00:00:00 なぜ価格設定がサプライチェーンに属するのか
00:00:49 サプライチェーンを経済の流れの支配者として定義する
00:02:10 中小企業が価格設定と購買を自然に結びつける理由
00:03:51 大企業がサプライチェーンから価格を分割する方法
00:05:30 主流のサプライチェーン理論が壊れている理由
00:06:51 エンタープライズ ソフトウェアが価格設定の隔絶をどのように増幅したか
00:08:44 サプライチェーンにおける所有権と実行の意思決定
00:10:19 断片的な意思決定と S&OP の限界
00:13:30 サプライチェーンのあらゆる意思決定を部門間で分割すべきではない理由
00:15:57 価格と価格設定ポリシーの明確化
00:17:21 価格設定ポリシーを成文化する必要がある理由
00:18:23 eコマースとB2Bにおける動的な価格設定
00:19:43 サプライチェーンのレバーとしての B2B と B2C の価格設定
00:22:47 プロモーション、割引、およびそれらの長期コスト
00:24:01 返品率にブランドへのダメージを含める必要がある理由
00:26:29 ブランド要素とサプライチェーンの決定
00:28:29 高級ブランドと割引による経済的コスト
00:30:06 主な障害としての官僚主義の惰性
00:32:09 既存のサイロ間の組織的なギャップ
00:34:02 サプライチェーンが独自に価格設定を行うべきでしょうか?
00:34:18 サプライチェーンは数値レシピを所有する必要がある
00:36:37 販売、財務、マーケティング要因をサプライ チェーンの計算に統合する
00:37:16 ファッション小売における価格設定の諸経費
00:38:20 価格変更の最適化と簡素化
00:39:13 現実的にストアは何回の価格更新に対応できるでしょうか?
00:41:15 手動による価格設定ポリシーが脆弱である理由
00:44:08 機能予測と価格設定ポリシー
00:45:20 価格変化に対する需要応答のモデル化
00:47:23 価格を意識した予測のためのデータ要件
00:47:47 リードタイムと迅速化のための機能予測
00:48:49 統合された予測による収益性の向上が期待される
00:51:44 価格設定、購入、割り当てのどこから始めればよいでしょうか?
00:53:23 意思決定の量、複雑さ、経済的影響
00:54:05 長期の航空契約と価格設定のリスク
00:55:04 サプライチェーンの重要な手段としての価格設定
00:55:52 ソフトウェアベンダーが価格サイロを強化する理由
00:58:20 最後に思ったことと今後のライブディスカッション
まとめ
価格設定はサプライチェーンから切り離されたものではありません。それはその中心的なレバーの 1 つです。価格は、需要、マージン、在庫リスク、プロモーション、購入の決定を左右します。中小企業はこれを自然に理解していますが、大企業は価格設定、予測、供給をサイロに分割しています。その断片化により、官僚主義、誤った予測、隠れたコストが生じます。解決策は、サプライチェーンが価格を決定することではなく、財務、マーケティング、販売からの情報を統合して一貫した意思決定を行うための数値レシピを所有することです。在庫ポリシーなどの価格設定ポリシーは、習慣、政治、または切り離されたソフトウェアに任せるのではなく、経済的にモデル化する必要があります。
Full Transcript
Conor Doherty: Okay, so Joannes, good to see you again. Now, today’s topic is probably one that I think will intrigue just about everyone in supply chain, and it also runs the risk of being very hypothetical. So, I’m going to do my best today to keep that as practical and concrete as possible because the topic is that pricing is part of supply chain.
In fact, your position historically has been that pricing is as much a part of supply chain, even if it’s not acknowledged as such. It is at least conceptually as much a part of supply chain as forecasting demand is part of supply chain, as allocating stock is part of supply chain. Now, obviously, that is not the world we currently live in. So before we get into potentially why, at a high level, why is pricing part of supply chain?
Joannes Vermorel: So, obviously, it begs the definition of supply chain. So, if you approach supply chain concisely and then go where you say it’s mastery of the flow with an economics viewpoint, so you want to optimize rate of returns. So, that’s the flow with the economic perspective.
Then, obviously, the price at which you’re selling is very, very consequential. The higher, the greater the margin and the lower the volume. You know, that’s just a law of supply and demand. So, if you assume that it’s about the flow and that the economic perspective is relevant, then obviously it is one of the major, pricing is one of the major levers that is at your disposal to just govern the flow, shape the flow.
It is just as consequential as the capacity for your warehouses or your factories, or how much you actually buy of raw materials or semi-finished goods if you want to transform them afterwards. So, that’s obviously one of the major levers. And that’s why I say, again, if we say that mastering this flow is what supply chain is about, then with this economic perspective, then obviously pricing is one of the major levers, and thus it belongs into this overall mastery.
Conor Doherty: And that’s why it is common practice in the vast majority of companies and has been throughout the entire history of capitalism. Is that what you’re saying?
Joannes Vermorel: No, absolutely not.
Conor Doherty: Is that what you’re saying?
Joannes Vermorel: No, that’s the strange thing. When I met CEOs of smaller companies, let’s say below $50 million per year of annual revenue, it’s completely obvious for them that the two things go hand in hand.
