00:00:00 Introduction
00:01:49 S&OP: Quantitative alignment and decision making
00:04:08 S&OP for company-wide synchronization, not value alignment
00:06:56 S&OP: Team collaboration, unified forecast, and deliverables
00:09:36 Commitment agreements and bureaucratic challenges in S&OP
00:13:12 S&OP: Slow process, qualitative insights, and information flow
00:18:59 Evolution of companies, product complexity, and supply chains
00:23:19 S&OP: Costly, slow, outdated, and software-mediated
00:26:37 S&OP: Origins, formalization, and impact of technology
00:30:33 S&OP: Need for alignment, critique of meetings
00:33:30 S&OP: Comparison to cancer, forecast generation, and auditing
00:36:35 S&OP: Informational problem, data flow, and software necessity
00:40:52 Focus on unchanging fundamentals, critique of unified forecast
00:45:11 Consequences of stock out, repetition, and e-commerce questions
00:48:42 Difficulty of decision-making, critique of S&OP formalization
00:52:47 Value of qualitative analysis, critique of S&OP deliverables
00:56:43 Termination of S&OP division, removing unnecessary divisions
01:00:00 Upskilling, importance of accessible information, and data access
01:03:40 Converting S&OP team into data lake team
01:05:33 Conclusion of the interview


Lokad CEO Joannes Vermorel critiques the Sales and Operation Planning (S&OP) process as outdated and inefficient for modern businesses. He argues that S&OP, designed for simpler times, struggles to keep pace with today’s complex, fast-moving business environment. Vermorel criticizes the process’ slow refresh rate, its reliance on meetings and spreadsheets, and its focus on unified forecasts. He suggests that large companies should eliminate their S&OP divisions, likening them to obsolete fax machines. Instead, he proposes a shift towards improving access to information, transforming S&OP teams into data lake teams.

Extended Summary

Conor Doherty, Head of Communication at Lokad, engaged in a thought-provoking conversation with Joannes Vermorel, CEO and founder of Lokad, about the relevance and effectiveness of Sales and Operation Planning (S&OP) in the modern business landscape. Vermorel offered a critical analysis of S&OP, arguing that it is an outdated and inefficient process that fails to meet the needs of today’s complex and fast-paced businesses.

Vermorel began by explaining that S&OP is a corporate process designed to create alignment within businesses that have extended supply chains. It involves making decisions about production, storage, and transportation capacities ahead of time to ensure consistency within the company. However, Vermorel criticized S&OP as a simplistic solution to the problem of company alignment, arguing that the process is slow and struggles to refresh even on a monthly basis. He questioned why the alignment process can’t be real-time, suggesting that the need for synchronization, alignment, and cooperation would be better served by an hourly process.

Vermorel also pointed out that the S&OP paradigm was developed 50 years ago when companies had fewer products and less complexity. He used Procter & Gamble as an example, noting that the company had around 100 products 40 years ago, compared to over 20,000 today. He argued that the S&OP paradigm is outdated and ill-suited to the complexity of modern companies, criticizing the reliance on meetings and spreadsheets for decision-making, which he described as slow and inefficient.

Vermorel further criticized the S&OP process as a bureaucratic tug of war, where different departments have conflicting interests. He suggested that these numbers are often ignored in practice, with each division operating independently. He also dismissed the idea that S&OP meetings are about strategy, stating that they are more about agreeing on numbers.

In response to Doherty’s question about the software requirements needed to implement an effective version of the collaborative approach, Vermorel insisted that the solution must be fundamentally a software solution, as information cannot flow through people. He gave an example of how companies in the past might have had one inventory manager per SKU, which is not feasible today.

Vermorel also challenged the idea of a unified forecast as a solution in S&OP, calling it a waste of time. He argued that the numbers change constantly and it’s pointless to try to get everyone to agree on them. He criticized the focus on unified forecasts in S&OP meetings, arguing that it leads to discussions about the wrong things.

In conclusion, Vermorel suggested that large companies should terminate the S&OP division, arguing that it’s pointless bureaucracy. He encouraged CEOs not to be afraid of removing the S&OP division, comparing it to a fax machine, a technological dead end. He suggested that people who have specialized in S&OP should focus on improving access to information, converting their team into a data lake team.

This interview offers a critical perspective on the traditional S&OP process, challenging its relevance and effectiveness in the modern business landscape. Vermorel’s insights underscore the need for businesses to adapt and evolve their processes to meet the demands of today’s complex and fast-paced business environment.

Full Transcript

Conor Doherty: Sales and Operation Planning, or S&OP, has been the very bedrock of business planning for at least the last two decades. But despite its popularity, it is not without criticism. Here to explore them with me today is Lokad founder, Joannes Vermorel. Joannes, welcome. So, S&OP has been around for a very long time, but when we talk about it in 2023, what does it entail?

Joannes Vermorel: S&OP is essentially a corporate process to create company-wide alignments in companies that have extended supply chains. The basic motivation, the underlying reason why people want to engage in this sort of process, is that if you have a large company with an equivalently large supply chain that deals in physical goods, then to make it simple, you need to sell what you produce and you need to produce what you’re about to sell.

These sort of alignments need to happen and due to the fact that you have many stakeholders, potentially many divisions, it is non-trivial. Thus, there is a popular recipe, nicknamed S&OP, Sales and Operation Planning to do just that. It’s fundamentally a set of corporate practices to address this core problem of alignment, synchronization, and proper cooperation within the company to achieve basic business goals.

