00:00:00 Why planning should not happen in an ERP
00:01:13 Perishable goods: inventory value changes over time
00:02:45 The ERP as a system of record
00:07:52 Comparing products with different shelf lives
00:14:19 Soft constraints and customer expectations
00:21:52 Why allocation decisions do not belong in the ERP
00:25:56 The limitations of ERP planning modules
00:31:05 Clinical supply and competing priorities
00:32:39 Why service-level targets provide poor guidance
00:36:40 Measuring the economic cost of shortages
00:43:55 Cornering scarce supplies and competitive strategy
00:51:33 Planning across millions of possible futures
00:53:18 Commodity processing under uncertainty
00:57:39 Avoiding premature production commitments
00:58:26 Why fixed time-series forecasts struggle
01:05:49 The danger of planning around averages
01:09:08 Probabilistic thinking in everyday decisions
01:12:33 Manual overrides signal improper system design
01:15:01 Separating resource management from planning
Summary
An ERP records what has happened; planning evaluates what might happen. Confusing these functions creates rigid decisions based on averages, fixed rules, and arbitrary buffers. Perishable goods, clinical supplies, and commodities all involve uncertainty, competing priorities, and economic trade-offs that an ERP cannot adequately represent. Planning should therefore occur in a separate probabilistic system that evaluates possible outcomes and their financial consequences. The ERP should record and execute the resulting decisions. Manual overrides merely signal that the planning process itself has not been properly engineered.
Extended Summary
Much confusion in business begins with asking a system to perform a task for which it was never designed. Enterprise resource planning software is valuable precisely because it records facts rigidly: inventory received, invoices issued, payments due, goods transferred, and taxes owed. These records must be stable because accountants and auditors depend on them. But planning concerns an uncertain future, not an established past.
This distinction becomes obvious with perishable goods. Two units of the same product may have different economic values because they have different remaining shelf lives. Their value also depends on customer requirements, transportation times, available sales channels, and the risk of waste. An ERP can record expiration dates, but it cannot adequately decide whether a particular unit should be allocated to one customer, discounted for another, or held for a better opportunity. Hard rules such as FIFO merely conceal these trade-offs rather than resolve them.
The same problem appears in clinical supply. A universal service-level target gives a simple answer where reality requires a comparative one. When three programs each require three units and only five exist, no target can make nine units appear. The relevant question is which shortage produces the greatest harm. That harm may depend on therapeutic consequences, possible substitutes, dosage requirements, project delays, and competitive considerations. However uncomfortable it may be to assign economic values to such outcomes, refusing to do so does not eliminate the trade-off. It merely causes the decision to be made blindly.
Commodity processing adds uncertainty in yields, quality, market prices, capacity, and lead times that may extend beyond a year. Planning around averages cannot solve this problem. An average collapses many possible futures into a single number, including futures that may never actually occur. Adding an arbitrary safety margin does not restore the information that was discarded.
These examples point to the same conclusion. Planning should be performed by a separate numerical system capable of considering probabilities, economic consequences, and changing opportunities. Its decisions can then be transmitted to the ERP for execution and recording.
Manual overrides are not evidence of human sophistication so much as evidence that the underlying decision process has not been properly engineered. Repeated adjustments consume skilled employees without creating a reusable asset. A numerical recipe, by contrast, can be examined, improved, and redeployed.
An ERP need not be defective to be unsuitable for planning. A ledger is useful because it is rigid. A planning system is useful because it can adapt. Confusing the two does not remove uncertainty; it merely ensures that uncertainty will be handled badly.
Full Transcript
Conor Doherty: So Joannes, good to see you. Today’s topic is blunt, but I would say important. Stop planning in your ERP. Now, what I wanted to do today was slightly or is slightly different to how we would usually approach the problem.
What I’ve done in preparation for this is find as much relevant information as possible related to different verticals. that might be LinkedIn threads, conversations, case studies, not just from us, from competitors. Basically, I just scoured the internet to try and find as much information as possible about concrete planning in ERP situations. So, basically the before section of traditional case study. I’ve put these all together and made composite stories or like vignettes. essentially one vertical, a context.
Here’s a problem, an anonymized character. I’ve just made them up to make it easier for discussion. I’m going to pitch them to you. And then we’ll go back and forth concretely discussing what’s going on in this situation, how might we analyze that, how would we do it better, etc., rather than just an abstract discussion.
So, I presume you’re okay with that.
Joannes Vermorel: Let’s proceed.
Conor Doherty: Let’s go. Let’s. So the first vignette the context this is perishable goods brief context. So, this is a mid-market perishable goods distributor.
Rough order of magnitude, let’s say 1 to 5,000 active SKUs. That’s a huge range, I know, but just for the sake of discussion. Annual turnover, the mid hundreds of millions, multiple sales channels, stock value depends heavily on remaining shelf life. That’s a key constraint as we will discuss.
