational Excellence in Oil & Gas Summit, Novemeber 3-5, Houston, TX.