Conor Doherty: So when you say ’the alignment process’, specifically, do you mean insights or sales figures? What exactly do you mean by alignment between the different functions?

Joannes Vermorel: The alignment, from the S&OP perspective, is mostly about a quantitative alignment. There are plenty of sorts of alignments. People can try to align themselves on a certain vision for the business, on a certain culture, on a certain way of doing business. This is not what S&OP is about.

S&OP is really about the quantitative alignment. When I say quantitative, I mean ‘how much’. So, if you think that you’re selling widgets, the question is: should the factory produce a thousand widgets, a million widgets, a million a week, a million a month, a million a quarter? Should you expand the factory so that you can produce more? Should you shrink the factory and close maybe a production line because you need less? Do you need a bigger warehouse, a smaller warehouse? Do you need more transport capability, less transport capability, and where?

So there are plenty of questions where the various moving parts of your company need to size up the sort of efforts and commitments. And due to the fact that supply chains have delays, a lot of decisions have to be made ahead of time and thus the company must remain consistent with itself.

If you decided 6 months ago that you were about to produce a million widgets, well now, 6 months later, the widgets have been produced and you have to make sure that you can sell them. You have to store them somehow you have to transport them somehow etc. That’s what is underneath the term S&OP. The sort of problems that are tentatively tried to be addressed through those processes are those company-wide alignment, synchronization, and coordination. And we are really talking about the quantitative stuff.

Things like aligning the company-wide or divisions on the same values, for example, a certain way of paying attention to your job, some companies have very strong cultures around what it means to work for this company. This is not what S&OP is about. S&OP is really about the mundane quantitative alignment, not having everybody on board with the same values, for example.

Conor Doherty: So when we’re talking again, we’re squarely within the realm of quantitative figures. We’re talking about the hard numbers: what we’re going to sell, what we’re going to build. How exactly do you align very disparate departments on this? I mean, that’s surely a qualitative procedure. Is that meetings, is that emails, is there software for that? How is that handled?

Joannes Vermorel: The classical view on S&OP, the problem statement is that there is possibly half a dozen divisions, possibly more, that have to take decisions ahead of time compared to a future that is not perfectly known. S&OP is this process that is supposed to provide an answer to those divisions so that they can take these decisions ahead of time. Now, the way S&OP approaches this problem is very specific.

Essentially, S&OP proposes that teams are going to collaborate to establish a unified, company-wide forecast. There are different ways to go about it, but it’s really what it’s about. It’s a unified forecast for the company, and not any kind of forecast, an incredibly specific forecast. It’s going to be a time series, a point time series forecast. So literally, the company is going to decide on a certain granularity.

It can be the product, it can be the SKU. There will be a granularity in terms of what we measure, but there will also be a time granularity which can be per day, per week, per month, per quarter, possibly. The company picks a granularity according to S&OP, and then forecasts are made and potentially iterated upon through the S&OP. The idea of S&OP is that since there are many shoulders, all those people have to meet and discuss and revise.

The output of S&OP is this unified forecast for the company. This is literally the deliverable of the S&OP process. It is a company-wide forecast which is then considered as the truth, the baseline, and everybody has to do what they have to do to meet these targets, which is not just a statistical forecast in the sense of being a number that is accurate about the future, it is also a target for everybody as well.

Conor Doherty: On that point, and I don’t want to put words in your mouth, but you just said that people will collaborate under the S&OP umbrella on the demand forecasting. Now that obviously is a quantitative process. Are you suggesting that people in this articulation of S&OP will weigh in? Like if I’m in sales, we’re going to sell more than that, we’re going to sell less than that. How exactly is there collaboration in a very quantitative process like demand forecasting?

Joannes Vermorel: From an S&OP perspective, yes, the numbers are technically a forecast, but in practice for the various teams involved, it is a commitment. So it’s a very different sort of numbers. Yes, it’s a time series, but from the S&OP perspective, it is a commitment. That means that people commit their division to be able to fulfill this target. These commitments mean something different for every division. So if we have a product and we say next month 1,000 units, if you’re from the sales division, it means that you commit yourself to make this happen. There will be orders from clients that will be roughly equal to 1,000 units.

If you are from the warehousing division, that means that we have the capacity to store those 1,000 units. So that means that we do what it takes so that when those 1,000 units will flow, we have the storage capacity to make it happen. And then production, they would say we have acquired the raw materials so that we can produce those 1,000 units. So again, those numbers are as much forecast as they are commitments. And as part of those discussions, collaboration, it’s not the statistical aspect that dominates, it is more like the agreements on the commitments that dominates.

Conor Doherty: And these agreements, or the unified vision that you described, this is reached once per quarter, once per year? What’s a typical timeline or cycle for this kind of process?

Joannes Vermorel: When we have to consider the timeline, we have this problem statement, the need for alignment, which is when you look at the simplest possible solution, that would be craft numbers that are commitments or forecasts, but they are more on the commitment side. And people need to meet, and then you want to keep this information fresh, so that makes sense. So most companies would want ideally to be able to revisit that, let’s say, weekly. The reality is that S&OP is invariably something that devolves into a bureaucratic effort and that’s very slow.

