00:00:07 Intro to min/max inventory approach.
00:01:45 Benefits of min/max: simplicity, automation.
00:02:29 Limitations of min/max: write-offs, non-optimization.
00:04:39 Min/max issue: oversimplified SKU-centric perspective.
00:07:09 Why min/max gained traction; easy ERP integration.
00:09:20 Min/max limitations in supply chain management.
00:10:58 Overcoming min/max limitations: hacking settings.
00:12:15 Advocacy for modern inventory systems; transition difficulties.
00:14:22 Alternatives to min/max: prioritization, nonlinear constraints.
00:17:00 Risks of certain business decisions.
00:17:25 Min/max approach: potential drawbacks.
00:18:30 Financial impacts of inventory mismanagement.
00:19:01 Advice for managers: moving beyond min/max.
00:20:43 Closing thoughts.

Summary

In an interview, Joannes Vermorel, founder of Lokad, discussed the min/max approach to inventory management. The approach utilizes minimum and maximum stock levels for automated replenishment, improving productivity but failing to account for fluctuating demand, leading to overstocking and potential inventory write-offs. Vermorel argues it oversimplifies supply chain management and neglects the interconnectedness of inventory. Despite these limitations, its simplicity and ease of implementation have led to widespread adoption. A workaround for discrepancies in supplier order quantities was suggested. Alternatively, Lokad recommends an approach prioritizing value from the client’s perspective, considering nonlinear constraints, and basing decisions on projected future demand.

Extended Summary

In this interview, host Kieran Chandler discusses the min/max approach of automated replenishment with Joannes Vermorel, the founder of Lokad, a software company specializing in supply chain optimization. The min/max approach is a simple yet prevalent method of inventory management.

Vermorel describes the min/max method as exceedingly simple, involving two parameters: the minimum (min) and maximum (max) stock levels. The min parameter is a trigger point, indicating when a stock keeping unit (SKU) is due for replenishment. The max parameter denotes the target stock level, indicating the amount to reorder to bring the SKU stock level back up. This system results in a ‘sawtooth’ pattern of inventory levels, where stock slowly decreases until it hits the min level, triggering a replenishment to reach the max level, before decreasing again.

One of the main advantages of this approach, according to Vermorel, is automation. Once the min and max parameters are defined for an SKU, the system automatically generates purchase orders when the stock level hits the min value. This automation can enhance productivity by reducing the manual work involved in monitoring and ordering stock.

Despite its simplicity and automation benefits, Vermorel highlights significant limitations of the min/max method. The system doesn’t account for changes in product demand over time, potentially leading to unnecessary stock build-up for slow-moving or dying products. The min/max approach will continue triggering replenishments, regardless of whether the product’s demand is gradually vanishing. Consequently, companies using this method could end up with significant inventory write-offs due to overstocking of products with declining demand.

Therefore, while the min/max approach to inventory management offers simplicity and automation, it also has inherent limitations. It doesn’t account for fluctuating product demand, leading to potential inventory write-offs. It also offers no guidance for setting optimal min and max levels, potentially requiring constant manual adjustments to achieve optimal inventory management. These shortcomings indicate that relying solely on the min/max approach may lead to suboptimal inventory management and potential failure.

Vermorel first criticizes the min/max inventory approach, asserting that it overly simplifies the complexities of supply chain management and artificially narrows the perspective to a single SKU (Stock Keeping Unit). This approach prompts businesses to focus on individual SKUs, disregarding the interconnectedness of their inventory and the broader supply chain. The primary question isn’t when and how much to reorder of a specific SKU, but rather, given limited resources, where should investment be prioritized across the entire range of SKUs.

Vermorel describes this SKU-centric view as “poisonous” as it ignores the competitive nature of inventory management, where all SKUs vie for the same investment. This mindset pushes businesses into a cycle of constant tweaking of the min/max parameters, while neglecting the broader perspective of how to best serve clients who require multiple SKUs. As a result, companies may struggle to optimize their service level and overall inventory investment strategy.

Chandler then prompts Vermorel to explain why, if such limitations exist, the min/max approach gained widespread adoption. Vermorel identifies its simplicity and ease of implementation as primary reasons, particularly in early relational and database systems. A min/max inventory policy could be added to an ERP (Enterprise Resource Planning) system with a few SQL queries, making it an attractive feature for systems starting from scratch.