So the story, Marta, not a real person, Marta works in a perishable goods business with very long lead times. Key detail. Their ERP records inventory, purchase orders, expiry dates, customer orders, and transfers. Allocation and exceptions are partly handled through spreadsheets and planner overrides.
The ERP may show that stock is available by SKU, location, and expiry date. The issue is that units of the same SKU are not necessarily economically equivalent. Some may still qualify for higher value customers. Some can only go to lower value channels.
Some may require discounting. Some may become waste if not allocated quickly. So the planning question is how to rank the options when remaining shelf life, customer eligibility, margin, expected demand and waste risk are all interacting in the same situation for the same person essentially every day. So join us in this first vignette.
The question is what is an appropriate level of responsibility for an ERP in that situation and what parts of the planning decision process should be outside of the ERP scope.
Joannes Vermorel: The ERP should only again its misnomer should have been enterprise resource management should only be responsible for lot tracking. And when I mean lot, it means that this set of units that belong to the same lot and they will all have the same expiration date and thus that’s your you’re typically tracking in this sort of situation perishable goods products by lot. So you have lots that comes on and it’s like a shipment of banana. Boom, one lot.
And if the supplier next day gives you another shipment of banana, it’s going to be another lot. And again, this lot tracking makes a lot of sense because there you have the same expiration date and if something goes bad, it’s typically the whole lot that need to be thrown away. If you have like a sanitary problem on one item, it’s the whole lot, etc. So, so the ERP should definitely be tracking that. that’s what it should be concerned but it should not be concerned very much with anything that is forward-looking.
This is maybe at the point in time when you’re passing an order to a supplier the supplier gives you an ETA as an acknowledgement for when you will receive that can be in the ERP. That’s okay. But more than that I think is a mistake. The ERP should absolutely not be forward-looking.
The problem is that the future becomes fuzzy. You see for example the projected future value of your goods perishability it’s very much dependent on the shelf life. But it’s not it’s not an exact science. It really depends on the what you can negotiate with your various channels. it’s here we are assuming that it’s going to be probably something like a B2B business as opposed to be B2C and showing the stuff.
So there will be an element of negotiation with there might even be auctions in certain places. so this is this is not known guaranteed and thus it’s going to be a mismatch. So my take is that focus on the past. you need to have like an accurate reflection in ERP of what happened what is and this is it and all the things about that is essentially a projected future estimate needs to live elsewhere just because it’s your ERP is going to make it very crappy especially with this sort of dynamic where you have like a decrease of value where it will sharply decrease day after there you might even have phenomenon I don’t know the word in English la which is when you buy products that are for like vegetables water evaporate so you’re selling in kilogram but every day you don’t even have the same the things that are waiting the same stuff it’s for typically fresh produce you have to think that every single day you will lose even in refrigerated conditions about 0.5% of the mass of your product that would just evaporate. so that those shenanigans and again it is not exactly like a super precise because it depends of humidity in the air it depends. So that’s why I say essentially you want your in this sort of situation. So for this for this company I would say if you if you have stuff that is just facts and then calculation that are just rules but not heuristics that are hard rules like where it is something where there is it’s not an approximation.
It’s literally for example let’s say you import bananas and there is a tax on whatever produce that you produce that you import that’s good it belongs to ERP you want to obviously reconcile the payments that you issue to your suppliers with the balance of the supplier this sort of thing it belongs to the ERP and when I say it should not be forward-looking I would say It should not be forward-looking in the future if there is uncertainty. Counter example would be you buying now with a supplier and you have payment terms with the supplier that say 60 days after ordering you pay then okay then you know that you have a payment that will be happening at this date in the future this is a fact. I mean it’s not 100% guaranteed but let’s say 99% of the time it will happen exactly as anticipated because it’s super mechanical. All of that is the responsibility of the ERP.
Anything else lives outside the ERP. It should live outside the ERP because otherwise it will just completely crappify your ERP.
Conor Doherty: A key detail here that was surfaced during the during the description of the situation for Marta, again not a real person, but Marta, is that the economic value of the inventory really does depend on the available shelf life. And the thing is how should be it in an ERP or without an ERP or outside of an ERP, how exactly should Marta’s company be approaching or considering the tradeoffs between two units that they’re the exact same SKU, but they have different shelf life. So one has three weeks of shelf life remaining, one has 15 days of shelf life remaining, another has 10 days. The ERP may not have that level of granularity, but you’re suggesting that it would be critical to have that in terms of making better decisions.
Joannes Vermorel: Yes. And that’s where you have really to differentiate. Yes. The ERP should follow give you a rough approximation of the economic value of what you have.
But for an accounting perspective, you see, so you see the value of your ERP is just like a ledger. It’s an extension of the accounting system. It’s an extension of the accounting system and it’s just there to make sure that you’re not being defrauded that you know that money doesn’t evaporate, inventory doesn’t evaporate. So you need for your auditors and you need for your accountant to have this information.