Thus the vast majority of companies have achieved kind of quarterly refresh of their S&OP plan. They nearly all dream of going to monthly updates, but the reality is that it’s so slow that it’s an immense struggle to get a monthly refresh. And the weekly refresh or even the daily refresh is just completely outside the realm of what sounds possible.

Conor Doherty: So people, in theory, aspire to meet on a monthly basis to revise this unified vision. Are we hitting these numbers? Continually tweak the formula.

Joannes Vermorel: We have to revisit again, why do we have any delay? People often bring up S&OP, but the problem is that S&OP, by implication, is a simplistic solution. We have a problem and we jump on the most obvious solution, which is to have a unified forecast slash target slash commitments that bring the whole company together. It’s the simplest idea you can have, but it’s not necessarily the best or the most efficient. It’s just the most expedient and naive way to go about it.

Now, you end up with a problem. If you decide to execute it that way, you end up with something that needs to be refreshed. In practice, it is very difficult to have this process iterate, to have a full refresh of this process on a monthly basis. Most companies only succeed at doing that on a quarterly basis.

But if we step back and look at the problem statement, which was to have alignment within the company, why should this thing be anything but real-time? There is no reason. These things could be happening every day, every hour. Why do we have a delay on that?

There are plenty of things that are incredibly difficult and can be done in milliseconds. For example, computing the value of the cosine function, it’s something very complicated. Until a century ago, you had to be a university professor to do this sort of calculation. Nowadays, a pocket calculator or a smartphone can do millions or billions of those calculations per second. So, the question is, why should it take time at all?

When we think about S&OP, we went from the simplest solution, which may not be fine in reality, but let’s suppose for a second that it is. Then you end up with a process that is very slow, where it’s a struggle to even reach a refresh rate of one refresh per month.

But if we look at the problem statement, the initial problem statement is the need for synchronization, alignment, and cooperation. These sorts of things would be better done by the hour. There is no reason not to do that, except if your process is just incapable of doing it.

Conor Doherty: Just to push back a little bit, an S&OP advocate might say that while doing it hourly would be great, there are certain qualitative insights or market insights from specific departments that are difficult to translate into a numerical form. Thus, it’s easier to do that in a meeting. That’s why they would have meetings to relay information about finance or sales, here’s what we’re going to do. The warehouse is going to be shut down, we’re decreasing it. There are certain things that just have to be communicated orally rather than easily expressed numerically.

Joannes Vermorel: That’s where there is a profound misunderstanding about what you can communicate at all through humans. Humans are low bandwidth creatures. The amount of information that can go in and out of a human body is limited. If you look at information theory, expressed in Shannon bits, how many bits of information can you have in and out of a human person? It doesn’t matter if this person is super smart, super educated, has tons of insights. The reality is that per second, we are talking about a tiny few bits per second.

So, what does this have to do with the problem? Well, it turns out that large companies are super complex. When we are talking about companies that are selling tens of thousands of products or hundreds of thousands of products that have millions of SKUs, the amount of information that needs to flow through the company is very large. We are talking about megabytes of information, and potentially, in large companies, gigabytes of information. This is irreducible information, the raw information that needs to somehow transit through the company.

What people don’t naturally realize is that you can’t convey this information verbally. Sales cannot communicate all the information there is to be known about the sales to the production division. In a large scale company, we are talking about at least megabytes of information. Even if you do a three-hour meeting, we are talking about kilobytes of information that will flow. So, we have three orders of magnitude of discrepancy between the amount of information that can flow in a meeting and the amount that needs to flow.

What is happening in reality? The information is flowing, but through Excel spreadsheets. The Excel spreadsheets do contain the megabytes of information that need to flow. This really begs the question of what is it that you’re really communicating.

In the end, the problem is that people discuss these numbers, but these numbers are incredibly thin-grained and there is tons of information that, by design, cannot be discussed. The S&OP was a paradigm that emerged at least 50 years ago in a world that was simpler, in a world where companies had at least 10 times fewer products and very frequently 100 times fewer products.

For example, 40 years ago, my parents were working at Procter & Gamble. Worldwide, at the time, they had like 100 products. Nowadays, it is like 20,000 references, at least. It was simpler times, with fewer references and much less complexity.

If we look at the way a very large company like Procter & Gamble was set up four decades ago, it was one country, one plant, and then you distribute to, let’s say, 10 warehouses that are your national retail chain, that would be the Carrefour and all the large retail chains, and you ship one or two full trucks a day to each of all those warehouses, and this is it. And every country is kind of the same, and you have a supply chain that is like super simple. It’s not the supply chains that we have nowadays. Nowadays, supply chains have grown immensely in complexity. We are talking now routinely, even for companies that are nowhere as large as Procter & Gamble, of tens of thousands of products, hundreds of suppliers, various modes of transportation, and multiple channels.

There might still be a few companies that say they do 10 billion euro worth of turnover with only 20 products and they have only like 10 clients and it’s super simple and they only source locally. Yes, there might be some businesses that are like that. I don’t think that it’s a majority. I think if there are still any businesses that operate like that, I’m not even sure that we can still find businesses that would say we do 10 billion, we are only local, and we have 10 clients that are also local and we source everything locally.

I think companies, once they reach this sort of size, they are super complex nowadays. Thus, the problem is that this S&OP paradigm was something that made sense for this low complexity world where, through a few kilobytes of information, you had everything that was needed. But nowadays, we’re talking about megabytes and information cannot flow. So, people do the next best thing, which is to go through spreadsheets.

