For the latest entry in our spare parts management interview series, founder Meir Weisberg sat down with Conrad Greer for a deep dive into parts classification and identities, the role of AI, and the myriad ways spare parts management can affect an operation (for better or worse!).
Conrad is the Founder of SPC Results Inc., a consultancy specializing in Supply Chain and Maintenance business process improvement. Drawing from his background as a former Marine Engineer Officer in the Navy, he brings extensive experience in materials management and SAP systems implementation to asset-intensive industries. His notable contribution to the field includes the development of SMI Templates (Structured Material Identities), a system for standardizing material master descriptions.
As a recognized expert in MRO (Maintenance, Repair, and Operations) inventory management, Conrad has led multiple SAP implementation projects for energy and utility clients, focusing on end-to-end material procurement and inventory management processes. He has shared his expertise at MAINTRAIN, the national conference of the Asset Management Association of Canada (PEMAC), where he presented his innovative approach to reducing costs and improving operational safety through better materials identification systems.
Note: we’ve made some slight edits to the original text to ensure consistency. These changes are only grammatical or format-related, and we haven’t made any changes to the contents.
Q1: How did you get involved in spare parts?
I started off my career in the Canadian Navy as a marine engineer officer. For the Star Trek fans, I was Scotty–that was my job. We carried spare parts with the NATO codification system. Spare parts was sort of a non-thing; there was a lot going on, but we knew what parts we had and where we could find them. It was easy!
In my second career, I went into the SAP implementation space as a materials management functional consultant, and the clients that I ended up working with were mainly oil and gas, utilities, railroads—they weren’t manufacturers. The direct supply chain wasn’t a significant thing. The people in the procurement warehouse were there to support the maintenance team to keep their assets running. That was the focus, so MRO and spare parts kept cropping up.
So there's a problem here, and as I went on to be an independent consultant, I had now worked in several merger projects. A client acquired a refinery and merged them together– both SAP clients, running many of the same equipment and with the same database, but no comparability between parts identities. With weak catalogue identity standards the records are incompatible. We do our duplicate checks and we get next to nothing, so what's going on here? This is where it started to dawn on me–it's not the ERP software. This isn't just an SAP story, this is all of the CMMS's and ERPs. There's something about the quality of the items. To me, it's the identity. It's just ended up being something that I got interested in and saw that this is a problem that needs work. I'm certainly not the only person that's interested in it. It’s a gap that's existed since the dawn of business systems. We went from the old experienced maintenance guy that knew the parts and they called suppliers and they got the right parts and they took crib notes about what to ask for. This went into local systems and then the chaos, in my mind, became bigger when we started centralizing on enterprise systems.
Q2: What are the most significant business impacts you see from not having the correct inventory identity?
Maintenance workers are the primary user of MRO parts or materials. So for them, I'll put it into wrench time.
For companies that are primarily reactive, if something breaks, we have to go get parts that are difficult to find. They search for the thrust bearing they need and they get 52 possible answers. Then they're back on the phone, so the system is not helping this situation. It's hindering it. So, that's the picture I'd put on the reactive side of maintenance. And then you get over to a lot of companies now that are pretty advanced and they plan their work. They don't schedule jobs until the parts are available. Their loss is the wait time from planning to material receipt.
This is sort of the beginning of the maturity curve. Ideally, it's all connected. Clutter in the catalog impedes its success. Crosschecking must be done. They may get the wrong parts because it looks like it's right but it's not clear. It takes longer to get the job scheduled and then once it's scheduled we still may be issuing the wrong parts. Then now, it’s at the job they find out ‘this isn't right, we have to stop this job.’
For example, in oil and gas, you have these pressurized systems with hydrocarbons, and they are very dangerous. So the specification of the materials is important even in something as small as a drain line. If it's not the correct material identified item, it's going to be a problem. They do positive material identification, they take steps, but it's like having a dirty cluttered working environment: you're setting yourself up for mistakes. With that, if you have the wrong material spec and then that drain line rots out. Sometimes, in use, it explodes, it starts a fire and you've got a life threatening environmental disaster.
So at the far end of the spectrum these mistakes can have some very low-probability but high-effect disasters. So that's the maintenance world that it harms. Over in the supply chain world there are two big problems for a lot of companies: in a lot of situations spare parts are supplied through the warehouse. We will stock it, if we need it, we'll put it in stock and get it off the shelf as needed. There could be some optimization to that, but that's a basic plan. I have not found a company that has an efficient warehouse. The common story is that working capital and the MRO inventory go up every year. It always increases. It never goes down.
We're adding more stuff, we're not consuming it as fast and we've got sophisticated algorithms and MRP–but the problem isn't there, it's on the master data. My primary theory on it is because these things are poorly identified, you get the situation that when someone can't find what they need, they create a new one, and then it's sort of like the sedimentary seabeds, you get layers. Seven years ago, we used part X for this. Then we forgot about it, but we still have them in stock. And then we created a new one and we use part Y. So you add, and each time you trip up and make a new record, you get new stock. The old stock doesn't go away. Often the old stock is fossilized. So inventory keeps growing as well as the space needed to store it.