I mean, they literally think, “Okay, if I buy this, if I reach this minimum order quantity, I get this price, and thus I can sell at this price, and does it click or no? I really, really need to go to 2,000 so that I can have a further discount and I’m more competitive, and then it will click, or I completely give up. I just buy 100 units. My per-unit price is really bad, but I’m just going to do a favor to clients who don’t want a second, I would say, alternate supplier.”
You see that this sort of thinking concurrently, at the same time, the price at which you’re going to sell and the quantities that you’re going to buy. Again, for pretty much CEOs that are still very hands-on, who are having like a small manufacturing business, small brand, or a relatively small retail company, this is completely obvious. That’s literally the way they think, and there is always, “At which price am I sourcing, and at which price am I going to sell?” jointly.
And now, that’s the thing, when we go for larger companies, then those companies tend to adopt the codes, I would say, the sort of practices, mindsets that are established through supply chain circles, and then it becomes much more bureaucratic. Then it’s as if suddenly we had like a fork, where pricing goes essentially to marketing, or sales, marketing, product development, potentially sales, and the quantity and capacity of the flow goes to something that is supply chain.
You know, there is this fork, and those things that were really considered as jointly tied up to $50 million annual revenue suddenly become completely decoupled and completely independent. And for me, that is very strange. There are plenty of historical reasons for that. But fundamentally, my point is that the people who are in the right are small business owners who have this completely joint vision. This is a correct vision.
Conor Doherty: So, just to be clear there, are you suggesting that below a certain level, where let’s say maybe CEOs have more skin in the game because it’s a smaller company, perhaps it’s family-owned, etc., they would operate the business completely differently? So they would have the same supply chain people managing PO lines and setting prices and looking long term in terms of the impact of pricing on brand, etc. So all of that would just be one giant responsibility.
Joannes Vermorel: Yeah. And really, again, the point that I’m also making is that the classic mainstream supply chain theory is just broken.
Conor Doherty: Mhm.
Joannes Vermorel: And it’s interesting. It’s when you have people who don’t approach the business through those broken lenses. They just do it very differently and very intuitively differently.
Conor Doherty: Mhm.
Joannes Vermorel: And that’s the interesting thing. It seems that those insights, as the top management is less hands-on, you know, because larger companies mean they have so many other problems to manage, that suddenly this sort of joint, this sort of coupling that you have between essentially pricing and purchasing, because you see, it’s really, “What do I buy in order to sell at which price?”
You know, that’s where you have this real coupling, because if you buy something that you cannot sell profitably at this price point, then your flow will be stuck. Or conversely, if you can buy a lot more to get a much better price, and then you will be able to sell a lot more, again, you will flow a lot more.
So, this thing, interestingly, just seems to get lost. And what has been happening in the enterprise software world over the last three decades has really, really amplified this disconnect. That’s the interesting thing, even in large companies.
Again, generation of my parents, starting in Procter & Gamble, late 70s, early 80s, in marketing, but the thing was, the reality is that what people were calling at the time marketing at Procter & Gamble, at the time, those people were also responsible for the supply.
Conor Doherty: Yeah, you discussed that before. They were making operational strategic decisions.
Joannes Vermorel: So, fundamentally, they would be completely responsible for all the flow. And again, I think with the introduction of enterprise software, those business systems, it seems that over the decades this sort of coupling was kind of lost.
And it was lost mostly because the software vendors themselves had nothing really, I would say, convincing to offer on this front. So they kept the stuff completely separated, segregated. And it turned out that also one of the contributing factors for enterprise software companies doing that was that all the supply chain textbooks, the ones produced by academia, very fake, the ones that have numerical recipes, you know, formulas in them, not the sort of consultancy vibe sort of textbooks, the books that come from academia about supply chain were systematically, no, it’s still the case, systematically ignoring pricing.
I mean, you can have like 800 pages, thousands of formulas, and literally zero line that even mentions that pricing and prices exist and are consequential.
Conor Doherty: Okay, I’m going to pull this back down into the concrete because, as I said right at the start, I want to be mindful, because I think the general framing of most of the discussions we have is there’s broad agreement when we discuss ideas on ownership, and we might disagree when it comes to execution. I think that’s the way to frame it.
So, if we have a discussion about, I don’t know, better purchasing decisions, we would all agree on the ownership of that problem. We would all agree, there will be no disagreement, that that is a supply chain issue. The disagreement will be on, “Well, we think you should maybe forecast lead times. You should use probabilistic forecasting, financial ranking at the SKU level, at the SKU-location level.” And other people might say, “Well, I’ve got ABC analysis. I’ve got service levels,” whatever. So the execution is where we disagree, but ownership we agree on.
Today, we are presenting something that I think is fairer to say that most people, I’m not saying they disagree with you, but they’re not even aware that there is a debate on ownership here. So, the execution is way down the line in this conversation. Right at the top level, and I mean that both in terms of the discussion and in terms of if you’re talking to executives, what is the concrete justification for saying, “Your company is currently arranged like this. You have this org chart with all these attendant responsibilities. You need to revise that. You need to internalize to supply chain pricing, and here is why. Here is the concrete justification for even listening to the rest of this discussion.” What is your response?
Joannes Vermorel: Right now, in most companies, the division of labor when it concerns supply chain is very broken. You see, it is fragmented among so many people. And pricing is just one example among many, because you see, you have so many other problems on the division of labor.
