00:00:07 Challenges in food industry supply chain and technology adoption.
00:01:14 Ancient origins of food trade and globalization.
00:02:04 Challenges in data management for the food industry.
00:03:48 Diverse types of food supply chains and their unique challenges.
00:06:02 Managing expiration dates in hypermarkets and ERP systems.
00:08:04 Data noise and complications in fresh food supply chains.
00:10:34 Quality control and the uncertainties of multi-sourcing in the food industry.
00:12:54 The auction mechanism and its impact on supplier selection.
00:13:09 The challenges of forecasting promotions in hypermarkets.
00:15:10 Promotion optimization as a negotiation between retailer and supplier.
00:18:32 How traditional retailers are underutilizing collected data.
00:19:44 Amazon entering the fresh food market and challenges they face.
00:22:38 Partnerships and acquisitions to overcome supply chain challenges.
00:22:59 Fresh food industry’s unique challenges and need for better tools.


In the interview, Kieran Chandler and Joannes Vermorel discuss challenges in the food industry, focusing on supply chain optimization and technology. They cover issues like the lack of standardization, diverse supply chains, global sourcing, and perishable item management. Vermorel highlights the unique nature of the fresh food industry, where poor supply chain management can lead to rapid waste. He explains that traditional software tools have struggled to offer effective solutions, with many companies only upgrading from paper-based systems to Excel. Vermorel hopes Lokad’s technology can provide better solutions for the fresh food industry’s supply chain management.

Extended Summary

In this interview, Kieran Chandler and Joannes Vermorel, the founder of Lokad, discuss the challenges faced by the food industry, particularly in terms of technology and supply chain optimization. Vermorel notes that the food industry is both ancient and large, with practices that sometimes date back centuries. Due to its age, the industry has developed its own solutions to various problems, even before the advent of modern technologies like the internet. However, this has also led to unique challenges when compared to other supply chains.

One of the main challenges in the food industry is the lack of standardization, particularly when it comes to packaging and barcodes. While modern supply chains are characterized by barcodes and packaged products, the food industry often deals with raw materials measured in quantities like kilograms or pounds. This can make tracking and managing inventory more difficult, as there is no standardized method of identification.

Another challenge faced by the food industry is the diversity of supply chains. Under the broad term “food,” there are many different types of products and supply chains, ranging from fresh produce to meats. These products often have varying levels of production locality and traditional methods of sale, further complicating the supply chain process.

Fresh produce and meat, in particular, are typically sold by weight and do not come with barcodes or packaging until the very end of the chain. This makes it difficult to track and manage these products in a standardized way. Additionally, certain products, like meat, still require human interaction for sale in many markets, where a consumer requests a specific cut of meat from a butcher.

The global nature of the food industry also adds complexity to the supply chain, as products are sourced from all over the world. This international supply chain results in a wide variety of products being available to consumers, but it also means that companies in the food industry must navigate numerous different supply chains and adapt to various local practices and regulations.

They cover the challenges faced in managing perishable items, particularly fresh and frozen foods, and how these factors affect the supply chain.

The interview begins with a discussion about the unique challenges faced in handling frozen food. Warehouses for frozen foods are extremely cold, which makes it difficult for workers to operate in these conditions. As a result, many companies are moving towards end-to-end robotization to minimize the number of people needed to work with frozen foods.

Next, they discuss the management of expiration dates, particularly for fresh food in hypermarkets. Many ERP (Enterprise Resource Planning) systems have a crude vision of stock management and do not take expiration dates into account. This creates complexities in the supply chain, as the stock needs to be managed at a more fine-grained level to ensure proper handling of perishable items. Another challenge comes from customers who may misplace items on shelves, causing stock discrepancies.

The conversation then shifts to complications arising from the supply side of fresh food. Retailers are dependent on suppliers who are, in turn, reliant on factors such as harvests and weather conditions. This makes it difficult to maintain a steady supply of fresh food. However, fresh food typically has a high rotation rate, which helps with statistical forecasting. The main complication lies in the unreliability of suppliers in terms of lead times and quality control.

Quality control is a significant concern for retailers, as they need to ensure that the products they sell are of acceptable quality. This may result in additional uncertainty, as some products may not pass quality checks and need to be discarded. In cases where a supplier is problematic, retailers may need to resort to multi-sourcing from different suppliers or even participating in auctions to secure the necessary goods. This can lead to further uncertainty in terms of pricing and supplier consistency.

He highlights that retailers initially approached promotions as a forecasting problem, trying to predict the demand uplift. However, Vermorel argues that promotions should be seen as a negotiation between the retailer and its suppliers. Instead of focusing solely on accurate forecasts, retailers should consider optimizing the outcome of these negotiations.

