00:00:03 Sales and Operations Planning (S&OP)
00:00:28 S&OP’s role in complex supply chain management
00:02:32 The origins and evolution of S&OP
00:03:19 Benefits of quantitative supply chain management
00:06:18 Pros, cons of S&OP and quantitative management
00:08:01 Forecasting complexity in S&OP: need for automation
00:08:29 Poor S&OP alignment and economic impact
00:09:02 “Sandbagging” in sales forecasts and S&OP
00:12:19 Successful S&OP implementations and business complexity
00:14:05 Tech potential in S&OP, quantitative management
00:16:02 Inventory management: financial implications and growth
00:17:14 Stock-out cost agreement’s importance among departments
00:18:14 Persisting outdated Sales and Operations Planning
00:19:03 Efficiency of the Quantitative Supply Chain
00:21:49 Improve processes, maximize automation
Summary
In a discussion with Kieran Chandler, Lokad founder Joannes Vermorel is examining Sales and Operations Planning (S&OP) and its significance for supply chain optimization. Vermorel is describing S&OP as a vital process for aligning internal operations with external market forces, stating it was originally developed by 20th-century FMCG companies. While traditional S&OP focuses on improving management and communication, Lokad’s quantitative supply chain approach is emphasizing data-driven solutions. This approach is automating forecasting for vast product ranges, producing probabilistic predictions that are enhancing efficiency. Vermorel further explores the concept of S&OP 2.0, advocating for automation and aligning economic drivers rather than market states. His advice to CEOs is underscoring computer-aided management of granular supply chain data and continuous improvement.
Extended Summary
Kieran Chandler, the host, and Joannes Vermorel, the founder of Lokad, engage in a discussion primarily about Sales and Operations Planning (S&OP) and its relationship with Lokad’s supply chain optimization approach.
S&OP, as Vermorel explains, serves as a crucial business process that executive leadership teams employ to ensure efficient resource management, especially in large companies with extensive supply chains. He emphasizes that it strives to foster alignment among all departments in a company to streamline the delivery of goods in line with market demand.
Vermorel identifies the birth of S&OP with large, fast-moving consumer goods (FMCG) companies of the 20th century aiming to cater to the mass market. These companies devised processes that later amalgamated into S&OP, representing best practices in supply chain management.
However, Vermorel highlights a nuanced shift in perspective when contrasting traditional S&OP with Lokad’s quantitative supply chain approach. While S&OP is rooted in the belief that supply chain challenges can primarily be addressed through better management and improved communication, Lokad’s methodology prioritizes quantitative and computational solutions.
Despite different philosophies, Vermorel concedes that alignment, a central challenge tackled by S&OP, remains a significant concern in Lokad’s quantitative supply chain approach.
Vermorel kicks off by shedding light on the difficulties that large companies grapple with when managing vast product catalogs. Large companies with thousands of product references make it unfeasible for any individual or group to maintain a mental image of what will happen for every single product reference. Therefore, individuals often concentrate on aggregated information, such as the product category level, leading to fragmentation.
Vermorel introduces the quantitative supply chain as an alternate lens to address these issues. The quantitative supply chain automates the forecasting process, creating a probabilistic forecast that can be shared company-wide daily.
Host Kieran Chandler segues the conversation to SNOP, a process involving all departments in problem-solving, offering a holistic perspective. Chandler questions whether the quantitative supply chain might engender a supply chain bias. Vermorel counters this, stressing that the quantitative approach treats the future state of the market as a given, generated by machine learning algorithms.
The discussion veers towards alignment within a company, specifically the potential for disagreement on economic drivers. Vermorel acknowledges this issue, outlining that different factions within a company often have varying incentives, leading to misalignment.
Vermorel goes on to delineate the shortcomings of conventional S&OP when navigating large, intricate supply chains. He advocates for a greater reliance on automated processes for tasks like raw calculations and forecasting.
Vermorel introduces S&OP 2.0 or “Quantitative Supply Chain,” where the focus during meetings shifts to aligning economic drivers instead of trying to predict future market states or demands, which he believes should be automated.
