00:00:06 Stock Keeping Units (SKUs) and their limitations.
00:00:30 SKUs and their role in inventory management.
00:01:58 How SKUs relate to basic supply chain models and formulas.
00:03:42 Issues with SKUs and real-world assumptions, such as expiration dates.
00:05:01 Challenges with SKUs in the fashion industry and substitute products.
00:08:00 The problem with SKU-centric thinking in supply chains.
00:09:33 Alternatives to SKU-based systems for various industries.
00:11:25 The gap between SKU abstractions and real-world needs in different verticals.
00:12:26 Industries where SKU-based systems make sense and those where it doesn’t.
00:14:55 The real problem with traditional ERP systems and the need for better mental models.
00:16:02 Advice for supply chain practitioners: expand horizons and rethink production flows.
00:17:00 Assessing software fit for SKU flexibility and avoiding vendor lock-in.
00:17:58 The future of SKUs and their impact on supply chain.
00:18:28 SKUs as training material for students and grasping basic formulas.
00:19:07 Closing thoughts.


In the interview, Joannes Vermorel, founder of Lokad, discusses the limitations of the Stock Keeping Units (SKUs) model in supply chain management. He highlights that SKUs are useful abstractions, but their one-size-fits-all approach can be problematic for industries like fresh food, aerospace, and fashion due to factors like perishability, substitutions, and varying lifecycles. He suggests supply chain practitioners should critically assess whether SKUs fit their business needs and consider alternative approaches. Vermorel emphasizes that while the SKU model is helpful in supply chain education, businesses should focus on real-world demands to ensure efficient resource allocation and better decision-making.

Extended Summary

In this interview, host Kieran Chandler discusses the concept of stock keeping units (SKUs) with Joannes Vermorel, founder of Lokad, a software company specializing in supply chain optimization. They begin by defining SKUs as an abstraction that reflects the same product stored in multiple locations. The concept of SKUs is important for bookkeeping and managing inventory, as it simplifies the process of tracking products in different locations. In some warehouses, each SKU may have a designated physical bin for storage. The main assumption behind SKUs is that all units within an SKU cannot be differentiated.

SKUs are helpful in developing mathematical models for supply chain management, as they provide a clear way to represent goods flowing in and out of a system. Early supply chain models like the economic order quantity, Wilson’s formula, and safety stock formula all implicitly work on a single SKU. Beyond equations, SKUs are implemented in the core of many supply chain systems, such as ERPs, WMSs, and eCommerce platforms.

However, the assumption that all units within an SKU are identical can be problematic for some industries. For example, in fresh food, expiration dates become crucial. If an SKU contains 100 units of a product with a daily demand of 20 units, it might seem that there is enough supply. However, if 90 of those units are set to expire overnight, only 10 units will be available the next day, leading to a potential shortage.

Vermorel argues that while SKUs are useful for categorizing common products, they may be insufficient when dealing with items that have a certain level of granularity or substitutability.

Vermorel highlights that even when dealing with seemingly identical items, the notion of an SKU can be misleading. He uses the example of a store selling white shirts, where each shirt SKU represents a different model. Although it may seem relevant for businesses to know the exact number of units per SKU in order to manage inventory and replenishment, Vermorel contends that this information may not be as useful as it appears. This is because customers don’t walk into a store with a specific SKU in mind; they walk in with a need, which can be fulfilled by multiple similar products.

The problem, as Vermorel explains, is that the SKU system doesn’t account for substitute products. By focusing solely on SKUs, businesses may not recognize that they have an excess of stock in similar items that serve the same demand. This can lead to inefficiencies and a misallocation of resources.

As an alternative, Vermorel suggests looking at the problem from the perspective of service units and units of demand. For example, in the case of car parts, there may be multiple compatible parts that can fulfill the same function for a specific car model. In this context, the question of availability should not be framed around individual SKUs but rather whether there is a compatible part available for the customer’s car.

