Summary
Moderated by Kévin Baumann, Lead Supply Chain Scientist at Lokad, this roundtable brought together senior executives from the retail sector for an in-depth discussion on the challenges and opportunities of the modern supply chain. Sébastien Roux, Director of Operations and Projects, and Christophe Lapotre, Executive Director of Omnichannel, both from Maisons du Monde, along with Bertrand Renault, Supply Chain Director at IskayPet, shared their expertise alongside Joannes Vermorel, Founder and CEO of Lokad.
The discussion explored the practical challenges faced by Maisons du Monde and IskayPet, including the complexities of managing intricate networks and the impact of AI and probabilistic forecasting on supply chain transformation. Revisiting the “old paradigms” of the sector, the participants highlighted how Lokad’s Quantitative Supply Chain approach is reshaping operations. Change management, user adoption, and fostering trusted collaboration with Supply Chain Scientists were emphasized as key factors for building a robust and resilient supply chain.
Full Transcript
The Moderator: Good morning, everyone. We have a large panel today, full of insights and reflections to share, so I won’t monopolize the floor. I’ll simply read the title of this session: “Quantifying Uncertainty: A Roundtable Discussion by Lokad on Forecasting and Planning for the Modern Supply Chain with Maisons du Monde and IskayPet.” Gentlemen, there are many of you here today, so I’ll let you introduce yourselves and dive into this 45-minute discussion. Thank you very much.
Kévin Baumann: Good morning, everyone. My name is Kévin Baumann, and I’m the Lead Supply Chain Scientist at Lokad. Today, I have the pleasure of moderating this roundtable alongside Christophe Lapotre, Sébastien Roux, Bertrand Renault, and Joannes Vermorel. Christophe is the Executive Director for Omnichannel at Maisons du Monde, Sébastien is the Director of Operations and Projects at Maisons du Monde, Bertrand is the Chief Supply Chain Officer at IskayPet and has spent nearly 20 years at FNAC, and Joannes is the CEO and founder of Lokad.
As mentioned, the theme of our discussion today is “Quantifying Uncertainty: Forecasting and Planning for a Modern Supply Chain with Lokad, Maisons du Monde, and IskayPet.” We are fortunate to have experts from quite diverse retail sectors. I’ll start by asking a few questions, and then we’ll open the floor to the audience for any questions they may have for our panelists.
As I mentioned in the introduction, we have representatives from quite different sectors here. Before we delve into planning and forecasting, it’s essential to properly identify the challenges and issues we aim to address. I’ll start by turning to Christophe. Christophe, could you outline the challenges Maisons du Monde is facing that necessitate the use of forecasts?
Christophe Lapotre: Good morning, everyone. It’s a broad topic linked to a two-year transformation plan for Maisons du Monde. This plan aims to simplify our organizations and focus on a network of 336 stores across 9 countries. Establishing forecasting capabilities is a fundamental goal to address customer satisfaction and performance challenges. Today, the plan includes a decentralization aspect, adding a very precise global-local SIG touch to our store-level forecasting system.
It’s true that we’ve hit the limits of our current tools and organizational structures. Forecasting is critical in this two-year transformation plan within a high-volume model. With 336 stores and a fast-moving online platform, we need much finer projections to revisit the entire value chain. It’s about having the right stock in the right place and collaborating with Lokad in the weeks ahead.
Kévin Baumann: Very good. Bertrand, do the issues Christophe just mentioned resonate with what you experience at IskayPet?
Bertrand Renault: Yes, absolutely. For context, IskayPet operates 320 stores across Spain and Portugal, along with an e-commerce platform that holds a significant market share. Speaking directly to the challenges, in terms of forecasting, we make nearly 2 million decisions daily based on store assortments and the number of stores we have between Spain and Portugal. Thus, demand forecasting is essential, especially since our logistical capabilities are somewhat flexible but not as much as we’d like.
Additionally, we must account for costs and logistical capacity to deliver the best products to our stores at the right time.
Kévin Baumann: So, Christophe highlighted the store network, and Bertrand, you spoke about the number of daily decisions. Indeed, the larger the catalog and network, the more complex planning and forecasting become. To illustrate this, Sébastien, could you elaborate on the scale of your operations at Maisons du Monde? We’ve mentioned the store network, but what about the catalog, online platform, and warehouses?