The information is mediated through software. Nowadays, there is no company in Europe or North America that is above, let’s say, a million-dollar turnover that doesn’t have an ERP or some kind of digital backbone for their supply chain and their operations. So, all the information is mediated through software. It can be mediated with ERPs or through spreadsheets, but it goes through the spreadsheet.

So, what is it that is truly going to be discussed in those meetings? In my experience, and I’ve seen probably over 200 companies in these S&OP processes, no matter how good the initial intentions were, they devolved into super bureaucratic affairs where it’s basically a tug of war. Sales wants to have the targets as low as possible so that they can exceed their commitment and get their bonuses. Production wants to have the targets super high so that they can invest in their capacity and it will be easy for them to produce whatever is asked of them.

This was developed in a tug of war where people defend their turf. It becomes very political and, in order to prevent cheating or abuse, companies put even more processes in place. This makes things that were too bureaucratic in the first place even more bureaucratic. In the end, these S&OP processes are super slow. Most companies refresh only once per quarter, so you end up with numbers that are always available too late. It’s a big charade.

Everyone pretends that it produces useful numbers, while every single division operates independently of those numbers in practice. They need those future numbers to be refreshed on a daily basis for plenty of reasons. So, these S&OP numbers are produced at great cost, and in the end, every single division just minds their own business and does it their own way anyway. In practice, the information is flowing through software, so it’s mediated, it’s not flowing through those meetings anyway.

Conor Doherty: I want to summarize a bit there because you’ve dropped a lot of little pins in that. Correct me if I’m wrong, but based on your description, S&OP goes back to, let’s say, the 80s. So, about 40ish years at a time where companies were operating on a scale of complexity of let’s say 100 products. Fast forward 40 years, there’s a couple of orders of magnitude more complexity in terms of products being offered, yet companies are now less agile despite the 40 years of technological advancement between the birth and present day. Is that a fair summary of that?

Joannes Vermorel: Yes, I would say S&OP dates back probably a century ago. I would date S&OP with the emergence of the modern 20th-century companies, a little bit more than a century let’s say, with the emergence of the modern large-scale company. Companies like General Electric and Ford were fairly unique at the time because they were the first time that you had very large private undertakings. Before that, if you were to seek anything that would be 100,000 people, the only places where you would find 100,000 people that work together would be the armies.

The beginning of the 20th century is the birth of the large modern company where you have tens of thousands of people who jointly work together for something that is not the Army. You can even find tidbits of that in the terms when you say chief executive officer, all those terms were borrowed from the Army because that was the only place where you had this sort of large hierarchy, the Army and maybe the church.

Conor Doherty: Specifically in the context of business though.

Joannes Vermorel: So, I would say as soon as you had that, I’m pretty sure that even if you look at General Electric in 1920, they had internally something that was very much like S&OP. People were producing light bulbs, some other people were storing light bulbs, some other people were selling light bulbs. I’m pretty sure all those people were meeting once in a while to decide. But the question was, how many different references did they have? And I’m pretty sure if you look at the time, we were talking of hundreds, maybe a thousand, significantly fewer than today.

I don’t know how many product references a company like General Electric has nowadays, but I would bet that they have over 100,000 and they operate in probably over 100 countries. So, it’s very different from the company that was General Electric 100 years ago. So, back to S&OP, the way I see S&OP, it was just something that was formalized, and I would say, four decades ago it was formalized. And why was it formalized? The question of why was it formalized is because it was productified so that it could be sold by consultants.

So, those sorts of things were done naturally. Again, that is pretty much S&OP. The way you see it, it’s the emerging process that you get in a large company when you just let people figure out whatever solution they can find when they are in a rush and they don’t really spend much time on any kind of elaborate thinking about the way you should even approach these sorts of things. And the reality is that unless you have modern computers, you don’t really have an alternative.

So, probably, this sort of pre-S&OP, something that was not packaged as something that could be sold by consultants, but let’s say this pre-S&OP from 1920 to 1980, well, for those 60 years, that was pretty much the only option available. Computers at the time were not really capable of providing any kind of alternative answer, so you had that, and that was just reasonable. And then the game changed progressively in the 80s because computers started to become super powerful.

And it becomes dramatically different starting from the year 2000s because with internet connectivity, it really dramatically changed the way you could think not only about processing information but also the flow of information just because you had this commoditized internet. Yes, the internet is dating from the 70s, but it took some time until it was really commoditized in the sense of being super cheap to have corporate data that just flow through this network. And I would date the age where it becomes very, very cheap to flow business data through the internet at the year 2000.

Conor Doherty: If the collaborative approach, like the in-person collaborative bureaucratic approach, you’re not an advocate of, okay, well then what exactly would the software requirements be to implement an effective version of this?

Joannes Vermorel: The thing is, again, you have a problem statement. People tend to confuse problem and solution. S&OP is a solution, this is not the problem. What is the problem? The problem is alignment, synchronization, coordination. So, you have the problem statement. If I go to the naked problem statement, the problem statement is half a dozen or more divisions have every single day thousands of decisions to make. That’s the raw problem statement.