Vermorel also emphasizes that there are numerous alternatives to the SKU system, but because the SKU framework is so pervasive in the industry, it can be difficult for businesses to recognize and adopt these alternatives. The key is to move beyond the narrow focus on SKUs and explore other approaches that can better address the complexities and nuances of supply chain management.

They discuss the concept of Stock Keeping Units (SKUs), the limitations of the SKU model, and the implications for different industries.

Vermorel explains that SKUs are an abstraction, and like most abstractions, they can be “leaky.” This means that while they simplify the process of reasoning and coding into software, there can be a gap between the abstraction and reality. The larger the gap, the greater the risk of unintended consequences. The suitability of the SKU model varies from one industry to another.

In aerospace, for example, the SKU model is mostly nonsensical due to factors like serviceable and unserviceable parts, as well as the varying lifecycles of individual components. In the fashion industry, it is also mostly nonsensical because of the numerous possible substitutions. For food, the SKU model is again mostly nonsensical, given the high degree of substitution and perishability.

However, the SKU model can work well in certain industries. For example, it is a fairly good approximation for Apple’s iPhone business, as customers want the latest iPhone and would not consider another model as an adequate substitute. Similarly, for companies like Procter & Gamble that deal with fast-moving consumer goods (FMCG), the SKU model makes sense, as their products are well-defined, and there is little room for substitution.

Vermorel explains that many supply chain software solutions have emerged from the FMCG industry, where the SKU model works well. Consequently, other industries have adopted the same approach even when it may not be the best fit. While supply chain practitioners are not necessarily happy with this situation, it is often difficult to pinpoint specific issues with existing software.

He emphasizes that the real issue is not the outdated user interface of old Enterprise Resource Planning (ERP) systems, but the flawed assumptions built into the software’s mental model, causing friction for businesses.

Vermorel suggests that supply chain practitioners should expand their horizons and critically assess whether SKUs truly make sense for their business, both in terms of production and demand. If there are more efficient ways to organize production flows, businesses should consider adapting their approach. Additionally, if clients think about products differently than the SKU model, businesses might want to create software implementations that reflect this reality.

He cautions against relying on vendors to make adjustments to SKU-centric software, as it would be extremely difficult to change the core structure of the system. Instead, businesses should carefully evaluate whether a software’s built-in choices are suitable for their needs.

Vermorel sees the SKU concept as useful for training purposes in supply chain education but believes it is important for businesses to keep their focus on the real-world demands of their operations.

Full Transcript

Kieran Chandler: Today, we’re going to learn a little bit more about its limitations and understand some of the edge cases where the concept of SKUs actually falls down. So, Joannes, perhaps we should just start as usual by sort of defining what SKU actually is.

Joannes Vermorel: The SKU, or stock keeping unit, is an abstraction designed to reflect that the same product can be stored in many locations. So, if you have, like, 100 different products and you have a thousand locations, then you will end up with 100,000 SKUs because you’re going to count every single product in every single location. You introduce the notion of SKU very naturally, and it’s typically a bookkeeping unit. Sometimes, if you’re looking at warehouses, it can literally be materialized by the fact that for every SKU, you have a physical bin that contains the stuff. What you assume is that all the units within the SKU cannot be differentiated.

Kieran Chandler: It’s a concept that’s been fairly ingrained in the supply chain. So, what’s interesting about the concept of the SKU, and what are some of the features it looks at?

Joannes Vermorel: First, it’s interesting at a minimal level. When you want to write down the first mathematical models to model how flows go through your supply chain, SKU is a very nice way to write down your equations. Most of the earliest formulas for supply chain, like the economic order quantity, the Wilson formula, or the safety stock formula, for example, all those formulas implicitly work on a single SKU. You have a SKU with goods flowing in and goods flowing out, and you want to control what happens within this SKU. It’s very interesting because it gives you access to a literature of simple models that have been conceptually developed with the idea of having a SKU front and center.