Sébastien Roux: Good morning, everyone. To provide more detail, as Christophe and you mentioned, it’s 336 stores, primarily two warehouses in France. We operate across three distinct activities with vastly different lead times: small decoration items, large decoration items, and furniture. Lead times range from simple to double or even triple depending on the activity. We also handle B2B supply with home deliveries directly to customers and B2C supply for store replenishments. This illustrates the complexity of our daily operations and the challenges Lokad has faced while supporting us.
Kévin Baumann: Christophe, could you share more details about omnichannel commerce and the complexity it entails?
Christophe Lapotre: Certainly. Regarding omnichannel, thanks to our CEO François Melchior de Polignac, we’ve restructured. Previously, we were quite siloed, with sales channels running independently. Today, we have a website operating in each country and a marketplace, which is a major focus for Maisons du Monde. Managing 336 points of sale and clustering 15,000 active references is where the challenge lies. How can we dynamically adapt assortments across this network of stores and the omnichannel environment? That’s where we rely on Lokad’s expertise.
Our goal is to align these channels, something that wasn’t fully achieved before, by integrating a commercial and strategic animation plan. The challenge is twofold, or even threefold: to align all these channels effectively.
Kévin Baumann: Very good. Joannes, you’ve heard these challenges from Maisons du Monde and IskayPet. What’s your take on them?
Joannes Vermorel: Stock optimization within a distribution network is an old problem. Operational research from the 1950s and 60s began providing solutions, with paradigms emerging in the 1970s that are still widely used today. You’ll find many practitioners relying on these methods, like time-series forecasting, safety stock, and service levels.
From what’s been discussed so far, one might think this is what Lokad does, but that’s not the case—not at all. We don’t do time-series forecasting, safety stock, or service levels. We do something entirely different because, fundamentally, these ideas don’t work. I could delve into the details; I’ve even prepared a series of readings for those interested. We also have about 200 hours of YouTube videos explaining why these approaches don’t work.
The issue with these somewhat naive methods is that they don’t align with supply chain realities, particularly for distribution networks where demand is sparse, intermittent, and erratic at the store level. Lokad approaches forecasting from a probabilistic perspective, considering all possible futures—not just for demand but also for lead times. We’ve discussed how lead times vary by product, and this influences our approach.
For example, when we optimize for Maisons du Monde, we consider where the best payback in euros would be for each unit in a warehouse across all potential futures. Then we evaluate the next unit and so on, scoring every possible decision economically in euros, not percentages. This fundamentally different approach drives the optimization missions we undertake.
Kévin Baumann: Indeed, that’s a good description of Lokad’s philosophy. What also interests me is understanding how companies discovered this philosophy and what their journey was like to get there. Sébastien, what practices did you use before at Maisons du Monde to address the challenges you mentioned?
Sébastien Roux: Well, we operated in a very centralized model with expert teams, almost single-tasked, using somewhat outdated tools that we started to push to their limits. These tools weren’t necessarily suited for forecasting and anticipation. Furthermore, in this very centralized organization, we had many committees. Today, with Lokad’s involvement, we really wanted to simplify and streamline everything. And this was a real desire and expectation from the teams, as they’d been asking for it for years. We’re talking about millions of operations at our company, which has grown very quickly, and inevitably, at some point, our organization didn’t adapt as fast as it should have.
So now, we’re catching up. And today… how can I put it… the teams are very engaged, and the directions we’ve taken seem to be the right ones. At least, all the indicators are green.
Kévin Baumann: Bertrand, you spent nearly 20 years at FNAC. Do you have any lessons to share from your experience—any particular methods or practices you applied in Supply Chain?
Bertrand Renault: Yes, well, first, I believe there’s no magic formula, that’s clear. I think as maturity progresses, we start to understand which rules to apply. At one point, we experienced the industrialization of supply, meaning we began applying averages and a grid that wasn’t very precise in terms of coverage or safety stock. We talked earlier about minimum stock levels in stores, which in some cases weren’t necessarily aligned with merchandising rules.
So, I would say that at one point, we leaned towards industrialization to manage all this data in a certain way. Over time, we learned from the mistakes we made in demand forecasting. Moreover, I might be mixing concepts, but working capital has become increasingly important.