For example, the people who are producing the forecast are not the ones who are producing the flow decisions. So you have another problem of fragmentation, of silos. And here, pricing is just a big offender, but it’s not even the biggest. It just contributes to the overall ineffectiveness of the mainstream supply chain theory and the mainstream supply chain practice.
The way companies tend to manage, survive, and still make profits is that very frequently you will have the official division of labor, and then you will have some unofficial one. And it’s the unofficial one that prevents the stupid mistakes from being made. You know, the things where there will be somewhere, somehow, some people in the organization that will be able to say that if they are not confident about the price point at which they’re selling, or they project to sell, that the quantity should be steered up or down.
And it may surface through some sort of tug of war in the S&OP process, where people are trying to push the forecast upward or downward. But that is really, I think, a very bureaucratic and very inefficient way to approach this problem. You know, ultimately, you can bury the lead through the S&OP discussion on the fact that we have, in fact, a pricing discussion which is kind of represented through a tug of war on the forecast numbers. But that, again, is so indirect that it’s almost guaranteed to be an immense source of confusion and an immense source of approximations.
A lot of money on the table because stakes are completely unclear and not really communicated, because people are not even debating the price. They are debating the forecast.
Conor Doherty: Well, so right at the end, you got to where I thought the actual answer might begin, which was leaving money on the table. Because what I thought, my understanding here is that if you’re trying to increase or maximize the rate of return for all of your supply chain decisions, then the argument here is by internalizing this mechanism, the pricing as a decision, by internalizing that to supply chain, you increase, or at least probabilistically or hopefully increase, the rate of return for your supply chain decisions. So what I’m asking is, what is the bridge between those two points at an executive level? We don’t have to go into the coding or anything like that.
Joannes Vermorel: I mean, the thing is that, again, most of the division of labor is completely obsolete. They assume that this task, which is, again, those flow options, flow decisions, needs to be fragmented for every decision among many people. It’s not just that we have many decisions and each bucket of decisions will be done by a different person. It is that every individual decision will come out as completely fragmented, and that’s fundamentally the sort of problem that you need to fix.
If you have something that just robotizes end-to-end the decisions, then suddenly you don’t have to fragment through a divide-and-conquer approach the thing across many, many employees and many departments. You can even keep your existing departments. For example, you can keep marketing being able to have a strong voice in steering the economic drivers that will be reified in this economic recipe. So that will be literally encoded into the recipe.
It’s fine. But what I’m saying is that as long as you think of your supply chain as decisions where every decision is literally split over or spread over half a dozen distinct employees, it’s a losing proposition. You will have no consistency. It will be very bureaucratic, very slow. And if you look at companies who are doing things very smart, again, Amazon, they just don’t do that.
For Amazon, the idea of this coupling between the price, the volume, the stocking strategy, it’s all in one. And it has been like that for two decades. That’s the sort of thing that is obvious for very small companies. It is completely obvious for, I would say, smart but very large companies like Amazon. And we have this sort of no man’s land where it’s very bizarre for the mainstream supply chains, where many, many companies appear to be completely lost.
Conor Doherty: I’m going to push forward with this, but I am going to return to that point. But I realize that I want to be very clear when we talk about pricing, what we’re saying, because again, historically, we’ve used pricing to refer to a few things.
Because I know in the book you talk about every internal option at your disposal before you make a decision carries a price. There’s an opportunity cost to that expected return. That’s not the same as the price that’s externally facing, like markdowns, prices, etc. And again, here we’re talking more specifically, when we say pricing, of the pricing policies.
Joannes Vermorel: Yes. That’s also a distinction. Not your prices, but your pricing policies.
Conor Doherty: Okay, there we go. Because, for example, you can decide, “My pricing policy is to match what my competitor is doing to the last cent. That’s my policy.” My prices vary constantly.
Joannes Vermorel: Yes, exactly. The price of competitors can be automatically monitored through competitive intelligence tools. So you can have your prices that are varying constantly, but your pricing policy is exactly the same. It is, “We are exactly at the level of the competitor.”
Conor Doherty: So here, when we say pricing, we’re talking about who is in charge of the pricing policies.
Joannes Vermorel: And in many companies, this thing is not even codified. So, in fact, there is a pricing policy in the head of certain people. Something happens, then this, but then they actually update the price instead of having a formula that is just updating the prices whenever the prices should be updated.
And again, it can be a smart formula. I’m not talking of stupid things. For example, it can be a formula that says, “If the price has already been updated in the last 48 hours, we don’t change it back again,” just to avoid having hyper, I would say, yield management sort of style of prices, where every single time somebody’s looking at the price, it is different.
But that happens, by the way, at Amazon. Amazon does that. You have some products where you just refresh the page.
Conor Doherty: Yeah.
Joannes Vermorel: And 10 minutes later, it is a different price.
Conor Doherty: Yeah. But for example, retailers, fashion retailers, can’t do that. Even just physically updating the prices.
Joannes Vermorel: Absolutely. Absolutely.
Conor Doherty: So there’s a threshold.
Joannes Vermorel: Yes, exactly. This sort of very price-dynamic prices are something that are going to be found in e-commerce or also for B2B. For example, if you are, let’s say, servicing parts that are for AOG, for aviation, you have your AOG desk.
Conor Doherty: Mhm.
Joannes Vermorel: Somebody asks you for a given part, you can say, “Okay, the quote is this much.” And now you are one part left, because airlines that have an AOG desk, they have to support their own fleets.