The new items will also pass duplicate checks when they are created, because the duplicate check is narrow–it’s on the manufacturer part number, and we're talking about stuff that may go back 30-40 years. Did they originally enter that same number? Then you have supplier numbers and other things that make it challenging for the duplicate to work. You need more structure and item comparability than just an MPN check. This causes working capital for MRO to go up and up.
Something that can happen when supply chain and maintenance functions are separate and report up their own chains of command is the frustration of working together can boil over and maintenance doesn’t trust the warehouse. They just add free text in their items and then they call the supplier like they did 50 years ago to get the right item. They do this because they don’t trust SCMs. Maintenance needs one part but they order two or three so they have one for next time. In our Canadian terminology, (may be American as well), we call these squirrel piles. It disappears from the system at the moment of goods receipt and is impossible to track.
These parts only get used if the person that knows where they put them in their locker and they’re around when the next job comes up. If they move on, if they're off shift or it's in a lay down yard and it's sitting outside, it doesn't last. So moneywise, this common practice has probably the biggest financial impact, and it gets looked at as this sort of culture battle between maintenance and supply chain. It's bad master data that is poking them both in the eye that is really the root cause of it.
They would get along with each other if it worked, but it doesn't work. So, now they're left to struggle. That's my business overview of it.
Q3: What are the key challenges when looking at a plant that needs to work across industries? Where do people struggle the most with spare parts management?
I see the root of all evil is uncontrolled spare parts identities. What we want to have is good master data that works for every MRO process replenishment, warehouse put away, kiting, and issue, and warranty; but that starts with part identities.
Companies I've worked with spend time on parts lead times, reorder points, max stock, min-max. They have people that understand that and work on it. The problem is that there's not enough of them at scale. Typically, I'm looking at companies, enterprises with 150,000 to 700,000 records, so hiring a person to clean up this amount of records is an impossible task. If you have 50% more records than needed, it is even a more difficult task.
Then there’s an operations and maintenance world and a capital projects world, and they don't connect. A lot of master data preparatory work is done and paid for in the capital project. What spare parts do we need? Often projects fund the first several years of spares. But the connection between Projects and Operations is broken and the intended benefit is not realized.
More recently, organizations are maturing. They're putting operations personnel into the early stage capital project. They're examining the maintainability. OEMs are a good starting point for what is needed for maintenance, but it's very much from the OEM point of view, not the operator's point of view.
You may have other assets that you could go share spares with, but it starts with ‘how are we going to maintain it? Do we take the whole unit out or do we have a pump shop that's going to overhaul these items? What are we going to maintain? Then what items do we need to achieve that?’
Maintenance strategy needs to involve procurement. ‘Can we get parts right away, or do we need to stock them?’ I've done a lot of work in the Canadian oil sands, which used to be at the end of the world. To run these mines and upgraders, they had to have parts on site to fix a breakdown because it was multi-day provisioning, even if it was directly on the shelf in a US warehouse.
The oilsands region had a few hundred billion dollars of capital investment invested so OEMs set up locally. Caterpillar has distribution there now. Parts are sitting 40 minutes away. If you're thinking about it from a supply chain point of view and trying to optimize, do we need to have this sitting there? When doing planning and scheduling, you need to consider emergency breakdown requirements. But if it's something that you can get in time, you can just buy it as you need and not have stock using working capital.
This all fits built on the foundation of decent data. It doesn't have to be perfect data, but decent data. Every business process that uses MRO items is poorly, negatively affected with low-quality identity data.
Q4: moving forward, how do you change the process, the mindset?
The company needs to understand the pain and waste they live with for ignoring poor MRO item identities. Too often the waste is accepted as a cost of doing business.
The simplest but least accurate method is a parametric approach assuming the size of the problem is proportional to the number of people involved with MRO processes. Make a simple guess of the annual cost of waste per MRO staff member and multiply by the number of staff and you have a simple estimate.
A more detailed approach is to list the harms that arise from poor quality MRO item identities, estimate a cost per occurrence, and the frequency of occurrence. Multiply them out and you have a size of prize estimate.
There is significant opportunity in addressing these inefficiencies.
Q5: After you've assessed the opportunity, how do you workshop with the customers? What tools do you give them? And how do you change their processes?
Currently, I work remotely with clients. We do assessment projects and collect data dumps. We look at maintenance reporting and try to put the story together. This serves two purposes. First, it paints a clearer picture of the opportunity, going beyond just saying "this hurts" or "this sucks."
Second, it helps me scope and plan the project. My method involves applying a structured taxonomy. We take all the data, put it into that taxonomy, and then rationalize it. We can see that it's not just different manufacturer part numbers, but the first five attributes of the description are critical.
We work closely to do a fuller comparison between item identities. This includes description attributes, manufacturer information, manufacturer part numbers, and for non-manufacturer items like pipes and fittings, safety-related specifications.
Key specifications and material requirements need to be part of the item's identity. In SAP, we can use classification to store these identity attributes. I then look at international standards: ISO 8000, ISO 22745, and UNPSC, which most people lean on.