Vermorel then talks about the vast amount of data that retailers collect through loyalty programs. This data can provide detailed information about customer preferences, habits, and even personal details. However, he notes that, disappointingly, most companies do very little with this data. They may occasionally use it for simple direct marketing campaigns, but generally, they are not utilizing it for supply chain optimization.

The interview also touches on the topic of Amazon entering the fresh food market through Amazon Fresh. Vermorel believes that Amazon is facing an uphill battle in this area, as it must compete with the established physical infrastructure of traditional retailers. Food supply chains present unique challenges due to the potential for mess, breakage, and hazards. Amazon will need to find solutions to these mundane problems while maintaining its usual mix of high tech and low prices to gain a competitive edge.

Vermorel notes that food supply chains can be thought of as many different worlds, including frozen food, dry food, produce, and more. He observes that companies like Amazon have tended to focus on specific segments within the food supply chain, rather than attempting to address all aspects simultaneously. This approach may be necessary for Amazon to compete effectively in the complex and challenging world of food retail.

They explore how the industry is lagging behind in technological advancements compared to others. Vermorel highlights the unique nature of the fresh food industry, as poor supply chain management can lead to rapid waste and loss of merchandise. This has resulted in a low tolerance for inefficient tools within the sector.

The conversation touches on a partnership between an unnamed French company and Monoprix, a well-established food retailer, to explore how they can bring a technological edge to the supply chain. Vermorel explains that traditional software tools have struggled to provide effective solutions for the industry, with many companies only managing to upgrade from paper-based systems to Excel-based ones, which still lack the required efficiency.

Vermorel expresses his hope that Lokad’s technology can offer better solutions for the fresh food industry’s supply chain management, though the interview ends without a clear conclusion on this point.

Full Transcript

Kieran Chandler: Today on Lokad TV, we’re going to discuss exactly why this is and understand what some of the challenges are that must be overcome in order to ensure the variety of fresh produce is available in supermarkets every day. So Joannes, what is it about the food industry that makes it quite so particular?

Joannes Vermorel: First of all, it’s super large and ancient, so those two combinations are relatively rare. There are some industries like cars that are super large but they are not ancient, I mean they are one century old basically. Here, it’s a huge industry with practices that are dating back several centuries. So obviously, these industries did manage to find their own solutions a long time ago. They didn’t wait for the age of the internet to figure out how to do it at a global scale. You even have discussions centuries ago about whether some countries would be allowed to import wheat or other things. So, the idea of globalizing the food market is literally going back centuries in the past. Because this industry did figure out solutions for those problems centuries ago, before the time of the internet and modern information technologies, they did come up with solutions that work. We are not starving anymore, so that’s a fair point. And also, there are plenty of specificities that kind of only apply to food. These specificities don’t only apply to food per se, but there are plenty of things that make food relatively unique compared to most other supply chains.

Kieran Chandler: From a data perspective, would you say that because some of these companies are so ancient, they may be not quite so advanced?

Joannes Vermorel: Yes, and sometimes for good reasons because it’s more difficult. Modern supply chains are characterized by barcodes; everything comes with a barcode, and everything is packaged and standardized. The problem when you’re dealing with produce or meats is that it’s a sort of industry where you end up selling the products by the kilogram or the pound, and things do not get packaged with barcodes. The end consumer version sometimes has a nice box for exactly half a kilogram of meat or half a kilogram of produce, but that’s really the end of the chain. Everything that comes before, you have to deal with these relatively raw materials where you’re measuring quantities, which is something that has nearly disappeared from pretty much all the other supply chains. Even chemicals now tend to have barrels or jerry cans, so you have your barcode and units of products, and you’re not counting things in kilograms.

Kieran Chandler: Let’s get onto some of those challenges then. What are the key challenges that the industry faces? I imagine it’s a really international supply chain these days, with people demanding things like quinoa and avocado coming from all over the world.

Joannes Vermorel: I think one of the challenges is the diversity of supply chains that you have to deal with because when you say food, you have tons of different types under this overall term. You have fresh foods, like fresh produce and meats, that are sold per kilogram and are very frequently produced locally or not too far away and with very traditional ways of selling them. Even in modern markets, you can typically pick anything from the shelves, except for meat where you still have one person and you ask for a piece of meat, and this person is going to take a piece of beef and cut it for you.

Kieran Chandler: The fresh food, with specialties like meat, fish, and produce, tend to be literally specialties on their own. Then you have everything that is frozen, so you have a whole supply chain dedicated to frozen food. One of the problems is that when you have ever been into a warehouse for frozen food, it’s freezing cold in there. So, it’s quite tough for people to operate. I mean, it’s literally painful when you have to spend the entire day at minus 18 degrees Celsius. It’s very cold, and if you have to spend the entire day, it’s not comfortable in the slightest. So, people who operate supply chains for frozen food want to, and they have been very aggressively moving toward, end-to-end robotization so that they have as few people as possible operating with frozen foods. Let’s say we sort of tackle the idea of expiration dates then, particularly with fresh food. How are these big hypermarkets managing that? Because it’s a complete logistical nightmare if you think about the different products they’ve got to manage and the different expiry dates. It must be quite challenging.