So if somebody else the same night, for example, comes and asks again for a quote for the same part, you might actually give a substantially higher price because now you are already depleted from one unit. And now there is a second client that shows up asking for a second unit, and that could endanger your capacity to actually keep your own fleet operating normally. So you have various scenarios where the prices can be super dynamic. It’s typically either e-commerce or fancy B2B.
Conor Doherty: Yeah, that’s actually a point I hadn’t written down, so full improvisation here. I hadn’t considered the difference between the applications of pricing as an organ of supply chain in a B2B context versus a B2C context. And you can branch out B2C in a moment because we can get into, let’s say, supermarkets versus luxury and the differences of elastic and inelastic prices.
But just at a high level there, do you see one as being more accommodating of supply chain pricing than the other? So, for example, would this be easier to install in a B2B context than a B2C context? Because there’s going to be pushback no matter where, but one might be a little bit more hospitable.
Joannes Vermorel: Obviously, I think in B2C the automation of pricing tends to be more, I would say, on the agenda just because it is very tedious to manually update thousands. Again, if you’re B2C, chances are that you are exposing thousands of references. Chances are it’s not the case always, but thus the amount of price points that you’re exposing is large. And that’s just manually updating all those prices can be very tedious.
Also, you tend to have a lot of competitive intelligence data because if you are B2C, then your competitors are also B2Cs, which means that their prices are most likely public. Thus, if they’re public, you have competitive intelligence, price watch, and you can have price-matching strategies. B2B varies a lot, and in B2B I would say the complexity that can go into pricing can go from very simple, similar to what you have in B2C, to insanely complex.
For example, insanely complex would be multi-decades-long maintenance contracts in aviation for a fleet. That is a very, very complex contract, and this contract will come with a price at the end, and there is an element of negotiation in that. So that needs to be taken into account.
So, obviously, when I mention that numerical recipes should be used, I am leaning more strongly in this sort of situation where hundreds of times a day, a price has to be emitted one way or another. It can be just through a quote. But I’m talking of situations that have a certain intensity of repeat. If it’s something where you’re negotiating a multi-decade-long contract, but you only sign two of those contracts per year, then it’s a completely different discussion.
Conor Doherty: There are a few points written down here, but I’m going to go to promotions because I know that a huge part of pricing, especially in a B2C context, is going to be, let’s say, markdowns, promotions, discounts, perhaps price spikes. These are all exercises of pricing.
But you have many, many times sat in exactly that chair and argued against many of those things, saying things, again, I’m paraphrasing, but one of the problems with discounts is you instill the expectation or the behavior of expectation in clients. So then the discount that you’re willing to accept today actually has a ripple effect.
Joannes Vermorel: Yes.
Conor Doherty: So the price that you pay on that pricing decision stretches farther in time than you think. But in parallel to that, there are contexts where companies have to consider brand. So then you introduce the tension where if your goal is to maximize rate of return at all costs, that can come at the expense of brand because a discount can tarnish the perception you have of a brand.
That was a lot, but how do you basically smooth the edges on that problem, where pursuing maximal rate of return, including pricing, can come at the expense of, let’s say, things like brand perception?
Joannes Vermorel: I mean, usually that is the exact opposite that is happening.
Conor Doherty: Okay.
Joannes Vermorel: That is the exact opposite. When people don’t pursue rate of return, they do things that damage their brand because the damage is not priced. You see, if you do not price the damage that you’re doing to your brand when you discount, then your teams are going to do too many discounts. That is very mechanic.
That’s a typical misunderstanding that I get. People say, “Oh, if you have this sort of economic perspective where you go for rate of return, that’s going to make you super short-termist and super naive.” I say, “Why is that? It will make me naive if what I put on the scale is itself very naive.” On the contrary, an academic perspective lets you have a lot of factors, including very long-term factors. Yes, it’s going to be guesstimation heuristics. Absolutely.
Conor Doherty: This is part of the drivers.
Joannes Vermorel: This is part of the drivers, exactly. But this will be integrated. And on the contrary, that’s what I see, is that a lack of integration, for example, creates all sorts of problems.
For example, for B2C retailers doing promotions, if you don’t take into account all the costs that you incur by having those spikes on your supply chains, by the bad habits that you generate for your customers, and the fact that generally you can also create quality of service problems, such as, for example, you start a promotion on something where you were already running low inventory-wise. So it’s really a bad move, but it does tend to happen frequently when companies don’t consider their promotions completely coupled with the supply.
You end up with a lot of friction, of cost. So that’s why I say the pricing policies, and indeed that includes promotions, need to be completely coupled so that you’re looking at the full picture with a unified rate of return perspective that integrates the long-term factors.
Conor Doherty: So I just want to paraphrase that and correct me if I’m wrong. Are you essentially arguing that there’s no functional overall difference between a brand decision and a supply chain decision, assuming that you have factored some sort of, it might be loose, but some sort of economic driver that represents the potential damage of discount?
Joannes Vermorel: I mean, when you say brand decision, the thing is that when we say supply chain decision, I mean something very specific. I mean an allocation of resources that impacts the flow. So, I send this amount of money to a supplier to acquire 10 units. That’s a decision because I allocate resources.