Near the end of the discussion, Vermorel offers advice to CEOs seeking to enhance their internal processes. He suggests an emphasis on status quo improvement and better use of computers to manage granular information about the supply chain status. This shift, he believes, can circumvent scalability issues and enable employees to apply their high-level intelligence to solve complex problems.
Full Transcript
Kieran Chandler: Welcome to Lokad TV. This week, we’re going to be discussing Sales and Operations planning, more commonly referred to as SNOP. This is a business process where the executive leadership team comes together to decide how to best manage a company’s resources. While it’s not a particularly new concept, it is rarely well understood and can be very difficult for a business to implement. So, Joannes, I guess a good place to start is exactly what is SNOP? Could you describe it in a little bit more detail?
Joannes Vermorel: Absolutely. SNOP is the answer to the need for complete alignment in large companies that are running with a large supply chain. It means everyone - from marketing, sales, logistics, warehousing, production, purchasing - needs to be aligned to deliver the goods that the market seeks. This includes complete internal alignment and also an alignment with external forces, essentially the demand as observed on the market. Without this alignment, you might end up in a situation where the sales team is selling something that you can’t produce or you’re producing things that you do not sell. So, alignment is essential, both internally and externally, to meet market demand.
During the 20th century, large consumer goods companies invented ways to serve mass markets with this internal alignment that also happened to be in line with what the market demanded. This industry collectively invented a series of processes which has been crystallized under the name SNOP, representing an ideal of best practices in this respect.
Kieran Chandler: So, it’s basically a way of getting a lot of internal departments to communicate better. Where did this concept actually come from?
Joannes Vermorel: I believe it was crystallized by former executives turned consultants who started to have a well-rounded recipe to implement SNOP. So, we’re talking about consulting materials on how to replicate the processes that had been set up in some large companies into other companies. In the end, it was really a better way to manage the supply chain from a strategic management mindset.
Kieran Chandler: That’s kind of similar to what we’re doing here at Lokad. We’re talking about things as a whole business process, improving communications, and so on. You said “so far so good”, what’s not so good?
Joannes Vermorel: Well, it’s subtle because there’s a paradigm shift when you go from SNOP to what we do at Lokad, which is quantitative supply chain. The core of the challenge has shifted. The core focus of SNOP is that supply chain problems can essentially be solved with better management and better practices in the way different divisions in your company communicate and share information.
The perspective of quantitative supply chain is very different. The perspective is that information is primarily flowing from one computer to another, from one machine to another. It doesn’t involve as many people. If you have ten products that you’re producing and selling, you can bring people to a room and then they can manage. However, with the rise of digitalization and the increasing complexity of supply chains, this becomes more challenging.
Kieran Chandler: We’ll discuss what is the most likely outcome for the demand for certain products. For example, let’s consider a small company with just ten products. The shareholders can reasonably form an accurate idea of what the future holds for each one of those ten products. Now, let’s consider a typical situation for a large company nowadays, where it’s not ten products, but 100 thousand product references. Suddenly, it’s completely inhuman; nobody can hold in their mind what is going to happen for every single one of those references.
Joannes Vermorel: Indeed, you end up with a situation where this process, where we are supposed to communicate with each other, becomes fundamentally very difficult. People tend to focus on information that is more aggregated, for example, at the product category level. However, this approach leads to issues of fragmentation, both vertical and horizontal. This situation creates problems of its own. The quantitative supply chain primarily addresses this issue by taking a different perspective on the entire supply chain challenge.
Kieran Chandler: I understand. The real benefit of a Sales and Operations Planning (S&OP) process is getting all of the departments to contribute to the problem and looking at things through the eyes of every single department. But isn’t the problem with a quantitative supply chain that you’re just looking at it with a real supply chain focus, essentially having a supply chain bias?
Joannes Vermorel: Yes and no. The question is, if you bring people together, what are you going to agree on? From an S&OP perspective, people want to agree on the future state of the market, the demand to be served, and create alignment of everyone on that. The perspective of a quantitative supply chain is different; the state of the market is given by the machine, it’s kind of a given.