So, the sales will be, which prospect do I pursue? Which client do I recontact to make an offer? The pricing division would be, for every single product that we are selling, should I raise or lower the price point every single day? If you’re dealing with replenishment, every single day, what do I need to order, etc. So, those are the problem statements, all of those decisions. And we are talking, again, if we’re looking at a large company, we’re talking of millions of decisions per day.

And this is the flow of the supply chain. Those decisions are very repetitive and they need to be aligned because there is a whole sequence. You buy the raw material, you produce, you transport, you store, and then you transport again to do the final delivery. So, there is a whole sequence of that, but it’s very repetitive and that’s why we have a flow and that’s why we have this supply chain.

Okay, so the naked statement is, how do we achieve this collaboration or this alignment for all this on those decisions? That’s the naked problem statement. The solution which says, let’s have meetings for that, it’s one possible solution, but I would say it’s a very, very crappy solution. It’s going to fail on so many fronts. I mean, again, the meetings, it’s low bandwidth, so there is very little information that can possibly flow through that.

So, if you think that you can create alignment through those meetings in terms of those quantitative decisions, no, it doesn’t work. And people, consultants, would say, oh, but we have to align on the strategy. And I would say, if you think that S&OP is about to discuss strategy, you’ve never been to S&OP meetings. This is S&OP is really a tug of war on numbers. If there are, you know, if people have to agree on a shared strategy, this is not the S&OP division that is going to do that.

And by the way, I’ve rarely seen processes that are so little concerned with high-level strategy than S&OP. So, S&OP is absolutely not about this strategical things. It’s not in an S&OP meeting that you decide whether you should move your brand toward premium, charge more, and have more quality. This is not in an S&OP meeting that you discuss these sorts of things. Because you see, that would be a possibility where you say, if we go premium, will you be able to sell it at a higher price? That would be a question for the sales. And then, if we go premium, you could ask production, will you be able to have higher quality, etc.

But this is not the sort of question that gets realistically asked. So, you see, these sorts of strategic questions are absolutely not the sorts of questions that are dealt with in S&OP meetings. S&OP meetings are quantitative, so people want to agree on numbers. So, this is not, this is the kind of the opposite of the strategy questions. And the deliverable is this kind of one unified forecast which has, again, nothing to do with strategy.

Now, the difficulty is that once you have, it’s very difficult to think of a problem without thinking solution. So, people are kind of stuck, you say, okay, but if we remove that, what do we put in place? And here, I would have several answers, but the first answer would be something that is not especially nice. It would say, once you remove a cancer, what is a surgeon putting in place in the patient?

You know, I would say most companies could remove their S&OP division and it would just work exactly the same. Because, again, the reality is that information flows, but not through this S&OP process, it flows through the software layers that exist in the company. And that’s it. Because that’s not, and just to give you the sort of ideas, is that how does the sales team generate the forecast?

Well, the sales team takes the sales of last year, they remove 20%, and they say this is a new target. And then they are going to exceed expectations because, well, the company is going to be more or less doing what they did last year. But since they have lowered the bar by sandbagging by 20%, they’re going to exceed their target. And then production, they do the same thing. They look at the sales history, they see what was needed last year, they do their own calculation on forecast, and that’s how they do it.

And yes, there is a meeting with an S&OP process where everybody does a tug of war on the numbers, but at the end of the day, all those numbers are discarded. And the way I know it is that, once we at Lokad, we have frequently audited, I mean, frequently, we did that at least a dozen times, we did audit the numbers that were produced by S&OP processes. There were absolutely stunning errors in the forecast.

I mean, not in the sort of statistical error, errors where numbers were absolutely wrong because it was like data integration issues, and those errors, so that the numbers were completely off by sometimes several orders of magnitude. And it had never bothered anybody because, actually, the reality was nobody was even using those numbers. So, you can have numbers that are off by a factor of 1,000, and nobody even notices because nobody is even trying to use those numbers.

And that’s where I say it’s a charade because if you can produce numbers that are off by a factor of 1,000 and it doesn’t create any adverse consequences, it has no adverse consequences for the company, it means that nobody is actually using those numbers. So, it doesn’t, accuracy doesn’t matter, it’s just irrelevant.

Conor Doherty: There was a comment made about sandbagging. We can’t speak for a process and say that everyone who’s going to participate, or even the majority of people who will participate, will have bad intentions. But it seems fair to say that in a bureaucratic process like this, it opens the door to inefficiencies. You can drop into that bucket the idea of sandbagging, bad intentions, whatever. But as a system, the more people you involve in it, the more collaborative you make it, you increase the surface area for that kind of inefficiencies.

Joannes Vermorel: We have to go back to the physical reality, and when I say physical, I mean informational reality. We have a problem that is about processing information. This is not a physical problem like moving a box from one location to another. This is purely about information in, information out. This is an informational problem.

The question is, how much bandwidth do we need to solve this problem? It’s a problem that needs to flow megabytes of information in a midsize company and probably gigabytes of information if you’re looking at very large companies. That’s the problem statement. You can’t bypass this. It doesn’t matter what you wish the situation to be, this is what needs to happen.

If you think that having a process that is driven by people to flow this information is going to solve it, it’s not. If you have a process that looks to be driven by people that seems to be solving the problem, it’s not through people that information flows. It’s not possible. It flows through the software, possibly through spreadsheets. But the information is not flowing through people. That’s my statement. It’s just not possible.