If I move units from point A to point B, I allocate resources. If I put a price point, I allocate resources because it means that I give up the opportunity to retain the inventory for longer, for a higher price. So those are supply chain decisions. They are resource allocation.
A brand decision, I’m not too sure what it is. I mean, we could say, for example, that a brand decision is we decide to register a trademark in this country. That’s a brand-level decision.
I would rather say it’s a brand factor, or a brand driver. If I do a promotion, it will have some kind of economic impact that I can measure at the level of the brand. But this is not a brand decision in the sense of this is a brand-consequential thing, but not a brand decision.
Again, a brand decision is really something that would be, if you’re thinking in terms of branding, we redo the logo. That’s a brand decision. Or we change the messaging to the market. That’s a brand decision. So it’s not the same world.
Conor Doherty: I agree. So what I meant, and I don’t want to get lost in semantics there, but what I meant in the world of pricing, if you take, so for example, I’ll sketch out a scenario. You’re Louis Vuitton and you decide, “I’ve got all these bags. I’m just going to experiment with pricing because my algorithm said if I want to sell as many as possible, I need to reduce everything by 20%.” That’s going to have an impact on the perception of that brand. Louis Vuitton is no longer exclusive. They’re dealing in discounts.
Joannes Vermorel: But that’s very simple, is that the brand damage for a luxury brand to actually do discount is huge.
Conor Doherty: Absolutely. That’s what I’m saying.
Joannes Vermorel: So it’s very simple. You’re going to write your equation and you have this factor that is so huge that you say, “Okay, this discount costs me like 10 years of revenue.”
Conor Doherty: Yeah. Yeah, true.
Joannes Vermorel: Okay, it’s not an infinity number, but that’s a number that is so high. Okay, I don’t do that. You know, the rate of return is abysmal as soon as you have an economic driver that just reflects that. And again, for a luxury company, doing discounts is really, really bad.
So you just put a large number, like 10 years of annual revenue, as a counterweight of any discount, and boom, it’s never going to happen. If you just price some negative actions as very expensive, then any numerical recipe that is not completely broken is just going to stay clear of those options. It’s just steering numerically what is happening.
Conor Doherty: What do you think is the major resistance to installing this? Because again, as I anticipated right at the outset, nothing here sounds, within the vacuum of this conversation, outrageous. It sounds perfectly reasonable, but once you try to institute that—
Joannes Vermorel: But the thing is that there is no resistance. There is just bureaucratic inertia. That’s a very different thing. It’s not that people don’t want that. The problem is that, and by the way, that is the reason why silos are so difficult to fight. There is nobody to fight against. Your problem is essentially a gap between the silos.
This gap has no employees. Nobody has a title owning this gap. So you’re fighting against an absence. You see, it’s much easier in a large corporation to point to a person and say, “You’re doing it wrong. You need to change the way you’re doing things,” as opposed to someone who is a non-existent employee, because the division of labor is making a concern invisible. And even the division of labor can make a concern impossible.
There is just, you’re in the gap. There is nobody to blame, and the way you frame your own organization makes it impossible to actually add someone that would be responsible for this joint coupling that we were talking about. So you end up with this sort of situation where it’s very much the inertia of the company that defines your blind spot. That’s why I think a lot of companies struggle, is that this thing is owned by technically nobody. It is seen a little bit by everybody, but the path forward is super, super unclear for every individual person.
And this sort of problem did happen in history many, many times. For example, many retailers literally completely failed into having an e-commerce division early enough. They let Amazon become a giant before they even started to really respond, because again, there was a gap. Conceptually, the whole thing was absent.
You could not blame, for example, someone who is responsible for a given store for the fact that they don’t have online sales because they are not responsible for online sales. So you see, if you have a gap, then it is very unclear. And here it is even worse than e-commerce because the gap is really in between stuff that already exists.
Because it’s in between stuff that already exists, you cannot just, like for e-commerce, where it would be an easy case where you just have to proceed additively, where you just create a new e-commerce division and you’re good. Here you’re into something where you have to essentially rethink a little bit how you’re approaching the problem.
And this rethinking, unfortunately, has to go all the way to the top because it goes really up to the CEO level, because that’s where the division actually starts. For example, the supply chain director cannot solve that through an internal reorganization within supply chain because pricing is not under his command. And the same thing, marketing director cannot solve that because again, it’s a different division. And the only guy that is on top of head of marketing, head of supply chain, head of finance is the CEO himself. That’s why it becomes so difficult to solve this problem that, when you think about it, is completely obvious.
Conor Doherty: Is it your argument that supply chain departments should own the pricing mechanism or have more of a say, like a feedback loop from the marketplace?
Joannes Vermorel: No, I think what they should own is the numerical recipe and its proper execution. So numerical recipes that steer the supply chain decisions, but they don’t have to own the parameters. You see, that’s the way Lokad operates.
Conor Doherty: Yeah. Very similar.
Joannes Vermorel: Supply chain is responsible for generating purchase order quantities, inventory transfers, production schedule, prices that are updated every day. But is supply chain owner of all of that? No. It is owner of the execution, of the integrity of the numerical recipes. But every other division has their say on the economic drivers that matter to them.
So, for example, the cost of money is an economic driver. Supply chain is not having the final say on what is the adequate cost of money. It’s going to be finance. Fine. Finance is responsible for having the final say on what is the economic driver that must be used for cost of money.