But it doesn’t have to be like super precise. It’s it is actually better if it’s rough and simple and very straightforward to understand. Again, what is the point is think about your auditors. Your auditors, they just want to make sure that nobody’s stealing from the company, that there is no fraud that is happening.
There is no nothing that where the company could be blamed for not paying the proper taxes and whatnot. So you and yes, there is depreciation and all of that needs to be reflected in ERP. But again, simplicity is king here. What you want is effectiveness of your I would say of your of your accountant and auditors and that’s it. for supply chain purposes and decision-making purposes, you probably want something that is much more refined that is but also I would say probably a lot more clever and thus more opaque less suitable for very precise auditing.
I mean here you’re playing a different game. If you’re the auditor, what you want is to make sure that if inventory evaporates, it’s really water evaporating, not people stealing from the inventory. That’s the sort of things that you want to make sure and you want to make sure that there is not like for example a fraud where one of your employee would systematically discount vastly a piece of the inventory and then sell it to a company that is managed by his cousin. you know that’s the sort of shenanigans that you want to really look for. Is there somebody who is just say oh this inventory yes let’s have like a 90% discount on it because ah I think it’s not really fresh and then I sell it to my to my cousin.
You see that’s the sort of problem. So your ERP should be a first line of defense for inspection, discovering this sort of shenanigan, but this is not about making like super clever supply chain decision and those clever supply chain decisions, they will have a lot of heuristics, a lot of things that are I would say that cannot really be explained to auditors or it would take a lot more time and in fact a level of finesse that is way beyond what you need to just detect frauds. and I would say problems that are on the controlling plane of your company here. So that’s those sort of things they should not live in the ERP and they should be really like finely crafted numerical recipes that are just there for the benefits of increasing the profitability of the company. for example to have small heuristics that say maybe depending on the season the those products will not in term of value lose value at exactly the same pace that’s a sort of finesse that should exist in the supply chain numerical recipe but certainly not on anything that is exposed to accountant auditors and whatnot
Conor Doherty: But isn’t this where the expertise of people like Marta would come in where they have in the current context they have manual overrides they track or at least try to track this and estimate it through spreadsheets so that they can intervene exactly what you’re saying.
Joannes Vermorel: But I would say why do you want to override anything ERP? You see, I would say this thing do not belong in ERP period. Don’t even try to override anything ERP. Again, if you can override, you’re going to make the life hell of your auditors and consultants. what do you mean this value?
It just change like dynamically. Whenever anybody thinks it should change, it change. Oh, incomprehensible. No, for your auditors, you should have like a super basic rule that say okay super fresh lose 30% of its value every single day.
Think strawberries. Yeah, that you know that is something that is auditable. Nobody tweaked that. Maybe once in a year people will revise.
Oh, strawberries we have like a new variety that can last a little bit longer. So it’s not 30% a day of value lost. it’s 25. Okay, fine. You do your override.
But if it’s something that you feel that you have to do like daily overrides as a supply chain planner, this is this is your ERP is whole is doing something it should not be doing or you’re doing something with your RP you should not be doing. those things should not ERP should not even be concerned with this sort of things.
Conor Doherty: You kind of alluded earlier to I think the term would be soft constraints because here we are dealing with the B2B context where again Marta’s company let’s say receives they have long lead times so they receive the perishable goods it took five weeks longer than they thought it would so you have to shave that off of the shelf life then you have to arbitrate all of that between your suite of customers some of them are going to be let’s say A-grade B-grade whatever however you want to classify them your A-grade customer wants minimum of 5 weeks of shelf life. That’s what his minimum is. But I mean, others might take four weeks if I relax the price a little bit. So there’s a constraint, but it’s not a hard constraint.
And that I think starts to surface the economic problems with the these kinds of decisions. They’re multi-dimensional. So it’s not just it’s not just SKUs; it’s customer or your own customer profiles. It’s the expiration date, how that interfaces with the prices themselves that you would charge.
Thoughts?
Joannes Vermorel: Yeah, I mean all those soft information I think do not belong in the ERP. You see?
Conor Doherty: Well, why not? because currently they do in this situation.
Joannes Vermorel: The thing is that you have to think any change that you want to do in your ERP cost 50 times more if not 100 times more than if you do it elsewhere. Why? Because whenever you touch ERP, you can actually break the accounting of your company. You can literally create a situation where you’re violating the tax code just because you make a mistake.
So anything that is the ERP is the beating transactional heart of your company. It is extremely sensitive. It is very sensitive to downtime. If you just shut it down for a minute, things stop flowing.
So maybe you can shut it down during the night but maybe not because maybe you have like an e-commerce service that depends on the ERP being live and whenever the ERP is not live the e-commerce is shut down etc. So my take is that The ERP is due to the fact that it carries a transactional core it is extremely critical and thus you should not touch it willy-nilly certainly not for things that are very transient that could be changed. You know for example you decide today that you want to represent every customer comes with a freshness target. But then 6 months from now, you realize, oh, I know it’s actually more subtle.