However, none of these standards intend to solve the maintenance planner's specific problem. The goal is to help users rapidly identify the exact item they need for their job. UNPSC, for instance, is just a grouping methodology that puts things in broad commodity code buckets.
My approach involves building a client-specific taxonomy. Clients validate it, and I use an offshore outsourcing company to codify and enrich the data set. Often, when we define five attributes for an item, two will be missing in three-quarters of the entries.
An important aspect is handling "indeterminate" items. For items without stock, we flag them for deletion. If some are on the shelf, we need someone to investigate. Old or obsolete items get removed.
The ultimate goal is to build a rational MRO catalog with one item for each real-world maintenance item needed to run the business. An often-overlooked issue is obsolescence, which currently lacks an effective process in most systems.
Obsolete parts are clutter. They take up space, can be confusing, and provide no value. While removing them is important, it's not the front-and-center problem. There are more critical issues to address first.
Q6: after you leave a client with a clean defined catalog, a good taxonomy, what's their next step? Beyond being responsible for maintaining that taxonomy, how do they move on from there?
Once you create a rational catalog, you want to maintain it. The key is using the taxonomy used to rationalize the catalog for every future input.
This applies to company acquisitions, mergers, and scenarios where you're adding 50,000 new items. You must classify these items to the taxonomy, rationalize them, and integrate them. The same is true for new capital projects.
The goal is to include operations throughout the duration of capital projects. When introducing new items, potentially with new technology, they must go through the taxonomy. There might be synergies to explore.
A large cause of catalog clutter comes from day-to-day, one-by-one part requests. While they typically do a duplicate check, the manufacturer part number (MPN) check is often weak, allowing many duplicates through.
I've observed with clients that the master data support desk receives a significant volume of new material requests. However, they haven't added any substantial new capital projects. The volume of items being added has no correlation with their actual operations. It's essentially just adding more copies of existing items.
Interestingly, many of these are safety-focused environments with respected change management processes. For most changes, impacts must be studied and acknowledged. Yet for spare parts, there's a disconnect. The attitude is: "We need a new part? Sure, we'll add it." There's no consideration of how these parts have been maintained over the past 15 years. The system is fundamentally broken in this sense.
Meir Veisberg: Give the project an M&A, they'll have to kind of apply the taxonomy and all of it. Maybe get some support to do it, but then the day might still creep up on them. Someone make a request and then the taxonomy would not be applied, the material will be created and then that could still happen.
The goal is to make use of a structured taxonomy with controlled identities. The taxonomy isn't rigid. It's based on centralized, observed data, with limited opportunities for arbitrary creation.
The right group of people can identify improvements. Clever field personnel might suggest additional attributes for specific material classes. If they have a compelling reason, the taxonomy can be adapted.
For instance, if we discover an important attribute for 2,000 items, we could theoretically extract, change the taxonomy, and populate those values. This isn't always practical—especially if checking requires complex investigations like reviewing drawings.
The key point is that the taxonomy shouldn't be fossilized. It can be improved when there's clear value. There's no absolute precision in this process, just better ideas.
The approach is flexible. By improving the taxonomy, you move closer to a theoretical maximum efficiency for MRO processes. Quality master data helps achieve this theoretical optimal state.
Q7: What role will technology, especially things like AI and machine learning, play in the space in the coming years?
I believe the problem of uncontrolled item identities will be solved in 10 years. Companies with highly uncontrolled identities will become rare. The conceptual framework has been the primary barrier to solving this issue.
People have attempted data cleanses out of frustration, but these efforts haven't truly fixed the problem. The interconnectedness of items hasn't been the focus. Typically, these attempts just made individual items more structured without considering their relationships.
Now, the conceptual basis of the MRO identity problem is clear and will be resolved. Powerful tools—particularly AI and machine learning—will drive this transformation.
While "AI" is the current marketing banner, it's fundamentally a machine learning problem. Applying machine learning correctly will help solve the issue at scale.
I see the MRO maintenance world evolving to a point where uncontrolled identities will seem unbelievable. Machine learning and AI will be the primary engine driving this mass movement towards better data management.
Q8: What do you tell that material manager that maybe doesn't have the budget to get new software consultant? What should they do to improve their work kind of in small incremental steps?
In incremental steps, I have recent experience with this process. Most places aim to find a level of consistency in their items. This effort often focuses on moving forward, particularly on new individual material requests or "onesie-twoosies." The value lies in bringing in thinking that's going to help you going forward.
You can't touch the 150,000 records that are there and unstructured. However, what commonly happens is when I ask companies how they name their materials, the most standard answer is: “we call it what the requester wanted to call it, because he's a maintenance technician, he's a maintenance planner, he's in reliability, he's somebody that knows more about these things than I do. So if he says it's a right-handed wing nut, it makes sense.” Again, it's this missing piece–this useful standard to apply to it–that's what needs to be brought to bear.
We hope you enjoyed the article and gained further insights into expert spare parts management!
Follow us on LinkedIn for more content on how to elevate your spare parts knowledge.