Then, when you go into the reality of supply chain software, SKUs are everywhere. Most ERPs, WMS, and ecommerce platforms have notions of SKUs one way or another. It’s not just a concept that is useful to write equations, it’s also something that is implemented at the core of many supply chain systems. So, when you want to reason about it, it’s kind of handy when you think about SKUs. It just so happens that the software you operate has SKUs as well.

Kieran Chandler: We mentioned in the start that there are a few assumptions being made with SKUs. What are some of these assumptions?

Joannes Vermorel: The key assumption is that within the SKU, the individual units or physical products cannot be differentiated. That’s actually a fairly big assumption, and many supply chain practitioners are probably relying too much on this assumption for their own good.

Kieran Chandler: What kind of real-world assumptions are there?

Joannes Vermorel: It depends on the vertical, but let’s pick one for example: fresh food. What matters is that you have batches of products with expiration dates. So, if you look at the units that you have in a SKU, you can say, “Oh, I have 100 units in stock, and all is well because my demand is only 20 units a day, and so for tomorrow, I’m good. I have 100 in stock, and tomorrow, I will have 20 units of demand.” But what if out of those 100

Kieran Chandler: So, what we’re talking about here is the limitations of SKUs in the context of supply chain management. Joannes, can you elaborate on that?

Joannes Vermorel: Yes, I can. The issue with SKUs is that they work well for generic items, but as soon as you have items with a level of granularity, problems start to arise. For example, let’s say you have 100 units of an item, but 90 of those units are set to expire tonight. This leaves you with only 10 units in stock, but you’re expecting 20 units to be needed in the system. Clearly, this is not enough.

Kieran Chandler: Okay, so how does this lead to the issues you’ve observed?

Joannes Vermorel: Well, there are other issues with SKUs as well. For example, units cannot be assumed to be undifferentiated, even when they are all the same. The very notion of SKUs can be quite misleading. For instance, in the fashion industry, if you have a store with white shirts of a certain model, and you have three units left, you might think that you know when to replenish your stock. But the reality is that customers don’t walk into a store with the intention of buying a particular SKU. They have a need, and if that particular model of white shirt doesn’t fit, they might find a similar shirt that fulfills their need.

Kieran Chandler: So, what you’re saying is that SKUs are a way of cutting off your stock, but it’s important to be aware that it’s actually framing the way you look at the business.

Joannes Vermorel: Exactly. It’s important to be aware that SKUs might not be taking into account substitute products. You might have three units of one SKU and three units of another SKU, but they might be serving pretty much the same demand. In the end, you might have a large excess of stock, but it’s not very visible when you look at the stock at the SKU level.

Kieran Chandler: So, how do you deal with this problem?

Joannes Vermorel: One way to deal with it is to link all the SKUs together. The first thing to realize is that there are many alternatives. The problem is that when you start looking at the problem in a certain way, and you see that all the software that you interact with on a daily basis is taking this assumption front and center, it’s easy to think that there’s no other way. But there are many alternatives.

Kieran Chandler: So, there are plenty of alternatives for car parts, and you want to think of what is the unit of demand and what are the unit of service. What do you have to service the unit of demand, and you want to be agnostic of which part is being used because if they are all equivalent, it doesn’t really matter to think of different buckets, you know, different SKUs if in the end they all serve the same need. So, what you’re kind of driving at is that every kind of vertical should have its own kind of distinct idea of what a SKU is that works for that kind of vertical?

Joannes Vermorel: Exactly. I mean, SKUs are an abstraction, and like most abstractions, they are leaky. What you gain by having an abstraction, something simplistic like a SKU, is that it’s very simple and easy to reason about. It’s easy to write formulas and code that into your software. But beware, your abstraction is leaky. It’s a map, not the terrain. You can have a gap between reality and your abstraction, and the bigger the gap, the more you put yourself in danger of facing unintended consequences of this leaky abstraction.