For produce and vegetables, they have two freshest targets and they are not the same. And then you realize six months later, oh no, in fact they are willing to have freshness targets for a small percentage of their of their delivery that is outside the target and whatnot. You see, it’s really you’re modeling the kind of expectations of your customers and those are shifting and it’s really because it’s B2B, it’s ongoing discussions that is forever ongoing and you’re not having like a millions of clients. You probably have like 20 clients that represent you know the bulk of if you’re B2B of what you’re selling and you have those ongoing discussions.
So those parameters that govern you know the perception of quality of quality of service by the client and what is and also willingness to pay. Yes. Those things do not belong in the ERP. It’s way too elusive, too soft.
It can be revamped. You know, you have to think the ERP should be like an immutable bedrock. Any invoice that was produced cannot be undone. you’ve produced an invoice, the only thing that you can do is in produce another document that cancel this invoice or these sort of things. know you have to think that the ERP is really by design extremely rigid and you should not I would say increase the burden of this transactional core with stuff that has completely non-transactional with stuff that is completely up to change at any point of time that is in large part completely subjective and it’s completely okay you know I if two person have a different opinion This should not have any impact on ERP. You see ERP is not there to be the consensus of various subjective opinions.
What I’m describing here the parameterization of numerical recipes to govern you know decision-making processes. You can have a lot of things that are not completely consistent and it’s fine. You can even have like I would say at the same time things that are completely inconsistent because you’re A/B testing two different strategies. Don’t try to have anything that would be like inconsistent to your ERP.
That’s going to be hell. your ERP if you have a rule that say this product is worth that much. Nothing should contradict that. You know, you can’t AB test various valuation of anything into your that’s going to drive your auditors mad, your accountant mad. And yet that is daily I would say bread and butter for the supply chain practice.
So you see that’s where I say those things the those parameters they need to live outside. It can be just a tactical app you know a very small app that just record all of that but it doesn’t have to be the it can live on the site doesn’t have to be part of those six years long upgrade and everything.
Conor Doherty: When we’re talking about perishable goods, there’s always going to be a degree of waste. Like no, no, there’s like no solution, no approach, no software is going to completely eliminate waste. But and certainly not if you’re dealing with very long and varying lead times. So by definition baked into that just that operation is going to be an amount of waste.
It’s just that’s just the way the cookie crumbles. Okay. But there’s still a portion that’s within your control. And I know let’s say in Marta’s situation, one of the big drivers of waste is allocation and allocation in this context is very much either handled by hard rules.
So the context like FIFO first in first out who what customer needed it first doesn’t matter where geographically they are located. So for example, let’s say the perishable goods take a long time to arrive in their destination country. they then still have to be transmitted across this country. So you might have somebody who requested I don’t know the product first, but they’re the farthest away unless there’s a manual override from Marta or their team. That food will then be transported, the perishing food will be transported across the country, shaving even more time off, which of course has downstream effects in terms of waste, negative impacts on relationships with customers.
So if that’s not handled through override and just hard rules, what is the way that they should be at least considering it?
Joannes Vermorel: So the mistake here is that allocation doesn’t belong in the ERP. You see, this is a decision-making process. This do not belong in the ERP. What belongs in the ERP is acknowledging that the allocation decision has been made.
Conor Doherty: But what if they have a planning module? If there’s a planning module attached to their ERP, you know, come on. That’s come on. Come on now.
Joannes Vermorel: Yes, but it’s the sort of things that you should not have. You see, just it’s you see this is a class of ingredients that just do not belong to your ERP. I love marshmallows, but if the architect tell me, oh, you know what? I think I’m going to use this incredible material called marshmallows to renovate, you know, a portion of your building.
I say no I love marshmallows but it do not belong to this class of you know undertaking. Yes. So you see that’s so you have to recognize and that’s the trick is that ERPs there is a fine line towards you know is it something really mechanical because you see ERPs have a lot of rules to apply automated behavior. You for example if there is a tax to be paid you have this transaction bam tax applied automatically.
So you have all this automation and when it’s things that are very mechanical where it’s not it’s not eligible it’s not an eligible behavior like you pay the taxes sorry it is mandatory you do it there is no I can opt out if I want to so that belongs in the ERP but allocations is not you see the problem is that people tend to implement this sort of decision-making policies because from afar it seems that it fit a little bit this sort of mechanical workflows that’s where the information lives that the ERP can implement but you should not do that because fundamentally ERP is absolutely not designed for that and yes you can implement a FIFO allocation rules and whatnot that’s going to be a version of this decision-making policy but it’s going to be like a super crappy naive one and what I Okay, just don’t do that. Do not have this sort of automation inside the ERP. Let have another the numerical recipe should leave outside and then you can do things that are way smarter.