Once you acknowledge that, then you need to acknowledge that whatever solution you will find will be fundamentally a software solution because that’s how the information is flowing. It cannot flow through people. So if you say, “Oh, people are smart, people are insightful,” I would say yes, but they don’t have the bandwidth. It’s just not possible unless you literally take one planner per SKU, and then it becomes possible.

If we go back to General Electric in 1920, I would not be surprised that at the time they had one inventory manager per SKU. That would be very much the sort of thing that companies would do at the beginning of the 20th century. So yes, if you have one employee per SKU, yes, you can do that. But you’ve solved the bandwidth problem by just having a number of people that, according to modern standards, would be insane. No company nowadays could afford to have one person per SKU to manage inventory, and yet this is what was done a century ago.

Conor Doherty: It occurs to me that even potentially a software alternative to this, something that consciously limited the amount of human touch points, would still be predicated upon at least one person per department inputting this information. And isn’t that still vulnerable to sandbagging? Couldn’t I just mess around with the numbers?

Joannes Vermorel: Let’s clarify. We are reasoning by implication. I said that the easiest and simplest thing was to have an agreement on a unified future that becomes a commitment for everybody. That’s the intuitive and easy solution, but it’s not a good solution. It’s a terrible solution when you think of it. Why is that? Because those humans who have low bandwidth, you’re going to ask them to communicate information about stuff that changes all the time. And this is the opposite of what needs to be done. I know it’s a semi-philosophical problem but it’s a very important one.

If you have humans in the loop, you need to focus on what does not change. If there are insights to be acquired, this is about the fundamentals, what does not change. At school, you do not get taught about the weather forecast. The teacher does not walk into the classroom and spend the day teaching children about the weather forecast. The teacher could enter the room and tell the children about the temperatures that are going to happen in every single city in France for the next month and we could repeat that every single day. The teacher would come to the classroom and say we have these 20 cities, here are the temperatures for the next 30 days. Would the children learn anything? No, that would be an attempt to focus on what changes. So the teacher does the exact opposite of that. If he or she has to teach and communicate something, he or she will focus on what does not change, like arithmetic, grammar, poetry, whatever.

Why should sales reiterate the stuff that businesses usually change slowly, especially when they’re large? Why would you have sales reiterate their position every quarter to reset the exact same thing and whatnot? This is nonsense. There is no value in doing that. And to get the agreement on this unified forecast, this is the fallacy. This forecast is something that is going to be numerically unstable. We have the market erraticity, so numbers kind of change a little bit all the time, but it’s just noise.

If we go back to Procter & Gamble, people consume shampoos. It’s a very stable market. Yes, there are fluctuations, some brands rise and fall, but it’s mostly noise. Companies have been in the business of selling shampoo for a very long time. Yes, there will be new products that are introduced, they have introduced dozens of new products over the last decades, etc. But all those things are noise. The way you want to deal with a market like shampoo, there are fundamentals. And if there is one thing to be communicated, it would be about that.

If we go back to this S&OP, the problem is that as part of the solution we have said that the solution had to go through shooting for a unified forecast. We have said that a unified forecast slash commitment, and I really challenge that it’s the proper way to do that. My proposition would be this is complete nonsense. If you bring people to the room, and I would very much agree that there is value to bring people to those rooms together, it is absolutely not to get an agreement on those numbers. Those numbers change all the time. This is complete nonsense to waste everybody’s time to get an agreement or even share insight on those numbers. This is a waste, a complete waste of time.

But insights are very important and those can be communicated. Let’s give an example of a very important insight. Let’s say you sell diapers. Diapers are critical for hypermarkets because that’s one of the products where, when the hypermarket has a stock out on diapers, customers, young parents, go nuts. They want their diapers and they are very sticky in terms of brand just because their baby is used to that.

This is the sort of product that is quite expensive and if you’re out of stock for a specific brand of diapers, parents would just go to another hypermarket and do their weekly shopping. So the hypermarket would not only lose the sale of the diapers, they would lose the entire basket. And they might actually lose the customer forever because then they get used to going to another hypermarket and they never come back.

This is a very important insight. For diapers we need very high quality of service. Okay, this is a very important insight, but do we need to reiterate that at every meeting? Are we going to revisit quarterly the fact that diapers are a product where quality of service needs to be super high? No, that’s complete nonsense. At some point, we need to have this being really conveyed and that might be a meeting once where this information is conveyed. But then for the next meeting, we are not going to endlessly reiterate the same thing. That would be a very poor use of the time of everybody to restate what has already been stated and should be embedded in the company in a different way.

If you have new classes of products or market evolution that changes the way you should think about quality of service you can revisit that and that would be an opportunity to bring people together in a room to discuss, for example, what does quality of service for diapers mean in the age of e-commerce? It will be a new question that needs to be addressed maybe once and maybe revisited every couple of years, but that’s not a cause for revisiting this question on a quarterly basis and certainly not on a monthly or weekly basis.

Back to those S&OP meetings, due to the fact that as part of the recipe that is being proposed in these meetings, the deliverable is this unified forecast. The things that get discussed are all the wrong things. What I’m saying is that if we say that the information flows through software by necessity, if there are meetings, it’s certainly not to discuss these quantitative aspects. The quantitative aspects, they flow through the software. Humans do not have the bandwidth to do it any other way. It’s wishful thinking to think that humans can deal with the numbers. The numbers will cheat you. You will deal with numbers that are so aggregated that it loses any insight. It’s not actionable.