And same thing, marketing, for example, can be responsible for having the final say to say what is the way to quantify the brand damage in euros or dollars when you do a discount. And then they are responsible for this part. And then supply chain is only responsible for the proper execution of a formula that is truly faithful to what marketing has just stated and confirmed.
You see, that’s where it becomes very different. Instead of having all people together in an S&OP to push the forecast up and down, it’s more like every division has the final say on their economic drivers of interest. And supply chain itself is responsible for a few economic drivers that are really flow-dependent: cost of transport or these sorts of things. And for the rest, they are just responsible for making sure that the numerical recipes are faithful to the statements made by the other departments.
For example, if the quality of service is expressed by sales, and sales say, “Okay, quality of service, this is the way you need to think about it. This is how much we lose on the long run when we do not serve clients on time and in full,” then the responsibility of supply chain is, I mean, supply chain doesn’t necessarily own these statements in being able to override sales. But what they will typically have to do is they will own the faithful integration of this statement by sales into their own calculation.
Conor Doherty: It occurs to me that, because I’m trying to imagine what this sounds like to people who’ve never heard it before, and then transpose that into reality. And the context that I’m looking at is, let’s say, a fashion retailer. It’s just the example I always go to in my head.
And I’m just thinking about, imagine a store, jeans, T-shirts. Everyone has been in a clothes store at some point, I’m sure. And then I look at it from the perspective of what Lokad would typically, or any company like us would typically, provide to that store. And it would be, per day, “Here are decisions. Here are decisions on what to order. Here are decisions on how to assort, how to optimize the assortment. Here are decisions on how to allocate within the network.” Those are updated every single day.
If you’re adding prices into that, yes, it occurs to me that there’s a different kind of overhead involved in that, because obviously I’m not going to change the prices on the clothes every single day, though in reality or in a vacuum it might be optimal to do so. That said—
Joannes Vermorel: No, that’s the point, is that we are back to optimization versus optimization. Again, from an economic perspective, you have to take into account the fact that there is an overhead expressed in euros or dollars for every price tag that you need to update.
Conor Doherty: Well, that’s exactly what I’m saying. Literally physically having to do that.
Joannes Vermorel: Your pricing policy must factor among all the economic drivers the overhead into changing a price compared to what it was.
Conor Doherty: Man-hours just to even go around the store and do that, printing it, updating the records internally, there’s the overhead. And that’s what I’m saying, the overhead associated with that means that there is presumably a minimally viable time horizon for that, so it wouldn’t be daily. You wouldn’t even do that.
Joannes Vermorel: And it’s not at the product level. Imagine a store. You can probably tell them on any given day, “We have 10 articles where we need to update the prices.”
Conor Doherty: Yeah, that’s fine.
Joannes Vermorel: Ten. Assume that we have like four units in stock per article. That’s 40 articles that need to be retagged, probably manageable. But if you say 500 on a given store on a given day, then that becomes completely unmanageable.
So you see, it is fundamentally this sort of overhead. It is not something strictly linear. It has super-linear curves where the more you ask, the more intense the pain, but maybe in a quadratic or super-quadratic fashion.
So what I’m saying is that, again, you just need to properly factor the economic drivers, and the pain and complication inflicted into the store needs to be factored in.
Conor Doherty: Well, also even just on the ground, your staff would have to do that as well. If your job becomes monkey work, where I just have to go around and multiple times per day update prices.
Joannes Vermorel: But again, that’s why I say you need to have a numerical recipe, and this numerical recipe needs to introduce all the factors. And whoever, for example, is responsible for the network of boutiques itself, this person would have the ownership of the proper modeling of the overhead.
For people, that would be the person that says, “What is reasonable? How many articles can we ask a store to refresh daily?” And probably the person in charge of the network will say, “Well, it depends on how many people we have in the store.” So it’s going to be a rule that depends on the size of the store, on this and that, and this person might have the final word on how you compute in dollars the overhead.
Conor Doherty: So, I want to be clear ahead of time. I know what you’re going to say to this. I’m assuming a position here, so don’t assume, “Conor, why are you asking me this silly question?” But Joannes, if we’re talking about in that context of a fashion retailer, only updating a few articles per day, slash even per week, why can’t that just be done manually with a pen and paper or an Excel spreadsheet? As far as I know the answer to that question, but go for it.
Joannes Vermorel: I mean, you can. You have no clue what you’re doing, but you can. You can also roll a dice and four, five, six, you increase the price by 5%, and one, two, three, you decrease the price by 5%. Just roll a dice. It will give you something.
That’s why I say you need to have a pricing policy. So you have a price that is just inertia, that is what you had yesterday, but what is your pricing policy? When do you decide to raise a price or lower a price?
And if you just say, “Oh, my policy is Bob. Bob has been with us for 20 years. Bob sits in a room. He thinks very, very hard. Then light comes and, bam, Bob tells us what we need to do.” I would say, praise to Bob. But then what will happen when Bob leaves the company?
You need to have something that is a little bit codified, because otherwise it just means that you are one employee leaving away from a problem of just not knowing how to actually keep managing your prices. And if people say, “Oh, but no, don’t worry, there are five people. It’s really under control.” I say, “Okay, fine. So if you have five people, that means that you are able to train a sixth one. Thus, the thing can be codified.”