Joannes Vermorel: Yes, I mean, there are many challenges. Some are accidental, and some are not so accidental. On the accidental challenges, you have that many ERPs have a very crude vision of the stock, meaning that they are thinking in terms of SKUs. In a previous episode, we were discussing the issues with SKU (stock keeping units). What’s wrong with them is that the notion of SKU does not embed the expiration dates. So actually, when you say you have 20 units from a hypermarket perspective when it comes to food, it’s not exactly that. You may have, for example, four units that expire in one week and the rest that expire in a month. You really want to have this fine-grained vision about your stock. That’s the accidental complexity. You have a business angle that should be managed by your system but is not, so that creates a lot of supply chain complications that are purely accidental.

And then you have the not-so-accidental complications. One of these complications is that customers are kind of messy. They might pick something on the shelf, look at it, walk for five meters, and finally spot something that they like better and put it back, but not exactly where it belongs. This can lead to stockouts on the shelves, but if you look at the electronic record, it’s not necessarily that the product is gone; it might be that the product is misplaced. This can happen to people in other supply chains as well, for example, for retailers selling clothing. But hypermarkets are especially noisy environments, with a lot of noise in terms of data.

Kieran Chandler: And how about the complications for fresh food from the supply side? Because you’re very reliant on these suppliers who are reliant themselves on things like harvest and weather. So how easy is that to manage?

Joannes Vermorel: That’s interesting because what makes it kind of easier is that if you’re dealing with hypermarkets and food, you’re typically dealing with relatively fast-moving products. For example, for fresh food, it doesn’t make any sense to have something super fresh if you can’t even sell one unit a day, or you have to be like a mini-market with an 80% gross margin to support that. But other than that, you need rotations; otherwise, you’re just going to discard your unsold expired inventory all the time. So by definition, you have decent rotation rates, which from a statistical perspective really helps to get good forecasts. It’s much easier to forecast the demand for a product tha

Kieran Chandler: So, in this respect, food is relatively on the easy end of the spectrum due to volume, but indeed, where things get more complicated in terms of supply chain forecasting and optimization is when your suppliers are a lot less reliable. Can you elaborate on this?

Joannes Vermorel: Of course. Your suppliers can be unreliable in multiple ways. The classical way is just lead time variability. For instance, you have suppliers you place an order with, and they can take two days to deliver or five days. You need a lead time forecast that is probabilistic to account for this uncertainty. This also exists for fresh food.

However, when the supplier does deliver, you have quality control issues. Most large retail networks nowadays have quality control checks at the distribution center level. They identify products that are not of high enough quality to be put on the shelves. These can be products like produce that don’t look good; they wouldn’t be a threat to customers, but if they don’t look good, they’re not going to be sold. So you need to find alternative uses for them, like using them in cakes.

Kieran Chandler: In that scenario, if you’ve got a problematic supplier, how easy is it to multisource and use a different supplier? Or is it a case that there are many different retailers competing for stock, which adds extra complications?

Joannes Vermorel: That’s an excellent question. You may have a supplier that delivers 100 units, but you have to get rid of 20 units because they don’t pass your quality controls. That creates some uncertainty. The reality is that, very frequently, large food distributors can end up in a situation where they have to go to auctions to buy merchandise at varying prices. They might not even buy from the same supplier every day. In the food industry, it’s common to have 20 potential suppliers, and through an auction mechanism, you just end up picking a different one every single day. It’s like multisourcing but with a marketplace mechanism in the middle.

Kieran Chandler: Let’s move on to some of the other challenges a classic hypermarket might face. One of the real difficult ones is promotions. How are these agreed between the supplier and the distributor?

Joannes Vermorel: It’s a very intriguing question. When I started looking at the pain points mentioned by most of the general merchandise retailers I met, promotions were considered super hard to forecast. We at Lokad started doing promotion forecasts in 2008-2009, but it took me years to realize that it was wrong to even look at the problem that way.

You see, the way we were trying to look at the problem was: there is a promotion, we want to forecast the demand. People would tell me about various promotional mechanisms, like “buy one, get one free,” and for any given mechanism, you’d want to know how it would affect sales.

Kieran Chandler: Products that are going to benefit from this promotion, what is the demand uplift?

Joannes Vermorel: Where it’s completely wrong is that if you want to first and foremost, a promotion is a negotiation between typically the retailer and the brand backing the product. It’s a negotiation taking place where the supplier offers a very good price, but in exchange, expects the retailer to offer a very good price to the consumers. Typically, the supplier is going to promote the product on their own, maybe through TV, journals, and whatnot.