Conor Doherty: But if the again the obvious push back there is Marta or any of her colleagues heard that they might say but all the information all the necessary ingredients live within the ERP. or all the expir expiry dates or the quantity
Joannes Vermorel: Not all the necessary information we’ve just discussed those meta information such as sensibility of customers there is plenty of relevant information that do not live there you have competitive intelligence you have again I would say yes ERP is probably like 80% of the raw information that you need but there is 20% that are actually very critical and that should not be in the ERP so this should live elsewhere. And then again the problem is that ERP the only things that you can even implement is going to be very simplistic rule-based logic and for your numerical recipe whenever you want to deal with any kind of uncertainty or any kind of I would say comparative economics where you want to wait oh what is the value of sending this unit here versus over there your ERP will fall the ERP is absolutely not designed to deal with any of that. And when I say ERP, I mean all the ERPs of the market that are all built on the same SQL core, transactional core, which is like the de facto standard for almost like everybody.
Conor Doherty: And just for the record, the position there is that also extends to any bolted on planning modules, allocation modules that are just the point is that
Joannes Vermorel: If for your planning module to survive this constraint, it would need to be completely isolated and then at some point you’re I’m just asking if it’s done right, if the planning module of the ERP is done right, it is so completely decoupled from the ERP that it is effectively an independent product you see it’s just
Conor Doherty: Yeah I mean because of the requirements the computational requirements essentially is what you’re suggesting
Joannes Vermorel: Yes and also the problem is the life cycle again just imagine your ERP the think of term life cycle of software products the ERP is a massive pain to touch to modify it all your it will whenever you touch the transactional history it will drive your accountant nuts it will create massive audit problems. It will it will vastly complicate the life of your auditors. It will complicate everything. So the idea is that the transactional core is something that you want to touch very minimally.
And that’s also why most companies when they want to upgrade the ERP the problem is why does it take like half a decade? The answer is because you’re touching the transactional core and you have problem that cascade everywhere. So that mean that the ERP moves incredibly slowly you know it’s and by the way if you look for example even the ERP on the markets you can look at NetSuite and I remember starting to actually use NetSuite in 2002 and it still looks exactly like nowadays what it was looking 20 years ago I mean 25 years ago in fact so that’s the sort of things where technology evolves very slowly and that’s why you have upgrade that takes like half a decade and sometime a decade on the other hand what should be the pace of change of your planning module at Lokad as a rule of thumb the numerical recipes that we deliver to our clients are tend to be entirely rewritten every 18 months so it’s like a continuous process where it’s being re I would say edited And after 18 months, it’s essentially something that is 100% complete.
Conor Doherty: How frequently do you want to deploy changes into production for your ERP?
Joannes Vermorel: Once a month and even that can be a little bit chaotic for numerical recipe for your supply chain probably like twice a day and sometimes even more depending. It’s so you see it’s really different world. Yes, it’s software but it’s software that abide to rules that are completely different. The time horizon are completely different.
The criticality also is completely different because you see c can you go fast and loose on your transactional core? No. Can you be fast and loose on your decision-making policy? Yeah, sometimes yes.
Conor Doherty: Certainly not. Because sometimes you have just something big that happen. Yes. and you need to react or you’re guaranteed to lose a lot of money like including in aerospace. I don’t want to go off into aerospace but like for example you can’t sit around all day you have to recalibrate but
Joannes Vermorel: For example let’s say one of your you’re doing perishable products you are it’s overseas you have two main competitors for some reason there has been a spike of demand and your talk two competitors on let’s say let’s say oranges are out of stock so you are now for a very limited time frame maybe a few days the only like all sellers who still has oranges. The normal rules do not apply. You can you can massively increase your price temporarily. Sorry, it’s exclusivity and then you will come back to normal times.
You see, there are plenty of things and that would be leaving a lot of money on the table to not take advantage of that. So that’s why I say the ERP the time frame it’s like years decision making policy for perishable goods we are talking of something where you think next day and if there is a lot of money to be made by really taking into account the exact now conditions the conditions today then let’s proceed and it’s okay if you have to tweak your numerical recipes twice a day.
Conor Doherty: All right. I realize that we could spend the full hour on this, but I do want to push on because there are lots of the same way the perishable example surfaces the idea of, you know, available shelf life and that determines value in clinical supply, high service level or at least an attitude towards high availability is a core concern. So, let’s push on. Here’s the context for clinical supply.
So, a large pharma/life sciences environment, dozens of concurrent programs, hundreds to low thousands of active material codes, direct inventory value is only part of the risk since delays can create multi-million-dollar consequences. So, situation again, Elena, not a real person, works in clinical supply. The ERP backbone records item data, batch status, inventory, orders, and movements. Planning remains partly project-based with high, and I stress this, high service expectations, buffers, and manual scenario assessment.
Now the organization has a strong service level attitude because obviously shortages can delay important programs which have huge knock-ons in terms of availability of critical drugs. Excuse me. The environment has become much more complex in the last 12 months. More programs, more variance, more demand uncertainty and much more competition for the same constrained resources.