Thus, the room for maneuver is increasingly tight. In a way, with this learning process, we realize we’re tailoring more and more specifically—sorry, excuse me, I’m struggling to express myself in French. I used to speak one language well, and now I speak two poorly. Technically speaking, I sometimes struggle to find the right words. The tailoring we did before was more like four sizes that had to fit all store clusters. Today, with learning, we’re trying to make a tailored suit for each store size, truly adapting demand to the customer’s needs.
This means the customer provides the inputs, telling us whether a reference should be in the store and, if so, in what quantity. So, I think we started with industrialization and are now, in a way, stepping back by taking each case individually. As I mentioned earlier, in our case, we have to make nearly two million decisions daily.
Kévin Baumann: Joannes, you’ve just heard these two examples, these two stories. I think we’ve heard many others like these before. What comes to mind? What do you have to say about what you just heard?
Joannes Vermorel: To support the argument I’ve made, that traditional Supply Chain theories work very poorly? Most of my peers approach these problems by piling up errors, exceptions, alerts, and exceptions everywhere. You’re supposed to have a system automating your Supply Chain, but it ends up generating tens of thousands of exceptions per day wherever the digital recipe fails. Then you ask a human to come in and apply a manual fix. Essentially, the Supply Chain system uses operators as human co-processors.
You have the computer processor, and then the human co-processor to patch things up wherever it doesn’t work. My perspective is that if, for example, we take a store assortment problem, and we truly understand how to solve it—how to make this choice—then we can code it. If we can code it, we can implement this digital recipe for each store.
So, every time we rely on an operator, it’s a sign we haven’t fully understood how to solve the problem. The response of “I’ll make a decision when I face the problem” is not a real answer. It just means someone will make a decision somewhat randomly, without being able to justify it when the time comes.
Lokad’s vision is to fully automate these decisions, not with some out-of-the-blue AI that’s supposed to have superior understanding, but instead with people like Kévin. Now, Kévin hasn’t introduced himself in detail, but he’s part of our team of Supply Chain Scientists. These are people who interact with clients, understand the strategic goals at a granular level, and have the skills—if I may call it that—to translate high-level strategic visions into digital recipes that can operate at a massive scale.
That’s the expertise of Supply Chain Scientists. Fundamentally, the output is mechanization, and proper mechanization is something that works without creating chaos that requires constant manual intervention, which, to me, demonstrates poor digital methods, regardless of the approach.
Kévin Baumann: Indeed, let’s now delve a bit deeper into Lokad’s Quantitative Supply Chain philosophy. Sébastien, how has the journey been so far at Maisons du Monde? What details can you share?
Sébastien Roux: Well, after five months of collaboration since the RFP, we’re in a rich partnership with Lokad and have a strong presence from their team. We’re in the final testing phase, holding regular meetings and always having a point of contact, both on the technical side and in terms of evolving our workflows. This represents a profound organizational shift. Things are going well. Lokad has been very supportive with complex questions, providing simple answers. I don’t have much to add, but we’re satisfied and heading towards a highly fluid, resilient, and effective collaboration.
Kévin Baumann: Great. Christophe, how would you say the change management process has gone at Maisons du Monde?
Christophe Lapotre: It’s gone quite well, I’d say. As we mentioned earlier, we had obsolete tools and organizational structures that couldn’t keep pace with the transformation plan. Today, this will radically change the lives of our teams. To give you an idea, in Sébastien’s team, there are 16 people who used to spend 80% of their time crunching data that was difficult to translate into decisions. The goal is to transform these store planners into Business Analysts, enabling regionalized management and reconnecting with in-store teams.
As a retailer, my dream is for our teams to ask as few questions as possible. Today, we still hear things like, “I’m missing this,” or “I feel I’m not reaching my store’s revenue potential.” Tomorrow, the goal is for these questions to disappear, and for calls to Sébastien’s team to drop significantly. It’s a big change, and it’s been highly anticipated. Of course, with any change, there’s an adjustment period, but this shift brings responsibility to the forefront.
For us, this change is incredibly impactful. It connects store teams with headquarters to provide perfect responses to our customers every day and optimize decisions. The process is going well, and the company is ready for it. We took over the team in February this year, so changes are happening quickly. I echo Sébastien’s sentiment that the Lokad team has been extremely supportive, which is crucial for a project of this scale in the value chain.
Sébastien Roux: I’ll just add one thing. We’re retailers, and this shift also instills a retail mindset across the back-office team, which wasn’t necessarily the case before. As Christophe mentioned, we were really just crunching data without understanding the purpose. Now, we’re seeing a mindset shift, and that’s very exciting for us.