Historically, the preference for monolithic systems also played a role. These systems, capable of tracking inventory inflow and outflow, were extended to include basic replenishment intelligence such as the min/max feature. Vermorel suggests that the better approach would have been to integrate a separate system dedicated to replenishment intelligence. However, the path of least resistance was often to simply add on to the existing monolithic structure.

The min/max approach can be effective if supply chain practitioners are dedicated to daily tuning of the min/max values. However, this places significant workload and responsibility on the practitioners themselves, with the potential for suboptimal outcomes if this effort is not maintained. Min-max is a supply chain strategy hard-coded into many ERP (Enterprise Resource Planning) systems. This strategy works relatively well when the supplier’s minimum order quantity matches the system’s min-max settings. However, it becomes problematic when the supplier’s minimum order quantity is expressed in value rather than in units. For instance, if a supplier only accepts orders worth at least $10,000 across all products, it complicates the system that prefers per product minimum quantities.

To circumnavigate this issue, Vermorel suggested a workaround - a system hack. In a rigidly min-max system, one can set the min value at zero and the max at an infinitely high number, so it doesn’t trigger the order. When an order needs to be placed, the values can be adjusted accordingly to allow the system to emit the reorder signal, and then reset back to the initial values. Vermorel acknowledges the extra layer of complexity this adds, but he argues that if a company is stuck with an outdated system, it can be a reasonable short-term solution.

The conversation then shifted to the alternatives to the min-max approach, specifically focusing on Lokad’s approach. The key differences according to Vermorel include prioritization, consideration of nonlinear constraints, and basing decisions on projected future demand. Prioritization involves considering the value of each product from the client’s perspective. Nonlinear constraints include minimum order quantities from suppliers and other factors like warehouse capacity. Projected future demand involves generating probabilistic forecasts to assess potential futures and associated risks.

If the min-max approach is “kind of working,” it suggests that inventory isn’t strategic for the business, and stock is cheap. However, Vermorel warned that as soon as inventory management starts having a significant financial impact, a more sophisticated approach should be considered to avoid leaving money on the table.

To transition away from a min-max approach, Vermorel suggested starting with prioritization. This can even be achieved with a simple tool like Excel. Products should be ranked based on their urgency for inventory, factoring in aspects like demand forecast, criticality for clients, and inventory risk. This prioritization can improve productivity, as it allows supply chain managers to focus on the top of the list rather than scanning every product linearly.

Full Transcript

Kieran Chandler: Welcome back to Lokad TV. This week we’re going to be discussing the min/max approach, one of the earliest automated replenishment methods to be incorporated into inventory management software. The primary benefit of this approach is its simplicity. However, in today’s episode we’re going to explain why taking this path means you’re ultimately on a route destined for failure. So, Joannes, perhaps a nice place to kick things off this week is if we could just describe a little bit more about the min/max approach and what it actually entails.

Joannes Vermorel: The min/max inventory replenishment policy is exceedingly simple. You consider one SKU and you define two parameters: your min and your max. The min says that when your stock keeping unit, your SKU, reaches this stock level, you trigger a replenishment. And how much do you replenish? Well, you replenish just enough so that your stock level would go up to the max value. So, you have the min that acts as a trigger, and the max as the target. This results in your inventory having a saw-like shape, where your inventory slowly decreases, then replenishes, and then decreases again. It’s indeed extremely simple to implement. You can do that in pretty much any system, and pretty much every single supply chain system out there has some kind of min/max equivalent.

Kieran Chandler: It seems very simplistic. What are the other key benefits to having this approach?

Joannes Vermorel: The one benefit, assuming that you have a software-based inventory management system in place, is that you get automation. It means that you could literally do nothing and purchase orders will be generated once you hit your min, which represents the trigger. So, in terms of productivity, it’s relatively good because it delivers something that can be valuable, which is complete automation.

Kieran Chandler: So that’s all the good points. How about now we talk about some of the bad points? It’s a technique that’s incorporated into many ERPs. What are its limitations?

Joannes Vermorel: The main limitation is that the inventory management that results from min/max is typically quite bad. One reason is that you will always end up with significant inventory write-offs because this min/max approach will trigger replenishment no matter what. So, if you have a product that is gradually vanishing, you will still place another reorder even after a year of slowdown. So, the problem is that there is no such thing as phasing out a product with min/max. You pretty much end up with inventory write-offs by design. Additionally, as soon as you want to optimize anything in terms of inventory, not just save time so that orders are being passed, the min/max method just tells you nothing. It just tells you that you could choose a min, you could choose a max, but the reality is that choosing those two conditions, which is when to reorder and how much to reorder, is entirely left to you. If you’re up to revisit all your min/max settings on every single day with min/max, you can implement any other policies that you want, but you’ve completely lost the automation benefits when doing that.

