John Deere successfully migrated its manufacturing landscapes from SAP ECC to SAP S/4HANA in one of the largest transformations of its kind. With an 18TB landscape supporting over 50 factories and 175 warehouses across five continents, the project was completed on time, under budget, and with minimal disruption to operations.
ASUG recently spoke with Jeff Thiele, John Deere’s Global Digital Product Manager for Manufacturing Operations, about this transformation and other innovations he and his team have led at the international manufacturing company, including directing operations through a massive technology modernization initiative.
This interview has been edited and condensed.
ASUG: What were the most significant business challenges and priorities you addressed when migrating to SAP S/4HANA?
Thiele: Four years ago, when I moved into this role, we learned that SAP ECC was going out of support by the end of 2027. This prompted us to ask: ‘What are our plans for migrating to SAP S/4HANA?’ At the time, there had only been preliminary reviews of the value proposition and what it would take to get that done. We realized the journey could take three to four years in some cases, leaving us with limited time to work while maintaining adequate coverage.
SAP S/4HANA is no little lift from SAP ECC. As a Fortune 100 company— recently No. 64 on the Fortune 500 list––with 56 factories worldwide, running those factories on an out-of-support ERP system would have been irresponsible, so we wanted to get those migrations done ahead of the deadline.
Our goal was to make that conversion, minimize the disruption to our operations, and allow factories to continue production. We needed to get this migration into a weekend, which was really a challenge. We have three different landscapes: the first one was Asia, the second was the Americas, which includes 35 factories—many of our most profitable—and then we concluded with Europe.
The technical migration was goal one with minimal disruption. Therefore, to do that, we changed as little as possible. We tried to mitigate the risk, so we did not do a lot of convergence or harmonization of processes during this phase—work that still needs to be addressed. We did not try to move to SAP Fiori front end in this phase, and we did not try to introduce new technologies, both of which remain on our to-do list. We focused solely on the changes that were absolutely necessary to make the technical migration work.
This migration included an 18 TB system that runs the Americas. At the time, I believe it was the third largest system to be migrated to SAP S/4HANA. We did not use a systems integrator to get this done, which also heightens a bit of potential risk, but it saves a ton of cost and provides a lot of control.
ASUG: Congratulations on your impressive migration to SAP S/4HANA. How is AI playing a role in your strategy?
Thiele: We use things like GitHub to help our developers code – even develop code in SAP’s ABAP [Advanced Business Application Programming] language. Our manufacturing operations have about 140 developers who are all ABAP capable. We try to do as few customizations as possible, but when the teams get stuck, they can use GitHub to enhance their capacity. We have already seen a significant increase in lines of code per hour or code per week with the use of GitHub, which functions as a copilot and assistant.
In addition, we use AI to help classify large amounts of data, such as business partner conversion (BPC) errors and other large data sets. AI was very useful here. AI is excellent at classifying data into buckets, which can speed things up. It is also very good at getting insights and can begin performing the work after observing a task a few times. In many ways, it can be an automation engine.
At times, our data transformation partner used AI to help us process a lot of data to get from SAP ECC to SAP S/4HANA. AI is getting very good at data work, particularly if your data is poorly structured. We were spinning off new data, so it was not always well-structured as we started to find errors.
ASUG: What other types of use cases have you developed through this process?
Thiele: We have some unique use cases. We have our own captive large language model (LLM) that sits behind our firewall. The LLM knows John Deere, and we can run it in our context.
We are also using AI to review plan and production orders to determine which items to cut using a laser. We use lasers to cut sheet steel and dynamically nest through a third-party solution designed for this purpose. We are running AI across the top of the system to create a more efficient collection of parts, which helps reduce scrap. It started at a 2% reduction, and it is growing toward 10%—a 10% skeleton reduction for John Deere that could be worth millions a year. We cut a lot of steel.
We also have an AI Machine Learning algorithm that forecasts indirect material consumption – things not in your bill of materials like eye protection glasses, gloves, or weld tips. By looking at many years of data, it can do a nice job of analyzing what’s coming forward and predicting what’s needed and when. It is doing a fantastic job.