Kévin Baumann: Very well. Bertrand, you’ve been familiar with the concept of the Quantitative Supply Chain for some time now. What resonates with you about this concept?
Bertrand Renault: As I mentioned earlier, I believe we’ve been engaging with the Quantitative Supply Chain for quite some time. I’d even argue that it has always existed. However, the sheer volume of data that we now have to process has led us into a form of industrialization. This industrialization, as I mentioned previously, has caused us to make several mistakes, such as clustering stores or classifying rotations at a global level. In our case, we operate two brands in two different countries, with geographic variations between urban centers and more rural stores, requiring assortments to be optimized based on the local demand within the trade area.
In recent years, we’ve been fully immersed in industrialization, trying to implement merchandising at the store level and clustering. In some ways, this was already a form of Quantitative Supply Chain. However, today, we are pushing things much further by striving for a much finer granularity, which requires a genuine focus on data. Not only do we need to embrace data, but we must also provide support aligned with the business.
On one hand, it is crucial to integrate merchants effectively. This must not simply remain a Supply Chain matter; we must collaborate with all the stakeholders within the company, especially the commercial teams.
Kévin Baumann: Indeed. We’ve talked about the Quantitative Supply Chain and change management. It is essential to gain the support of end users; otherwise, the project is almost guaranteed to fail, regardless of how advanced the technology is. My question now to Sébastien and Christophe is, how would you explain the Quantitative Supply Chain to someone encountering the concept for the first time in just a few words?
Sébastien Roux: I’ll keep it very simple: it’s about having the right stock, in the right place, at the right time, in the right quantities. This serves both our clients and our collaborators in their operational tasks. In our stores, operational gestures are significant, and behind them lie considerations of productivity and ROI that we need to address. So, that’s my simple explanation.
Kévin Baumann: Bertrand, would you like to add anything to this question?
Bertrand Renault: I might simplify it even further by saying it’s about asking questions and listening to our clients. This might sound basic, but by “clients,” I mean both our end customers and our internal customers—the stores. When we talk about having the right stock at the right time and place, we must also consider the associated costs. It’s important to avoid a disruptive Supply Chain.
For instance, when listening to stores, they sometimes complain: one day, they receive 10 pallets, the next day only two, and then five more the day after. We need to account for this, ensuring not just the right quantities in the right place and time but also taking into consideration the operational realities and associated costs. This becomes especially critical when working with low inventory coverage and focusing on working capital, making us more vulnerable to unexpected events.
Kévin Baumann: Joannes, is that how you present the concept when speaking to various CEOs? Or do you draw upon other concepts?
Joannes Vermorel: My presentations might be a bit long, but I’d like to build on the examples to help the audience understand. I fully agree with the idea of serving the customer and aligning with their needs. I’d like to invite the audience to think about this. Most of you are probably familiar with classical Supply Chain theories: time series forecasting, safety stock, service levels, and so on.
Let’s take a very basic example from Maisons du Monde. For those who have visited their stores, you’ll notice that products play multiple roles. Some products are there to be sold, while others are there to help sell, creating a warm and appealing shopping environment. A time series forecasting perspective cannot account for the idea that one product helps sell another. It simply doesn’t fit within the formalism.
This is one reason Lokad’s approaches might seem atypical—because we aim to adhere to the customer’s perspective. To achieve this, we need to discard rigid formal frameworks that can’t express straightforward ideas, like creating a welcoming atmosphere and ensuring customers can leave with the items they want. It’s a small nuance, but I encourage the audience to reflect on it. You cannot achieve this with time series alone. If your software relies exclusively on such methods, it simply won’t work. It’s mechanical.
Kévin Baumann: Thank you. One final topic: today, there’s a lot of excitement around AI in the Supply Chain, and Lokad positions itself as a pioneer in this area. For us, one of the most important aspects is the working relationship between the client and the Supply Chain Scientist. Sébastien, you touched on this earlier. Would you or Christophe like to elaborate?
Sébastien Roux: Regarding AI, we have high expectations. This tool offers an opportunity to create a simpler, shorter connection between the end customer and the supply back-office. We used to work with complex probability models, but AI is taking us further by showing that we can achieve equally good results with fewer resources. It enables hyper-efficiency while preserving the commercial and attractive nature of our stores. It also ensures transparency for our teams, streamlining their daily tasks so they can spend more time with customers and provide a seamless experience.