So while a high service target is understandable, if every project receives the same attitude of protection then buffers increase and scarce capacity is consumed. So the planning question is how to prioritize across competing needs when capacity, inventory, time and risk are all constrained. So Joan, I know historically like the service level perspective is one that we’ve been skeptical of. I think it’s fair to say in the clinical supply context at what point does a high service level at what a high service level is presumably justifiable to a degree but at what point does it cease to be the driving factor
Joannes Vermorel: From the start from the start you can’t have 0% service some amount yes but yeah you have to think of it the problem with the whole service level charade is that it’s a threshold you Just think of it. Imagine a high dimensional space. You have plenty of dimensions. Yes.
And whenever you have like a binary criteria like you’re in, you’re out. You’re excluding a half space. You know, you’re dividing your space into and there is an area that is like don’t go there. Okay.
So you can you can and you think when you cross all your constraints what you’re doing is you have your high-dimensional space and you every single time you remove half of the space you know just because it’s these constraints define there is a side of the hyperplane that is good and the other side that is not good I know it’s a little bit abstract highly visual for people listening by the way but think of it once you have this intersection yes of constraints what you end up with is nothing empty. Okay. So now you have applied all your constraints and you just realize that you don’t have any solution because all the sigma of all your constraints gives you the amount of potential solutions for allocation and whatnot is empty set. What do you do in practitioner terms?
It’s literally there is three guys that ask each one is asking for three units. Yes. I have only five. What can what can I do?
The answer is if you only think the above service level nothing. It’s you can’t satisfy those things and your tool just gives you a binary answer. So, so you see that’s the problem is that if you think your supply chain in terms of again hyperplane where it’s like the yes area or the green area and the red area and that’s it binary split you’re screwed. You see it’s this vision assume that you can have like a good solution that would satisfy all the constraints and the short answer is you can’t not being excessively expensive and it’s also completely not realistic because there are many situations for clinical supplies where it’s not yes and no.
For example, there can be if you have like I don’t know a bottle of 10 milliliters and they ask for 10 milliliters but you have a one that has 20 milliliters it’s still a valid substitute you see. So, so you have you have substitutes or maybe it is like Yeah, I see. It’s a there is also things when you have active products, it can be just packaged in different way with pills or with other sort of conditioning. it can be a package of again different branding. It can be a package that is exactly same package but it’s written in Spanish instead of being written in French.
So slightly annoying but not really the end of the not necessarily like a blocking problem. so there are plenty of situation what I’m saying what I say here in this sort of clinical supplies is service level is poor because it just give you a binary yes no answer. You see that’s where I say it is super crappy because it is not actionable. It is it’s in a perfect world where you can see the future perfectly you would orchestrate and then your service level would all be green because you never miss anything. Okay.
But in an imperfect world where the future is very fuzzy those sort of binary criteria are essentially useless. What you need is assessment of the exact cost of this service and at a super granular level. Back to the sort of situation, I had three clients that asking for three units. I have only five question what is the economical cost for the first client if I don’t serve one unit?
What is the economical cost for the second one if I don’t serve one unit and for the third? Maybe each one of them have a completely different perspective on the case. So my allocation need to reflect that.
Conor Doherty: There is not to cut you I agree but the thing is in this spec in I think it’s Elena in Elena’s context it isn’t purely an economic consideration because when you’re talking about availability of let’s say reagents for clinical supply trials which then impact let’s say potential fatalities or life preserving. There’s another dimension to this we have to I would say
Joannes Vermorel: You can’t opt out of the economic perspective. you see you cannot say there are fatalities thus it’s here it would be
Conor Doherty: What you’re saying is that oh we have like supra-economic goals. Yes. And the answer is no you don’t because as soon as you admit that you have supra-economic goals then the economic calculation just becomes irrelevant which mean that you could in theory have to pull all the resources of the company to solve this one instance. You know it is something if you say something is categorically superior to all economics then what you’re saying is that it is okay to dedicate to like ultimate sacrifice. you do everything for that. what you’re saying is that and the reality is companies do not do that even if you have a risk of fatality you need to put a very high price but that’s like a stockout penalty essentially like what is the un what is the cost of not we can not to be too not to cut you but again it’s just to put maybe more slightly more palatable terms on this because I agree with you
Joannes Vermorel: The reality is that you even if it’s a dire consequence you still need to price it because you need to be able to compare that to all the investment. You know, even if your company is rich and prosperous and etc., you still have a finite budget. Even if you’re a largest pharmaceutical company, extremely successful, let’s say for example, you still have finite resources even if they’re vast. So you cannot say this thing is like infinite value.
No. it’s for example, no probably no patient is worth a billion euro. Why? Because with a billion euro you can save hundreds of patients by revamping entire hospitals.
Conor Doherty: Well, this is medical ethics.