Kieran Chandler: So this is why we’re talking about this sort of method basically leading to failure. So, what are the problems that people are facing in the real world then by using this kind of approach?

Joannes Vermorel: I believe that the main problem is more insidious than just the min/max being way too simplistic. The main issue is that it frames the problem in a way that is relatively poisonous in how you want to even understand what’s going on in your supply chain.

Kieran Chandler: The force is in iteration, you know, the rest of the universe doesn’t matter. When you think about it, for most companies, they have thousands of product references or at least hundreds, and typically several locations. So things do not happen in isolation. The question is not so much of when should I order this SKU and how much, but if I have one extra dollar or euro to invest in my inventory, where should this euro go in priority?

Joannes Vermorel: You see, all your references, all your SKUs, compete for the same money, for the same budget that you could invest in your inventory. But the min/max perspective gives you a SKU-centric perspective on replenishment. So, all of this competition that exists within the SKUs that you manage does not even exist. And that’s where it’s relatively problematic, because then it means that you’re turning the problem into finding two parameters, min and max, and potentially revisiting those parameters every week, every month, every year, whatever. But it gives you a mindset where, fundamentally, it’s one SKU at a time, while what is really happening for your business is that you have clients that need many SKUs, and so all the SKUs compete for your budget. You want to really think of how can I offer the best service possible and not just a solution for this one SKU.

Kieran Chandler: And if it’s a sort of approach that is so problematic, as you say, why did it get so much traction originally? Why is it something that’s incorporated in so many ERPs?

Joannes Vermorel: I mean, it’s clearly something that is exceedingly simple to implement, especially when you think of the early relational systems, you know, early database systems. A min/max inventory replenishment policy can be implemented with like three or four SQL queries. You need one for the trigger, one for the quantity, maybe one to finalize the purchase order, and whatnot. It’s literally a handful of SQL queries that you need to add into your ERP to deliver this capability. So, it makes sense if you’re starting with nothing to add that.

Historically, there was this big approach on having monoliths, so you have an inventory management system that is perfectly fine at managing your inventory, so keeping track of things that come in and things that go out. But then you want to add a feature for replenishment intelligence. You start with something. The better approach would have been to say, “I’m not going to even venture in this sort of feature, so that I can have another system that is dedicated to this replenishment intelligence.” But it turned out that the easiest path was just to extend your big monolith and say, “Well, I can already manage my inventory, so I can add the first bit of replenishment intelligence.” Basically, you start with min/max.

If you assume that you will have supply chain practitioners that will revisit those min max values every single day to tune them, then the reality is that it can work relatively well. But actually, it’s just that all the intelligence and the effort is put on the shoulders of the supply chain practitioner.

Kieran Chandler: The problem is that a lot of our clients are very much bounded by the capabilities of that ERP system. So what are the options available to them if they’ve got something like min max that’s very much hard-coded into their ERP?

Joannes Vermorel: That’s a problem, because even if that can happen, that you have this sort of features that are kind of built into your system, it starts to create a lot of friction.

Kieran Chandler: When, for example, there is a mismatch with your minimum order quantities from your supplier. If your supplier has minimum order quantities expressed per product, you might have to order 50 units for every single product you order. That works relatively well with the min-max approach, because you just need to have more than 50 units difference between the min and the max. The problem arises when, for instance, a supplier tells you they only accept orders if you order at least $10,000 worth of products from them. In this case, you’re looking at the range of products you can source from this supplier, and there might be hundreds of them.

Joannes Vermorel: Yes, you don’t want a very simplistic approach which is, “I only order a product if I can place an order of ten thousand dollars for this product alone.” That’s the sort of friction that can happen. It’s possible to maneuver your way around these min-max settings, for example, if your system is rigidly geared toward min-max. We had a few clients like that. The trick is to set the min at zero so it never gets triggered, and the max that triggers oppositely. You game the system so that it doesn’t respond to anything. Then, when you want to pass a resolution, you’re going to tweak the values on the spot, let the system emit the reorder, and then shut it down again by resetting those values at zero. It’s very system dependent, and the idea is that you don’t want to do this manually. You would directly tweak the content of the relational database, so you can reverse engineer the behavior that you desire.

Kieran Chandler: The idea of hacking into a system and reverse engineering it might make a few of our viewers a bit nervous. Does this add an extra layer of complexity? Is this an approach that you would recommend?