At the core, if you want to think about optimizing promotions, you have to take the perspective that it’s a negotiation between the retailer and its suppliers. Thus, the question is, are you even thinking about the problem as optimizing the outcome of a negotiation? If you just think of it, as we were doing in the first years, as a pure forecasting problem, it’s completely wrong. It’s just not the right perspective on the problem, and this doesn’t matter if your forecasts are accurate or not. You’re kind of missing the crux of the problem.

Kieran Chandler: From a data perspective, I mean, a lot of the time now, these hypermarkets can see exactly what we’re purchasing because we’re using loyalty cards, and they can really track it down to the individual consumer. So how much is this data actually used, and what’s it used for?

Joannes Vermorel: Indeed, they have a lot of data, and I would say it’s incredibly rich. With purchase history, if you can access it, you’re maybe going to buy hundreds of products per year from this hypermarket, so that gives you an enormous amount of information for profiling. It’s incredibly rich. There are very few companies who could rival this. Even if you think in terms of your bank, your bank knows a lot about you, but they don’t necessarily have thousands of transactions per year that tell much about you.

On the contrary, in a hypermarket, you can really see which brand and type of products people buy. For example, it’s pretty hard to pinpoint your religion from your bank account, but if you go to a hypermarket and you’re buying halal food, it doesn’t take a data scientist genius to figure out your potential religion. So you have a lot of information. Then, what do you want to do with that? There is plenty of ways.

The media frequently portrays a degree of hysteria going on, on those evil corporations doing very evil things with the data. My own casual observation is, despite all the hype and all the things that you can read in the professional press, on average, they are doing about nothing with this data. It’s very disappointing. They have set up massive loyalty programs, invested a lot for decades, and literally, most of the time, they are doing nothing or relatively dumb things.

For example, you identify young women who are going to buy something for pre-birth, some goods for very young infants, and then they send promotional stuff related to infants, which is like the most basic approach to using this data. That’s just direct marketing; that’s not even supply chain optimization. As far as supply chain is concerned, I

Kieran Chandler: Would you say, to your knowledge, the vast majority of those companies are doing nothing with this data? When I say this data, I mean the baskets.

Joannes Vermorel: Yes, the vast majority of those companies are not doing anything with that data.

Kieran Chandler: Let’s talk about one of those companies that’s supposedly doing lots with data – Amazon. They’re now entering the fresh food market with Amazon Fresh. How is that going to impact the industry, and how are we seeing the growth of e-commerce in the industry?

Joannes Vermorel: I believe that Amazon is facing another uphill battle because they will be competing with the physical infrastructure of the entrenched players. Food is relatively difficult; it can be literally messy. For example, when you’re selling books, if a book falls on the floor in your warehouse, it doesn’t make a mess. Worst case, the book is damaged, and you can’t sell it anymore, so you just discard it. But when you have a bottle of wine, if you drop it, it creates a mess that you need to clean up. So you’re facing very mundane problems where high-speed automation is not super easy. One thing breaks, it gets messy. You have lots of liquids, things that can be explosive, like oils. If you’re piling bottles, it’s a major hazard in case of fire. The entrenched players have had decades to become relatively good at solving these mundane problems, which do not require super high-tech solutions. Amazon will need to figure out ways to solve those problems and have their usual mix of high-tech and low price. They will need to find solutions for every single one of those very mundane problems.

What we’ve seen with Amazon is that they tend to go mostly down the road of acquisitions or partnerships. In France, they’ve closed a deal with Monoprix, a well-known urban food retailer. So they didn’t decide to roll everything on their own; they’re taking a partnership with an established player. The question is, how will they manage to have this tech edge to the supply chain instead of just being a marginally better website?

Kieran Chandler: So, would you say that the main reason the fresh food industry is a little bit behind and more traditional in its approaches is because of the challenges it needs to overcome?

Joannes Vermorel: Yes, I think there is a huge diversity of challenges, and the traditional software tools were just very bad at it. In fresh food, if you’re bad at supply chain management, your products go to waste very fast. The penalty for being bad at supply chain is dramatic. For most other industries, if you’re just one week off, it’s not that good, but it’s not the end of the world. In fresh food, if you’re one week off, you’ve literally lost your merchandise. The degree of tolerance that retailers, suppliers, and wholesalers have for poorly performing tools is very low. What I’ve seen is that software has routinely failed, decade after decade, in delivering any truly satisfying solutions. They did upgrade from old paper-based solutions to Excel-based solutions, where Excel is just a glorified upgraded version of what was done on paper before. But fundamentally, they have still not figured out ways to do it much smarter. I hope that with the sort of technologies we have developed at Lokad, we have something better to offer now.

Kieran Chandler: Brilliant, we’ll wrap it up there. Thanks for your time, Joannes. So that’s it for this week. Thanks very much for tuning in, and we’ll see you again next time. Bye for now.