Joannes Vermorel: Yeah, that’s a good point. It’s so you see it’s so there is there is a tradeoff there is a value there is a trade-off and yes it can be what I’m not saying I’m not saying it should be cheap. I mean if you have a situation that is dire then the price can be very high and it will reflect that it indeed it is very dire and that you don’t want to take this decision but nevertheless you need to have this economic calculation and that’s how and you see unlike the so you need to have a quality of service that value the economic impact negative of this service yes and this will be unit by unit. It’s not service level and yes, no, you’re good or not good.
It will be super granular thinking is it something where I can I be a little bit below the quantity or will it not work at all? There are for example some drugs where if the patient takes stop doing that the taking the drug even once it completely disrupt the process that the drug was supposed to deliver. There are some other drugs where in fact a reduced dose is not ideal but still kind of okay. There you have you have situations where for example if you get half of the dose that you’re supposed to get you will still have like 90% of the therapeutic effect.
So it really depends and some where if you don’t have if you have half of the dose you will have 0% of the therapeutic effect. So you see all those subtle nuances need to be reflected and yes it is fairly complicated but the reality is that your service level is completely bogus that’s that gives you something that is extremely easy but also just essentially it’s a it’s a mathematical fallacy. It just gives you a green light that has the appearance of being like rational but it is not.
Conor Doherty: So again, much like in the in the perishable goods example, the kinds of I can’t remember the exact term you used earlier. it wasn’t hazy. meta, I think it was perhaps it was meta factors, something like that.
Joannes Vermorel: Oh, the meta parameters meta parameters.
Conor Doherty: There we go. Yeah. So for example, it’s not margin. It’s not something that’s very tractable and obvious.
The idea of I used the term stockout penalty. We wouldn’t frame it exactly like that, but it’s the same sort of mechanism between zero and one. what is the value to you the economic impact to you of not having that reagent at this time for this purpose? What is it?
Joannes Vermorel: And again it might be something that is completely nonlinear in the quantity that you’re missing. You see because maybe some clients some channels can actually you know survive operate with a shipment that is just divided by two. maybe they can they have like slack maybe they have ways to deal with situation and maybe not. So you see that’s where it’s it is not just again do not think a binary criteria think of the economic value unit by unit. You need to have this marginalist perspective and also it is not one product because sometimes you have to think of for example if a drugs need two compound to be actually of any use then if you send one unit but not the others the value of this one unit is just zero.
So, so you see you it needs to have this full perspective of what are we even what is exactly the structure of the perception of the unit of need and if the unit of need says I need to have this and that and that otherwise I have nothing potentially but in a pharma context. Yes. Exactly. Okay.
Conor Doherty: There’s one point that actually I hadn’t written it down but it’s one that I wrote down here as you were talking which is again like the meta parameters I know I wrote this scenario that in Elena’s context so again very large pharma company there’s enormous competition to produce very much the same drugs so everyone’s producing more or less the same things so there is a sort of FOMO factor to this as well so it’s not just high service like stock huge buffers to make sure that we have what we need when we want it, but also I might want to splash a little bit of my own cash to starve my competitors of the same the same resources that they need to compete with me. Is that also something that should be factored in and can you
Joannes Vermorel: Yes, absolutely. Absolutely. I mean, it was more too nefarious about it, but more than a decade ago, Lokad started working for, a company that was very successful in actually cornering supplies. And by the way, they were they were doing it for fashion, not for clinical supplies.
So they would literally sometimes famous brands would just undershoot for their next collection and they had like something that would sell very well and they would just go and buy all the inventories that they would have and then they would they would become a little bit by accident the exclusive provider for supplier for a brand and these products just because they had acquired all the inventory. So yes, cornering supply has a long history. it is absolutely it is a gamble. It is it is risky because you can also end up with a massive excess stock of stuff that depreciate. But yes it and that’s again that is something where you need to do a risk analysis and only the economic perspective can tell you if you have a positive rate of return on what you because you’re saying I’m going to allocate this many dollars or this many euros in securing this and then there is all the possible futures that range from huh I did this allocation for nothing.
The market is completely fine. I did not have manage to achieve any kind of squeezing cornering effect that I expected to. I did a massive corner and I can I can raise my prices for my products like crazy and I take market share from my competitors who cannot even operate because they can’t access the raw materials that they need. You see?
Conor Doherty: So, and that’s already moving past this the pure service level perspective because again you’re looking more broadly
Joannes Vermorel: And here you’re not you’re absolutely not thinking in term of service level. listing do not competitive. It’s a it is exactly a competitive behavior where you have to think of all those scenarios. and the idea being that you’re not about you’re not buying to fulfill service level. you’re buying to go way above because you say if I if I corner the supply then I can starve my customers my sorry my competitors from the supply that they need and thus their customers will come to me and the demand will explode for me. Yeah. So, so you see, but then you also need to think do I have the production capacity to you know if I corner all the supplies, can I still deliver?