If you have tens of thousands of different product references, you want to have a forecast that is completely automated. This forecast can then be shared with everyone in the company automatically, every single day. So, you don’t need to agree on that. But what if the forecast is not accurate? That’s another problem that is also solved by probabilistic forecasts, where instead of having a single number for the future, you have probabilities for all the possible outcomes.
From our perspective in the quantitative supply chain, there is no point in bringing people together to agree on the forecast because the forecast is way too detailed. We are talking about tens of millions of numbers, hundreds of thousands of product references where you have probabilities for every single day ahead for up to a year.
This massive number of probabilities needs to be forecasted and all of that has to be done automatically. So unlike S&OP, if you bring people together in a quantitative supply chain, it is not to agree on the forecast; this is a given delivered by the machine.
Kieran Chandler: Okay, but we talk about those economic drivers in the context of the quantitative supply chain, and we talked about having agreement on that. If you’re in a company which already has poor alignment in their S&OP processes, won’t they just have poor alignment and disagreement when it comes to those economic drivers?
Joannes Vermorel: Yes, and actually, one of the key reasons why S&OP is so difficult to implement is that different parties do not have the same incentives. When it comes to forecasting, they might even have adversarial incentives.
Kieran Chandler: I’ve spoken with many people who have worked in large companies implementing Sales and Operations Planning (S&OP). They all practice something called sandbagging. Could you explain what sandbagging is?
Joannes Vermorel: If you’re part of the sales team, your forecast effectively becomes your sales quota that you need to reach in order to achieve your bonus. As an employee, it’s in your best interest to forecast a low number so that you can easily exceed your target, surpass expectations, and secure your bonus. The original idea was to have the sales forecast originate from the sales teams, as they were thought to be closest to the market. They were, therefore, deemed responsible for producing the forecast. However, they have every incentive to get it wrong. This runs counter to the objective of accurate forecasting.
Kieran Chandler: Is there a way to address the sandbagging issue? Sales teams often respond well to bonuses. Could introducing a bonus for more accurate forecasts improve the situation?
Joannes Vermorel: While that might sound like a solution, it actually creates another problem. The easiest way to maintain an accurate forecast is to have very low expectations. A salesperson might stop closing deals once they exceed their sales quota to avoid losing their accuracy bonus. This wastes market opportunities. In practice, rewarding forecasting accuracy often rewards mediocrity. Additionally, the skills that make someone a good salesperson, like understanding the client and the market dynamics, do not necessarily translate into accurate statistical predictions. Therefore, your best salespeople may not be the best forecasters.
Kieran Chandler: If we agree that S&OP is challenging, are there any real-world examples of companies that are succeeding with these processes?
Joannes Vermorel: Yes, typically, large Fast Moving Consumer Goods (FMCG) companies have relatively good S&OP processes in place. They benefit from a relatively narrow range of products. S&OP works as long as the amount of information that needs to be transferred between individuals is not overwhelmingly large. It is effective if you have a few hundred product references at most. However, as the complexity of your business increases, or if you need to coordinate across many geographic locations for worldwide scale coordination, it starts to fall apart. For instance, you might have only 100 products, but if you’re coordinating across 80 countries, you now have 8,000 product-country combinations to consider. This doesn’t fit into the mind of a human. If you decide to organize your supply chain in silos with one unit per country, you lose worldwide coordination, which can lead to inefficiencies, such as having too much of one product in one country and too little in another.
Kieran Chandler: We’re now living in a technological age. Can we not just replace the human mind with computers and still successfully use these SNOP (Sales and Operations Planning) processes?
Joannes Vermorel: The essence of supply chain is indeed about using automated processing for calculations. However, it’s not about completely replacing human intervention. We can automate the raw probabilistic forecast, but we still need people to agree on the state of the market, which, for most markets, is better left to an automated process. This process is developed by engineers who are experts in both software and statistics. It’s not the machine working independently, rather, it’s about engineering a process to deliver statistical forecasts at scale.