Kévin Baumann: Excellent. As we near the end of our roundtable, I’d like to ask each of our panelists one final question. Christophe, what’s next for the Supply Chain at Maisons du Monde?
Christophe Lapotre: We’re currently in discussions. Our first goal is to complete the initial phase of downstream supply, finalizing testing in real conditions soon and expanding it to all areas at the start of the year. This is the first step in supporting our transformation plan.
We’re also working towards decentralizing roadmaps by country, tailoring strategies and concepts to specific clients in each region. Another major focus is leveraging the value chain with Lokad, both upstream and downstream, and activating other modules. We’re discussing the importance of pricing with Lokad, addressing this crucial challenge, and optimizing dynamic assortments. With 15,000 active SKUs outside the marketplace, our aim is to create the best possible assortment throughout the store’s lifecycle and across the year.
Lastly, we’re looking further upstream, all the way to purchasing. This involves integrating the entire value chain to break down silos and unite all stakeholders under one strategy. As you noted, cash flow management is one of the company’s most valuable assets, so accelerating the value chain is a key priority.
Kévin Baumann: Sébastien, do you have anything to add about how Lokad will support you in these transformations?
Sébastien Roux: We’re counting on continuing a productive collaboration, with both remote and in-person support, to onboard these new tools for assortments and potentially additional modules. So far, we’re very pleased and find Lokad to be highly supportive, which suits us well.
Kévin Baumann: Thank you very much, Sébastien. Bertrand, what’s in store for IskayPet’s Supply Chain in 2025?
Bertrand Renault: First of all, of course, continuous improvement, and I would emphasize two other points. The first is the management of promotions, which still gives us quite a headache. To give you an idea, around 40% of our sales are based on promotional sales, both in-store and online. Managing this demand is something we still struggle with, and, in fact, we’re still working on it today using Excel. So, we really need to focus on this issue because it has a significant impact on stock and sales. The third point is the opening of a platform in Portugal. We are currently working on the assortment that needs to be stocked locally to serve the stores in Portugal.
Kévin Baumann: Perfect, thank you very much. Thank you to all of you—Christophe, Sébastien, Bertrand, Joannes. It’s been a real pleasure to discuss with you today. We’ll now move on to audience questions. Feel free to raise your hand, ask your question, and let us know whether it’s directed at the panel in general or someone in particular.
The Moderator: So, will this stimulating presentation spark questions from you? It’s true that some practices were called into question, particularly by Joannes. If you felt challenged in your practices or tools, whether to defend your vision or otherwise, don’t hesitate to ask questions.
Well, maybe it’s me, though I might be the least relevant since I’m not a practitioner. But during your initial remarks, Joannes, after dismantling traditional indicators, you mentioned profitability, highlighting it as a central idea in business activities. Could you elaborate on this and explain how it could more fundamentally shape the way supply chain operations are structured and approached?
Joannes Vermorel: The classical vision of the supply chain optimizes percentages—service rates, essentially. The big illusion is that these percentages have absolutely no correlation with the return on investment you can achieve. None whatsoever. Whenever you see percentages, it’s extremely satisfying because percentages allow everyone to avoid risk. They create opacity and enable bureaucracies to thrive.
Percentages go up, percentages go down—it’s very convenient if you want a well-entrenched bureaucracy in a large company. You create KPIs everywhere based on percentages, and you can be sure that there will never be any ROI, and even if you lose a lot of money, no one will notice.
At Lokad, we financialize everything—not because we are financiers, but because supply chains are incredibly complex. There are many forces at play: inventory costs, stockout costs, transportation costs, MOQ requirements, price breaks (buying more for a lower unit cost), shipping decisions (air freight is costlier but faster), capital costs (how long you need to hold inventory), and so on. The only way to balance all these elements is to express everything in monetary terms.
This doesn’t mean adopting a simplistic, short-term financial vision where you make serious mistakes by looking only a month ahead instead of three years ahead. For instance, the cost of customer relationships, especially for a brand like Maisons du Monde, which is highly valued, is something built over decades. So, you need to take a long-term view. Financial indicators must be thoughtfully designed and long-term oriented. But ultimately, you need to express everything in euros to strike the right balance. That’s what Lokad does. And the surprising aspect is that this approach disrupts all these lukewarm, percentage-based classical indicators, where it’s percentage versus percentage, and in the end, no one has a real opinion about anything.