And then I don’t buy the argument that, “Oh, it’s so incredibly complicated we cannot put that into software,” because if it’s so incredibly complicated that you cannot put that into software, it’s probably completely bogus. There is no reason for pricing to be insanely complicated. This insane complication most likely is going to be super baffling for your clients. It’s probably wrong.
So your pricing policies have no reason to be infinitely complicated. And our experience has been that for almost all the clients where we have done that, it can usually boil down to less than 50 lines of code. People say, “Oh, it’s very complicated,” and bam, 50 lines of code later, we are done. We have their pricing policy. It’s not that fancy.
Conor Doherty: I could talk about the fashion retail one more, but I’m going to push on a bit because I realize we have been going for an hour, and I do want to get into at least a little bit of the technical detail. Because anyone who has listened this far is probably, “Okay, I want to hear a little bit of the technical detail.”
And I think the question would be, when we talk about demand forecasting, we’ll talk about probabilistic forecasting, generating scenarios, assigning a probability value to it. There’s a 1% chance that this amount will be sold, a 10% chance, and you add it up and you can make your decisions. Same with lead time forecasting. We talked last week that most lead times follow a bimodal distribution. See, I was paying attention. How does, like, visualize how the pricing is? Is it forecasted probabilistically? How does that work when you talk about folding that into the optimization?
Joannes Vermorel: Technically, what we’re doing is called a functional forecast. So you inject a function. The pricing policy is a function that is taken as an input. That’s how you will actually model the demand, is that you model the way your company will be responding to the demand by changing its prices.
So just to give you an example, if we know that we have a product where, if we lower the price, the demand explodes. That is very classic in, let’s say, electronics. There is a market price, and it’s a very, very competitive market because everybody is selling at the same price. Margins are super thin, but if you discount a little bit, so you’re like 1% or 2% lower price than your competitors, then boom, demand explodes because it is very, very tight.
So you can actually model that into your demand, and that will completely change your risk assessment on exactly what you will have as future demand, because you will take into account the fact that if you don’t have enough demand, you can actually lower the price and sell a lot more, albeit at a slightly discounted rate. That doesn’t mean that if you have a lower demand, you will not lose money, because probably the margin is still quite tight. But it shows a picture very different compared to, “I will be losing a little bit of money because I need to discount on an article where the margin is tight,” as opposed to, “I need to do a complete write-off.”
So the fact that your forecast is not only probabilistic but functional, so it embeds this sort of pricing policy inside, gives you much better trajectories that reflect the fact that if you have too much demand or too little, it anticipates how your company will itself respond by tuning its prices up and down as the situation unfolds.
Conor Doherty: And again, I presume that to install that requires the exact same data as any other. It’s just what is in your ERP. Dump it.
Joannes Vermorel: Exactly. And that means that you dump it to us, not dump it, not delete. It just means that the probabilistic forecasting machinery needs to be a little bit more fancy. But that’s really a technicality.
There are plenty of things where you need to have this functional perspective. For example, if you want to be really fancy with your lead times, you have the exact same problem. You need to take into account, for example, “Do I have the option of paying extra to expedite and shorten the lead time?” That will give you a completely different picture on how the future will actually unfold.
So, the policies that you have with regard to expeditions of the purchase orders is something where, whatever machinery you have to forecast, it’s the same sort of complications to be able to deal with a policy-driven lead time, or a pricing policy, same sort of complications.
Conor Doherty: Impressionistically, because I realize the question I’m about to ask sort of opens up the close, really, which is the discussion of the net benefits. But impressionistically, how much better does the decision become when you genuinely forecast all of these sources of uncertainty together, so demand, lead times, and you fold pricing into that?
Again, I realize it’s a total, you’re just taking a rough number, but are we talking about it’s an order of magnitude better? It’s appreciably better? It’s worth pursuing? It’s if you’re trying to squeeze the last few drops out of the orange? Where in the hierarchy of sources?
Joannes Vermorel: Again, if we’re talking of companies that are not superstars. If you’re Louis Vuitton and you’re already having your annual profit like 50% of your annual revenue, something that is completely staggering and extravagant—
Conor Doherty: Yeah.
Joannes Vermorel: I mean, most companies operate on, if you look at the profit generated, it’s what, like 7%, and that’s already like many companies operate with less. It’s not very good, but that’s just what it is.
And typically, this sort of inefficiency can literally increase the amount of profit that the company makes at the end of the year by like 50%.
Conor Doherty: Mhm.
Joannes Vermorel: Obviously, 50% of what they were previously doing. So you have a company raised from 7 to 10 essentially.
Conor Doherty: Yeah, exactly.
Joannes Vermorel: That’s that sort of thing. So it’s a few percent, but it’s meaningful. But when you think in terms of those few percent of annual revenue that are just direct overhead, when you compare that to the actual net benefit that the company generates, that is very, very substantial.
That’s why I think supply chain is usually underestimated in the capacity to really improve the profitability of the company, because those few percents of overheads actually hurt enormously at the end of the day, except if you’re a ridiculously profitable company to start from. Obviously, if we’re talking of gaining 2% extra absolute of your, if you’re already doing 50% profitability, it’s not going to be so impactful. But if you are a company that is really struggling and your profitability is like 2%, it’s really bad because sometimes you even have years where you actually lose a little bit of money. Then that can make an enormous difference.