SNOP 2.0, or what we see as the future of supply chain, involves bringing people together to agree on economic drivers, such as what brings money into the company and what costs the company money. For example, if you invest in extra inventory, it’s money that you cannot use for other purposes, like expanding your logistics capabilities. This is a cost to consider. Another example is the cost of not serving a client, or the cost of a stockout. It’s difficult to assess, but there needs to be a shared agreement on what these costs entail.
For instance, the marketing division might even generate a stockout with some probability if they run operations that the supply chain cannot handle. Therefore, there needs to be an agreement on how much it costs when this happens. This way, each division can balance the reward-risk ratio for every action they take.
Kieran Chandler: It sounds like there are a lot of reasons to move on from these outdated SNOP processes. So why are companies still implementing and using them?
Joannes Vermorel: SNOP has a lot of basic common sense elements that should not be discarded. For instance, seeking high-level alignment and avoiding internal conflicts in your company is still beneficial. SNOP also emphasizes that the CEO should be the one holding the company together and ensuring there’s a shared vision, which makes sense. There are a lot of things about high-level common sense and probably best practice for companies in SNOP that are still good.
Kieran Chandler: Information is supposed to flow between people. What kind of agreement should we be seeking in this scenario?
Joannes Vermorel: Historically, the agreement was on predicting the future state of the market demand. You start with a sales forecast, establish your demand planning, then your supply planning. After that, you consolidate all your manufacturing plans and iterate back on this loop monthly. Here, the idea is that everyone works together to establish a shared vision of the demand. However, I believe this process should be entirely automated.
Kieran Chandler: Can you explain more about this automation process?
Joannes Vermorel: Sure, there still needs to be alignment, but the work in progress is on the economic drivers. The good news is that we switch from a situation where people’s time was consumed to agree on the demand for the next three months. This process had to be repeated monthly, which essentially consumed people’s time in meetings that supported the assembly process. This is not a capitalistic process as the time of the people involved is consumed to support a synergistic supply chain. At Lokad, we have a different approach.
Kieran Chandler: What’s Lokad’s approach in this context?
Joannes Vermorel: We believe that when people meet, it should be to improve the economic drivers, which are then implemented into the daily data pipeline. This pipeline crunches the data from beginning to end to generate all the decisions, especially purchasing decisions or manufacturing decisions, and inventory movement decisions automatically. Therefore, when people meet, it’s to develop a better strategy that is immediately turned into software logic, which then runs automatically. This ensures that people’s time, the scarcest resource in a company, is used in a way that is highly capitalistic and capitalized over time.
Kieran Chandler: That sounds quite progressive. As we wrap up, if I’m a CEO looking to improve my internal processes, what would you recommend? Would you suggest an S&OP (Sales and Operations Planning) process combined with a quantitative supply chain? What advice would you give?
Joannes Vermorel: I would recommend starting with the basics. Reflect on how you use the time of your management. Do your managers spend their time maintaining the status quo or improving upon it? This basic question is a good starting point. Then, consider the role of computers in information flow. Do you expect the information to flow from people to people, or from machine to machine? Especially when it comes to granular information about the exact status of your supply chain - every single count of how many units or parts you have at every location worldwide, or how many grams of raw materials you have at every location.
Kieran Chandler: So, the goal is to leverage technology for efficiency?
Joannes Vermorel: Exactly. If you expect this granular information to flow from people to people, you have a scalability issue. So, rethink what can be delegated to machines to a large extent. This allows your people to exercise their skills and high-level intelligence at solving problems that actually require high-level intelligence. Instead of having people doing semi-manual data crunching with Excel sheets, delegate these tasks to machines.
Kieran Chandler: Sounds like a progressive approach. Any final thoughts?
Joannes Vermorel: I think when you carry on with this thought process, you end up with something that pretty much looks like a descendant of S&OP, something very similar to a quantitative supply chain. Of course, these are just names, but the idea is there.
Kieran Chandler: Great! Hopefully, today’s discussion goes some way to solving a few turf wars within companies. That’s everything for this week. Thanks very much for tuning in and we’ll see you again next time. Bye for now.