First, you have to realize there is a gap and how large it is. It really depends from one vertical to another. For example, reasoning in terms of SKUs in aerospace is mostly nonsense due to serviceable and unserviceable parts, and the fact that rotor balls have flight hours or life cycles left in them. For fashion, it’s also mostly nonsense because there are so many substitutions. For food, it’s the same.

However, some industries, like Apple selling iPhones, it’s a fairly good approximation. People want the latest iPhone and will not think that another iPhone is a good substitute. In this case, the SKU is a good notion, and it’s almost like a promise from the brand that there is no difference whatsoever between one box of an iPhone and another.

Kieran Chandler: So, for these kinds of industries where you talk about it being nonsense, have they moved away from using SKUs or is it something that’s still very much in use?

Joannes Vermorel: The interesting thing is that many of the supply chain software have emerged from fast-moving consumer goods (FMCG) where the notion of SKU actually makes a lot of sense. For example, for Apple or Procter & Gamble, selling shampoo or detergent, SKUs are usually something that makes absolute sense for those types of businesses. But when you move to other verticals, it might make less sense.

I have seen that many companies went with the same sort of recipes even if their businesses required relatively different ways of thinking about stock, following the trail of companies like Procter & Gamble, which were pioneers in having software-driven supply chains in the 70s.

Kieran Chandler: So, why is the industry kind of so happy to stick with this approach then? Is it just because it’s ingrained in these ERP systems?

Joannes Vermorel: To say they are happy is a strong word. I don’t meet many supply chain practitioners who say their ERP is a beauty or a jewel, and they are so satisfied with it. Usually, that’s not the case.

Kieran Chandler: There is usually a lot of ambient discontent about the state of things in supply chains. Frequently, people say the software is not that good, sluggish, and painful to interact with. But it’s not naturally easy to pinpoint the problem. They might look at an old ERP and say the user interface is plain text and looks old. However, the real problem is not the black and white screen or the text-based user interface. The actual problem is the core assumptions built into the mental model of the software that generate so much friction for the business.

Joannes Vermorel: It’s something very fundamental, but it can be hard to think of because it’s in your face when you look at a screen that looks like a computer system from the 80s. It’s easy to say it’s old and not good. But the more difficult thing is to understand that the very concepts in the software are not the right ones for the way you should think about and reason about your own business. That’s more game-changing.

Kieran Chandler: So, what’s the advice for supply chain practitioners? How can they find the solution or what should they be moving to?

Joannes Vermorel: First, expand your horizon. You have your SKUs everywhere, but think if it really makes sense for both the production side and the demand side. If you can easily reconfigure your production, repackage differently, and preserve a large amount of agility, then you might want to think differently about how you’re organizing your production flows and transit. If your clients reason about your products in a way that is not SKU-based, you might want to have first-class concepts and software implementation that align with that as well.

Unfortunately, it’s hard to go into specifics because they vary enormously from one vertical to another. Nonetheless, my advice would be, once you’ve expanded your horizon, start assessing the software you’re looking at to see if they are a good fit. If you’re looking at a piece of software that has SKUs at its core and is rigidly built around that, don’t expect the vendor to be able to adjust that. It’s not something you can add as an afterthought. You’ll have to live with that forever, so make sure it’s a good choice built into the software; otherwise, it’s going to be an ocean of pain.

Kieran Chandler: So, to conclude today, the concept of SKUs is ingrained in all the systems we use. Is it something that’s going to be here to stay, or is this something you can see changing going forward?

Joannes Vermorel: I think the best way to think about SKUs is as training material. It’s a nice concept to introduce students pursuing engineering studies specializing in supply chain. You can introduce them to the concept of SKU and then introduce all the basic formulas that come along with it. It’s useful to give them a taste of what it looks like. But it’s just a way to draw a map, not the terrain. My advice would be to keep SKU as training material, but for business, keep your eye open to the reality of the business and what it actually demands.

Kieran Chandler: Great, let’s wrap it up for today. That’s everything for this week. Thanks very much for tuning in, and we’ll see you again next time. Bye for now.