Conor Doherty: Okay. A reality-based questionnaire, because typically any sort of optimization project, it doesn’t matter who’s doing it, doesn’t start with an end-to-end perspective. It will start with, “Hey, we’re going to optimize purchases for this division, and we’re going to optimize allocation for those stores,” etc.
Do you think starting with pricing is more sensible than something a little bit more intuitive, like purchase orders, etc.? Or where in the journey do you slide that in? Because we’re not naive. It’s not going to be, “Day one, we do everything.”
Joannes Vermorel: I think your mileage really varies, company from company. Here, I would say, back of the envelope, companies need to really have a gut feeling of where the most money is left. You know, how much money is there on the table by just making those things better?
Here, it really, really depends on many, many factors, such as how competent are the people actually doing it. And that’s a sensitive question, but in many companies, I mean, most companies, all departments don’t have the same quality of people. That’s reality. That’s what we’re talking about, reality. You can have a company where finance is really the stars, some where it’s really marketing where you have the stars, etc.
So if the thing is superbly managed, I buy it manually. Okay, it might not be the biggest lever right now. And then there is another factor, which is the volumetry. I mean, are we talking of five decisions a day, five figures a day, or 50,000, or 5 million? Obviously, the bigger the number, the more impact you will get if you adopt something that is software-based, like a numerical recipe.
And then, how complex is your factor? Sometimes, at Lokad, we have clients where we are just doing like 10 decisions a year, but the decisions are so complex, so impacting, so consequential, then having a very, very detailed, highly structured quantification of all the economic drivers is super key. Otherwise, they don’t even know if they are losing or winning money.
That happens with long-term maintenance contracts in aviation. And just to understand, the key challenge here is that the profitability of a new client will really depend on your existing assets, the parts that you already have, the machines that you already have, your repair capacity that you already own, etc. How much you will have to outsource a portion of the contracts, and how exposed are you to future fluctuation of the purchase price for future parts, etc.
So here, you can have situations where it’s very, very unclear at a given price point if you’re actually earning money or losing money. And once you close a contract, you sign a long-term maintenance contract, you can be stuck for two decades with a losing proposition where year after year you lose money on the contract.
Conor Doherty: So if I were to summarize at a high level, obviously pricing is a key part of what shapes the demand that you ultimately will serve. The ability to install that, or rather the ability to immediately install and reap benefits out of that, will really depend on the vertical. You’ve given a bunch of examples. We’ve covered aerospace, fashion, the practicality of that. So as a lever to pull, it is, I think, fair to say, a potentially very consequential one.
Joannes Vermorel: Yes.
Conor Doherty: It may not be the first one you go to, but certainly one to consider.
Joannes Vermorel: Yes, exactly.
Conor Doherty: Again, we’re not selling dreams here with this. It’s simply being realistic. Something to be aware of as an option, and depending on where you are in your optimization maturity, you might want to pull that.
Joannes Vermorel: Yes. And my concluding thought is that don’t let, I would say, bad enterprise software vendors drive you into their silos. Because that’s the thing. What is broken with pricing is that the mainstream supply chain theory does not even acknowledge the existence of pricing.
I mean, literally, I have on my desk a series of very thick textbooks, mainstream textbooks for supply chain, and I’m talking the quantitative kind, the ones that have formulas in them. And typically there is nothing about pricing. None. So the whole concern is absent, never discussed. It’s never even clear that it could even be a concern. So it shines by its absence.
As a consequence, many enterprise software vendors, in fact, delegate entirely the thinking for their software to those sorts of books. They claim that they have super AI-powered software stack or whatever, the inspiration of the day for their marketing department. But the reality is that when it comes to engineering, they just delegate to the cheapest engineers they could find, and the delegation goes really like, “Here is a textbook. Just implement whatever is written, and that will be fine.”
As a result, you end up with a product where pricing does not exist. So you will end up with many supply chain software vendors who tell you it’s not a concern because their product doesn’t do it, or when they do it, it’s completely segmented within the product. It is completely disconnected, and thus you have a forecasting module that is logically completely decoupled from pricing and vice versa.
And again, that’s super bad because when you have this sort of product, it will enforce the decoupling inside your company. So it will worsen the problem of the bad division of labor because now this bad division of labor will be intensified by the software layer, I would say by your applicative landscape within the company.
Conor Doherty: All right. Well, I’m convinced. I don’t have any more questions today, but I do know that it’s a topic that I really do want to come back to definitely in a live format. I know soon we’re going to, next live event will be Hidden Costs of Fixed Lead Times, but I’m definitely going to come back to this because I want to invite, there are very specific people in my mind who I want to invite, who I know will show up, who work in finance, and they will have some pushbacks for you.
And I do want to get some external thoughts on this because, as I said right at the start, I think it’s a very intriguing idea. I also think there are ways to make it work, but I also don’t want to be naive about the operational realities or the operational constraints that many people at least have. And I would like you to respond to them directly because they’ll come in with a different perspective to my own.
But in any case, Joannes, as always, thank you very much for your time. I’ll see you directly after this episode. And to you, thank you very much for watching. As I always say at the end of these broadcasts, if you want to continue the conversation with Joannes and me, reach out to us on LinkedIn. You can connect with us there, or you can send us an email at contact@lokad.com. And with that, we’ll see you soon and yeah, get back to work.