For example, if I tell you that a company like Procter and Gamble is going to do this amount of sales in this region, if you don’t disaggregate per SKU, it’s not the level of information that is sufficient to take your decisions. You have your high-level information, but when you want to translate, you can’t turn a monthly sales forecast into daily decisions. There is just not enough information. You need to have the fine decomposition and you can’t turn a category forecast, let’s say a million units to be sold in the diapers category, into the split for the hundreds of SKUs that you have to maintain within that category.

So again, the numbers will flow through software no matter what and humans, they will be available for high-level information, but they are not numerical in nature. They’re qualitative and that’s very interesting. There is a lot of value. But again, if we look at what S&OP is, S&OP is absolutely not used. I’ve never seen that in any company to circulate qualitative information. I’ve never seen that, or never seen that under the name of S&OP.

Conor Doherty: Well, that’s the thing and that’s one of the benefits of the longer form conversations. At the start, it sounded very much as if you were somewhat against the idea of people being involved in any sort of meetings whatsoever. But it sounds more like there’s an objection to the formalization of waste that comes from S&OP. For example, you’re taking people who are not geared towards this kind of analysis and formally putting them in meetings where they have to discuss the kind of numbers that change on a regularity that they can’t possibly be expected to keep up with.

Joannes Vermorel: Yes, you see, the problem is that the deliverable, you know, you bring people together to deliver numbers and that’s where I say this is complete nonsense. This is absolutely wrong. This is not the proper way to do it in 2023. It was, and this is not even new, 20 years ago it would already have been a mistake.

The problem is that you are literally projecting a century old paradigm, think of General Electric in 1920, people get in the room and get those commitments because they are selling like 50 products. And yes, if you have such a small product range, you can have this quantitative information that flows through people. But nowadays, with the complexity, it’s not possible anymore.

I say it’s a waste of time, people know it and that’s by the way the reason why S&OP devolved into a bureaucratic nightmare. It’s because this is pointless, people know it. And so, what is the natural evolution in a company when you have something that is completely pointless? Do you think that the people that are the most engaged, the most committed, the sharpest, the most valuable people for the company want to be part of that? They know that it’s not going to be a game changer. They know that this thing is mostly a bureaucratic thing that just happens for whatever reason.

There is a self-selection mechanism at play. Over time, and due to specific factors, it’s apparent that certain activities don’t attract the brightest individuals. I’ve encountered numerous companies and can recount dozens of cases where a CEO told me about times when having people who performed “magic” made the company incredibly wealthy, or about strategies employed decades ago. For instance, we once managed to support a production process that no one else could replicate at the same cost, allowing us to dominate the competition.

We had a patented process that enabled us to produce a quality that was impossible for others to match at the same price, effectively allowing us to outperform our rivals. Similarly, there are stories of achieving such a competitive edge through unique strategies, but I have never heard of any company claiming that their planning process alone was so superior that it allowed them to crush the competition. I’ve never heard of that and I have never heard of any company that would tell me we outcompete our clients on planning through S&OP.

Bottom line is, it is wrong. People projected this paradigm that is a low bandwidth paradigm. Channeling those quantitative information through people, it works with low complexity and this is a low bandwidth solution to a low complexity problem. Now we live in a high complexity world that needs a high bandwidth solution. And that’s my point. If people meet, there is value. But when people exchange, what sort of value do you have? It is to convey things that resist quantitative analysis.

When you gather people in a room to discuss, for instance, what quality of service means for diapers in the context of e-commerce, you’re tackling an issue that isn’t solely about numbers. It’s a significant problem, but it largely defies direct quantitative analysis. There’s no straightforward way to distill this issue into just a few numbers. If it seems possible, achieving that simplification isn’t easy; it would require a considerable effort to generate those few numbers. Moreover, it involves an intensive internal process to ensure that these numbers accurately represent what you believe they signify. This need for alignment is important, but the deliverable is all wrong. And I think that’s the core about S&OP, it focuses on the wrong deliverable.

Conor Doherty: I understand all the criticism that has been made, but to give the devil their due, S&OP is pretty common amongst enormous companies. Now again, it’s not the appeal to authority, just because a Fortune 500 company uses it, therefore it’s good. But my question is, once you get past a certain tipping point in large companies, there will just be bureaucracy, there will just be meetings, it’s just inescapable. We’ve all worked for big companies. Now if you’re in a situation like that, you’re a Fortune 500 company, you have to have meetings, there will be some bureaucracy. What are some steps that could be taken to redirect that attention that is otherwise wasted with an S&OP process?

Joannes Vermorel: First, it doesn’t have to be that way. I have met some large companies that decided to just terminate their S&OP division entirely. Nothing bad happened. That’s it. You don’t, I mean at some point, what does a surgeon replace a cancer with? Nothing. If you excise a tumor, you don’t put anything in its place and it just works better. So first, I know it’s difficult to hear, but literally when you have a pointless bureaucracy, the best thing that you can do is just stop.

The tendency is for people to think in terms of additive solutions, where more components are added. This mindset becomes more pronounced as the size of the company increases, with a preference for adding elements to address issues. However, what about considering subtractive solutions? By simply choosing to stop certain actions, problems can be effectively resolved, and everything can function smoothly.