Joannes Vermorel: The recommended approach is to have a modern, cloud-based inventory management system that has an API and all the modern features. But the reality is, if you don’t have that, you might be stuck with a two or three-decade-old system that is completely entangled with the rest of your company. Transitioning to something much more modern might not be a reasonable short-term option. Yes, there is some extra friction, but the reality is that because min/max is very simplistic, hacking an override of min/max is still relatively simple. It’s much easier to reverse engineer because all systems tend to have some flavor of min/max. The trigger might be, for example, a date like every Monday I reorder up to the max. So I don’t have a min, I just have a max, and then I have a schedule that’s every Monday, or the first day of the month, or something else. You have many flavors of this min/max inventory, where it typically comes with one simplistic trigger condition and one simplistic kind of target.

Kieran Chandler: And with the more elaborate, semi-smart replenishment policies, it gets much more complicated to hack your way in?

Joannes Vermorel: Yes, that’s where the paradox is. When you have more elaborate, semi-smart replenishment policies, it gets much harder to hack your way in if the system has to stay in place and cannot just be deactivated. But let’s move away from the hacking and talk more about the alternatives to the min-max approach. What are the key differences between a Lokad approach compared to a min-max approach?

Kieran Chandler: So, what do companies need the most? I guess there might be some products where the margin is thin, but those products happen to be completely critical for your clients. If you don’t have them, you risk losing the clients altogether. So, would you say prioritization is the first step?

Joannes Vermorel: Indeed, prioritization is probably the first step. Then the second step is to think about all the nonlinear constraints you have in your supply chains. It’s very rare that you can completely ignore these. For example, you might have minimum order quantities from your suppliers. These reflect the fact that every time you pass a reorder, the supplier has to make a delivery. This is a fixed cost that is independent of how much you order, and it might be reflected in the pricing you get. Then you might also have other constraints such as if you have too many deliveries on a single day, your warehouse staff can get overwhelmed. There are many other nonlinear constraints that are not even reflected from a min/max perspective. It’s typically quite domain-specific, but the bottom line is that you want to make sure that your model is not ignoring the physical reality of your supply chain.

Kieran Chandler: So, you’re saying that it’s all based on projected future demand. Min-max inventory doesn’t say anything about how you even project the demand.

Joannes Vermorel: That’s correct. The future is uncertain, so the first differentiator is that you need to plug in some kind of probabilistic demand forecasts. This allows you to assess many possible futures and the risk associated with taking certain decisions, or not taking certain decisions.

Kieran Chandler: And a lot of our viewers might be using a min/max approach that is kind of working for them right now. So, what could the possible outcomes be if they don’t heed some of these warnings? Why should they change from that min-max approach?

Joannes Vermorel: Well, if the min-max approach is kind of working, it probably means that inventory is non-strategic and that your stock is very cheap. All you care about is automation. That can work if you have many products that need to be available, but the downside is that it’s very expensive if you don’t have them. However, if there is no specific downside to having too much inventory, like in the case of office supplies for example, the min-max approach can be fine. But as soon as it has a significant financial impact on your supply chain, you might be leaving a lot of money on the table by not optimizing inventory management.

Kieran Chandler: To conclude, we like our viewers to go away with a bit of a lesson. If I was a supply chain manager, what should be the first steps I should take in order to move away from a min/max approach?

Joannes Vermorel: The first step is to go for anything that looks like prioritization. You can even do this in Excel. The question would be, if you have all your products, can you rank them from the one that most urgently needs inventory to the one that has the least need for inventory?

Kieran Chandler: You’ve mentioned about mini demand forecast made of a moving average. It’s very crude, and then there are some kind of factors that would consider the criticality for your clients. Is this something that is an add-on or is it mission critical?

Joannes Vermorel: It’s something that includes certain factors. For instance, it can consider the inventory risk. The question is, is it something where you have fast obsolescence or is it something that will last forever?

With a very basic numerical recipe, you can already have a ranking that makes sense. The good thing is that in terms of productivity, it means that even if you’re still extensively manually processing your stock, as soon as you have a prioritization, you have a priority list. This way, you know where to look at.

Even if your prioritization is somewhat crude, the top of the list is typically a good place to start, if your recipe is not completely dysfunctional. This approach is better than just linearly scanning every single product every single week.

Kieran Chandler: Okay, great. Well, we’re going to have to leave it there, but thanks for your time today. That’s everything for this week. We’ll be back again next week with another episode of Lokad TV. Until then, thanks for watching.