The Moderator: Well then, let me turn to your clients. Is this clearly part of the initial appeal when embarking on a transformation plan? Does this play a role in the decision to work with Lokad?
Sébastien Roux: Yes, and what we appreciated most was avoiding being drowned in averages, which is often our default lens. What we loved about Lokad is their focus on very specific examples at the reference level, tracing the issue back to its source. Let me give you a simple example: we had a store with a series of vases. This store sold one vase, but we sent them seven without fully understanding why. Lokad helps us trace the value chain to see why we should perhaps have sent only two and why the error led to seven or eight.
What we liked was the ability to deeply analyze, understand, and act accordingly. In this example, Lokad helped simplify the approach, work on concrete cases, explain them, and naturally bring in a return-on-investment logic. That was a key point when we launched the RFP. What resonated most was being simple, effective, and making it tangible for 100% of our employees.
We discussed this a lot with our sales team. A salesperson needs to understand what’s happening. Averages and overly complex data don’t allow us to grasp the reality on the ground. We’re in a straightforward business—we have customers, we open stores, and we sell. We need to understand all the flows perfectly. Simplifying the approach and focusing on very specific cases at the reference level rather than averages—that’s what we appreciated.
The Moderator: To make a joke—maybe the first vase was broken. That brings us to the challenges of not coding but translating—the translation of ground realities into tools. You mentioned this earlier, how the role of the Supply Chain Scientist is to translate reality. Kevin, could you elaborate on your role?
Kévin Baumann: Yes, exactly. The broken vase example doesn’t apply at Maisons du Monde, but we do need to understand products in terms of their economic opportunities and costs. Let me give two simple examples. For plates, which don’t take up much space, we’d prefer to sell them in sets of six rather than individually because people buy full sets. Even if the stock seems high based on forecasts, we always aim to keep sets of six available.
A stack of plates, exactly. Conversely, for something bulkier like a mirror, there may be good sales potential, but stocking two or three mirrors might last four weeks, which isn’t unreasonable. However, the space they take in-store incurs logistical and storage costs.
Bulky items take up space that could be used for other products. For such items, we don’t necessarily want to push too far because they cost more to stock, both financially and in terms of available space.
The Moderator: Plates and mirrors break too. I’ll stick with this example for now. And there’s a question from the audience.
Audience Member: Thank you for the presentation. I have two questions. One for Christophe, Sébastien, and Bertrand: how do you manage change management? I imagine that if Lokad is aiding decision-making, some employees might feel deprived of their ability to make decisions. How do you handle this aspect of change management? And for Joannes, a question on your business model: how do you approach this topic? I understand there’s a mix of service and software—how do you position yourselves to support clients on the underlying economic model?
Christophe Lapotre: Thank you. Regarding the team aspect, today, indeed, one might wonder if the teams feel disempowered. Not at all. On the contrary, they now have more time to understand and analyze Lokad’s proposals. They have actionable levers where they can adjust the settings. And indeed, we’re also engaging with machine learning, meaning we’re fine-tuning the tool so that it learns a bit about our acceleration and deceleration logics.
We talk about offer localization. It’s also about acknowledging that consumption habits differ between the south and north of France, as do the products. So, we’ve organized our teams of Business Analysts by region to better respond to local demand and adapt the tool to observed discrepancies. As part of their work today, instead of spending a lot of time handling data, tables, percentages, dashboards, and the like, they focus more on human intelligence—what they recommend for stores and the corrections to apply.
That’s what’s interesting and why they’re so enthusiastic about the change, ultimately becoming business partners for the stores. Previously, they were purely operational. Now, they have regular exchanges with the stores to understand what they did before, why it worked or didn’t work, and to propose new ideas. So, they’re stepping out of their usual scope. Of course, this challenges preconceived notions and can be intimidating, as they leave their comfort zone where data manipulation was straightforward and averaged out outcomes were acceptable.
But now, they operate within a specific scope, and since we benchmark these scopes based on economic performance, it puts them in a different position, which is very exciting for them. When I say “they,” it’s because there are many women on the team. For us, though, it’s great because, with precise actions in one area, we can immediately see the added value we’ve created. In a highly centralized, heavily averaged system, such value gets lost in the volume.