I’ve had discussions with companies where, when speaking with the plant manager, I’ve asked about their S&OP process that produces targets. The typical response is that they receive the mail every quarter and simply delete it. When they want to do their planning for production, they extract their own data through the ERP, crunch their own numbers, and do what is reasonable for their production purposes. They are quite happy with the quality of service that we provide, so we’re good.

This is the sort of thing where I say it’s absolutely pointless. You can remove that. My first suggestion would be just to remove that. Yes, there will be a lot of people screaming. When you’re a cog, you have a lot of people that just are cogs and they don’t see the parts. But you realize that those sort of things can be entirely removed and it will not prevent the company from operating.

My message to CEOs is that don’t be scared by the idea of just removing that. Nothing bad will happen. I’ve seen a few CEOs do that, just remove this division, and everything was fine. The problem is that it is broken by design. You can’t salvage that. The mission statement that comes with it, the deliverables, are wrong. The question is, how can I improve my fax machine to become an email? There is no upgrade path from fax machine to email. At some point, you just have to take your fax machine and throw it to the garbage bin.

If you think that you can turn a fax machine into an email system, it’s important to note that, while some people actually achieved this around 20 years ago by adding options to fax machines allowing them to send emails based on the fax content, that’s complete nonsense. I’ve seen some people doing that, writing an email in Microsoft Word, printing it, and then using a fax machine to send the email to someone. That doesn’t make any sense. The same goes for S&OP. It’s like a fax machine. It was a solution that made sense at a point in time because it was the best technology available. But fast forward, it is not something that makes sense anymore. If your question is how can you improve your fax machine, you just can’t. It’s a technological dead end and it’s a dead end in terms of a paradigm as well.

Conor Doherty: And my last question then, because we’ve looked at the solution from the perspective of the CEO, but for the well-intentioned people who don’t sandbag, the people who have specialized in S&OP, not the vendors, just people who’ve worked, specialized, they’re diligent, they’re dedicated, they want to make the company better. If that disappears, what next? What upskilling should be done or what’s the pivot there?

Joannes Vermorel: First, people need to realize what the problem is. Coordination needs to happen and it requires information. The information is flowing badly. I’m not talking about fancy stuff like “we speak and they don’t listen”. I’m talking about super mundane things like wanting to produce a sales report of what has been sold over the last 12 months and all you get from your BI report is garbage.

I mean, not even noise, garbage in the sense of it’s broken, the software is not working, you know, just basic problems. Information has to flow, but producing the report takes hours because the ERP, the piece of enterprise software is garbage, and so producing a report takes hours, or the numbers are wrong because there was an update for the ERP, and all the numbers are now incorrect, or you see plenty of things, super basic things.

If you want to find a solution, just go back to the initial problem statement. People need to be able to access information. I’m not saying something super fancy, just transactional information. Can all people access the transactional information that they need in ways that are super convenient? If not, then you need to make this happen. Then make sure that the information is integrated in one place.

The problem is not the lack of digitalization, large companies are digitalized. The problem is that usually there is not one ERP, there are four of them or four systems and so people want to say, for example, let’s ask a basic question: you want to know how much stock do we have of this product? It’s a very simple question and you would like to know for every single SKU that you have, let’s say 100,000 or more, how much do you have in stock and most companies are not able to answer this question not without a lot of effort.

I know that because at Lokad when we start an initiative, it usually takes us weeks to get this information correct because the information is in four systems, there is the ERP, the WMS, there are partners, there is stock that is in transit that needs to be taken into account, there is stock that is reserved, there might be a backlog that is only found in the CRM. So there might be half a dozen systems and if you want to answer a question as basic as how much stock do we have, it’s a lot of information that is not easily accessible.

So I would say again, if we go back to the basics, we have the problem of making this coordination happen well, it needs access to the information and it’s through software and you have to make sure that the base layer of the software is done correctly so that this information can be accessed properly by all the parties that need that.

If you’re an S&OP team specialist, I would say convert your team into a data lake team. Data lake is just all the basic information of the company needs to be accessible in a programmatic form to everybody. This is not BI. The problem with BI is that people think the mindset is “I give you numbers for human consumption”.

But the thing is, again, bandwidth and you cannot, even if you can speak only, maybe, you can absorb by reading probably an order of magnitude more information than what you can speak. So you can read much faster than what you can speak. So, you can expose people to a lot of information, to a little more information, by just presenting those numbers on the screen through tables. But realistically, people can’t do much with that. So, I would say the difference between a BI team and a data lake is that the data lake perspective is you want to make the data available so that people can work with the data with tools.

So you see, that’s a very big difference. A BI, you present a screen, and it’s supposed to be someone who just reads those numbers. A data lake is, you let people extract a one million line Excel spreadsheet, and they just run their macros on top of it. You know, the mindset is if I give you a table, I’m going to give you a big extract with everything, and then you’re going to potentially in Excel, but run the macros that you have engineered on top of that to do whatever calculation you’re interested in. But if I give you a spreadsheet, it is not intended as a human will read the data line by line. They are more likely to do a calculation in Excel or something and then be done with it.

Conor Doherty: All right, well Joannes, thank you. I don’t have any more questions. You’ve covered everything in quite a bit of detail. So thank you very much for your time and thank you very much for watching. We’ll see you next time.