The AI algorithm can also determine when supplies are scheduled to arrive and create supplier agreements so you can just click and send. This is very efficient. We can have better fulfillment, fewer stock outages, and lower inventories as we are running this – the Holy Grail of inventory management.
For a generally conservative company, John Deere has been progressive in releasing AI and large language models internally where we can control—or even when the AI does act out or do something unpredictable—we can monitor and manage. We have been learning, but we have been learning fast.
In manufacturing operations, we have been using this captive LLM behind our firewall for about three years. That means that any insights we are getting are not getting out into the broader ecosystem. And it is allowing it to pick up John Deere’s proprietary data behind our firewall to give us better capabilities and insights—a really powerful method, in my opinion.
ASUG: From a human perspective, this is about building skills and making work exciting when team members get their hands on these use cases and begin developing them. Has that been a factor?
Thiele: It has. We tried to imagine AI use cases for manufacturing, and we tried a dozen of which maybe four have yielded value. Once you see four, you can well imagine 40 and once you see 40, you can easily know 400. So, you start to figure out what AI is good at and what it is not.
This is the same for me and my peers as we are using AI engines. I can have it rewrite emails for me, but my peers have had it do other things. To see what they have been able to do informs me. There is a collective on this. If you want to be one person in a silo in a corner, trying to figure out all things AI, it will take you forever. If you socialize where people are having successes, you can accelerate that I would guess by 100X in leveraging AI capabilities.
ASUG: After a successful implementation with minimal disruptions, what plans do you have for your team in 2025?
Thiele: Although we were under budget with our S/4 implementation, we spent significant money implementing it. And we did not do it for fun. We did it for the value it can bring to the company. Now, it is time to bring that value.
We are implementing new modules like SAP Extended Warehouse Management, SAP Service and Asset Manager, Resource Scheduling and Asset Performance Management, Material Requirements Planning Live, and SAP Production Planning and Detailed Scheduling. And there are the SAP Fiori applications and SAP Work Zone Advance, Robotic Process Automation (RPA) and Business Process Automation (BPA) engines, and SAP Build Apps. These are all opportunities that fit our business in some way. We are implementing some of those, we are starting to roll them out and bring those new capabilities and value to our businesses. And there are other new capabilities we are experimenting with to see if or where they fit into our ecosystem and if they would be good solutions.
Regardless, we have a contractual obligation. We wrote what is called an Authorization for Expenditure (AFE), and it has a return on investment. We’ve spent the money; now, we need to prove the savings and provide the documented value that we committed to bring to the enterprise. I have little doubt we will crush our ROI percentage because I believe there is significantly more value out there than we realized when we created the AFE.
It is now delivery time. We will be highly focused on delivering value this year and the next few years.
We are also going to take the opportunity to converge processes. We have a lot of needless variation around the globe, sometimes factory-to-factory, sometimes region-to-region, and as we move to these new capabilities, we are going to build it one way and converge much of that in our processes.
ASUG: You do not get work done across the globe under budget without alignment with stakeholders. What advice can you give others as they try to launch large implementations?
Thiele: We did some good work and had some good fortune. There were some things we walked into by happenstance that are really paying back now. One of them is SAP S/4HANA – SAP S/4HANA is very capable, and S/4 rolls off the tongue – it is easy. Everything we are talking about falls under our S/4 initiative—people have heard it, and they remember it. This S/4 tagline is easy to remember and build upon, both for the technical conversion and now for the value and convergence. It is all part of this SAP S/4HANA initiative. This has been very helpful.
Across our three landscapes, we engaged over 1,700 business testers. You are going to have issues as you do a tech conversion from SAP ECC to SAP S/4HANA, and they will surface either during testing or in production. Our goal was to identify and address as many of these issues as possible during testing, and we uncovered most of them. This required the dedication of these 1,700 testers who rigorously evaluated the system leading up to the go-live and during the transition. Their collaboration continued into the post go-live phase when they communicated with us during a couple of weeks of what we called our ‘hyper-care period’ to minimize disruptions.
What that leaves us with are more than 1,700 people globally who have a new appreciation for SAP S/4HANA and have some skin in the game with this initiative. This is not hurting us either. It was not intentional. It is a side effect of the way we approached the testing, but it is paying benefits as we start to move forward with delivering the value.