We’re just starting, but so far, it’s going very well. And I must say, they’ve embraced the model. Of course, we still have big discussions, as Joannes mentioned. We still want to reassure ourselves with tables, dashboards, averages, and service levels. But we’re starting to move beyond that. It’s true that it was initially a bit unsettling—and it still is, to some extent.
Joannes Vermorel: Regarding Lokad’s business model, it’s quite simple. We have a flat monthly fee that includes Cloud platform costs and Supply Chain Scientist services. What’s interesting is that all our digital recipes are accessible in terms of code to the client. This means that everything the Supply Chain Scientists develop—there are no black-box algorithms—is fully accessible to the clients. That said, our clients aren’t always interested in diving into the code.
We also provide something called the JPM (Joint Press Manual), which is a document written by the Scientists to facilitate the handover between Supply Chain Scientists. This document explains the rationale behind why we digitally coded certain aspects and how we did it. It explains the reasoning behind the modeling. Modeling involves choices based on our understanding of the client’s strategy, and there are many subjective decisions involved. If the client’s strategy evolves, the adapted digital model evolves as well.
The idea is that the client has access to both the code and documentation—not just what the code does (since you can read the code for that), but why it was done the way it was. This is all part of a SaaS model.
The Moderator: Perhaps another question from the back of the room over there?
Pierre F.: Yes, Pierre F., Léon Consulting. Thank you for the presentation. Joannes, you mentioned the approach of financializing the Supply Chain instead of using KPIs. We see that financialization is relatively straightforward for stock or transport costs, but for some topics, such as the cost of a stockout in-store, it can be more complex. A stockout might mean the customer buys nothing, chooses a similar product, or decides to return to the store later. How can we financialize the cost of a stockout in such situations?
Joannes Vermorel: Once again, classical Supply Chain theory tends to dodge such complex problems, pretending they don’t exist. This is true for issues like cannibalization, substitution (which are difficult), or stockout costs. It’s also true for the opposite problem, like the cost of discounts. For instance, you launch a promotion, train your customer base to expect promotions, and it takes years to establish and years to recover from.
Returning to stockout costs, my view is that it’s better to be approximately correct than precisely wrong. For stockout costs, you can start with a rough estimate, discussing with the teams to get their initial opinions. From there, you can use these initial assumptions to run decision optimizations. What’s interesting is that these economic assumptions prove incorrect when they lead to nonsensical decisions.
This provides a feedback loop: you test the assumptions by generating decisions based on them. If the resulting decisions seem absurd, it’s clear the assumptions were flawed. That’s the starting point. Later, with more time, if there’s access to loyalty cards or checkout data, you can analyze shopping baskets and statistically evaluate what happens—whether you lost customers or if they purchased less. But that comes later.
Generally, in my experience, starting with stockout costs involves discussion, rounds of experimentation, and generating decisions to calibrate and ensure you’re on the right track. Once operational, with more time, you can delve into analyzing shopping baskets, loyalty impacts, and so forth. It’s a staged process.
The Moderator: Thank you. We may have time for one last question.
Not quite. In that case, I’ll ask a classic: how did you first encounter Lokad? It’s true, a player like this, who breaks the mold and habits—how do you find them? This applies to Maisons du Monde and IskayPet as well.
Bertrand Renault: In our case, it was by chance. We conducted a market review, met with various stakeholders, and for me, the solution seemed fairly obvious, especially since the initial scope was quite limited.
In terms of investment and CAPEX, we were looking for something lightweight and manageable that would let us get started quickly. That’s exactly what we found, and it worked out very well.
Christophe Lapotre: For my part, it goes back to my previous professional life. At Celio, Lokad was already present. I participated in the tender process at the time. It was more than a winning move in my prior experience, so it felt natural to propose it to my colleagues in the commerce team at Maisons du Monde.
We then went through a natural tender process, and Lokad prevailed without any favoritism—I promise. But for me, it was already proven by example, having seen it transform and deliver a winning model.
The Moderator: Well, thank you all. We’ve reached the end of this discussion, but that doesn’t mean the conversation can’t continue at the booth, of course. I noted the location somewhere.
Joannes Vermorel: It’s just at the entrance, near the cloakroom.
The Moderator: Exactly. I saw it—it’s a practical indication, not expressed in terms of aisles or numbers, but it’s right at the entrance. Thank you all.