Because maybe if you if you have all this inventory of raw materials and then you are like oh crap I don’t have the processing capacity to actually deal with it. Thus I’m just sitting on that while doing nothing. Yes, you starve your competitors but you’re doing a massive disservice. and the market your clients might hate you for that because if you’re if you’re just playing games and creating just for the sake of creating problems for your competitors it’s not good. So you if you if you corner the supplies you have to make sure that you can act upon this supply or at least resell the supply to those competitor at a higher price.
I mean again there are plenty of implication but again that illustrate the fact that you need to look at the future with this those economic lenses. It is really a sort of game of behavior of expectation that is being played. It is not like a static a static rule to apply compliant I am above service level target or non-compliant red I am below.
Conor Doherty: Yeah, I think that’s certainly in this context it’s a very good way to demonstrate the criticism of the purely service level perspective which is traditionally it would be very much just I’m going to use the term inward-looking like what do I need in order to satisfy my current demand what do I think whereas the example we just discussed shows how it can be more outward-looking which is well what do what do my what do I want my decisions or how do I want my decisions to impact other players in this space and that completely independent really of me currently servicing my current
Joannes Vermorel: I mean in this book it’s what I describe you know the teleological view versus the rugged view the rugged view the teleological I say we know the future there is one forecast one plan and then we have compliance or non-compliance the rugged view is say we are creating opportunities we are assessing opportunities and there is risk and there is reward And it’s really a fine grain assessment of the of the economic values of those opportunities of those options and there is no fixed plan if situation it’s in a way the rug perspective is a lot more fluid in the sense of you let yourself be guided by the opportunities as they present themselves. It doesn’t mean that you’re not forward looking. you are very much forward looking but you don’t rigidify your the way you look at the future and saying I know the future now all I need is a compliant execution that matches the future I’ve stated the rugged view is more like I am constantly updating this probabilistic fuzzy view of the future and I am reassessing the value of my resource allocation continuously
Conor Doherty: Well closing thought before we move to the last vignette for today but it’s on that and it’s just to take the teleological rugged view and make it more operational because what I was going to ask to close was for Elena’s situation there’s manual sort of scenario basically if this then but it’s not very high-dimensional so in this closing question and this clinical supply context what would a better planning layer produce would just be scenarios decisions risk budgets like what exactly What exactly would you recommend to Elena and her team here?
Joannes Vermorel: You want something that is conceptually dealing with all possible futures. Okay, all possible. And I know it sounds strange because you say oh all it’s infinite and computers can only be finite in I say yes but morally yes you will only consider a finite number of future just because your computers cannot your computer cannot do anything but finite. But if we say finite meaning millions, yeah, in practice it feels like really something that is surveying everything.
So what I’m saying is that as opposed to handcrafted scenarios where you’re considering like five scenarios you see say no you need to have like conceptually an approach that makes the most of the computers that we have that are immensely powerful so that the baseline is we look at millions of potential outcomes weighted with their probabilities and you want to compose all the sources of uncertainties that you have demand future demand will vary future lead times, future prices of those commodities that can unfold. also the stated consumption because for example if you have to supply your own processes those people might even revise their own plans. Though you need to take into account that even what you project as relatively safeish and known which is your own plans can still be revised and thus there is a degree of uncertainty associated with associated with that even if it’s supposedly a plan that comes from your own company.
Conor Doherty: All right we already been going for an hour and I do want to I wanted to do at least three. So I’m going to push on. You mentioned commodities and so we’ll push on to this last one. So again this is commodity processing.
So the context here very large commodity processor hundreds to a few thousand raw intermediate and finished product references that’s important annual turnover well over a billion planning depends and this is where it gets very granular planning depends on raw material quality variable yields processing capacity and market price exposure. Now the situation Steven again not real works in commodity processing. Their ERP records procurement, production, inventory, sales, batches, transformations, invoices and shipments. Market view.
Yep. Very robust data. Market views, yield expectations, quality constraints and processing decisions are handled through very impressive specialist analysis, but also spreadsheets and manual judgment. Now, as you obviously know already, the raw material is variable.
Yield, quality, demand. Market prices can all change. Processing consumes capacity and cash, but may preserve future options related to quality, service, blending or commercial timing. Now, I should it’s not I didn’t write it down, but I should also mention all of these decisions are taken very far in advance.
Like the time horizon between, let’s say, making a purchase decision and even just receiving goods could be 12 plus months. So the planning question is whether today’s action and that could be let’s say process wait buy hold or reduce exposure excuse me does that improve the company’s expected economic outcome under uncertainty now in commodity processing and there’s also a dimension of perishability here what makes inventory planning different from simply tracking quantities and batches like let’s say in clinical supply that’s a lot of context I realize yes I
Joannes Vermorel: And here what you have is that you have to think that first you have the variability of future prices that needs to be factored in. again if you were able to accurately forecast the future in this area you would not do supply chain you would just become rich in the stock market. So you cannot really expect to accurately forecast the future. what you can do is assess the volatility. Yes. And that’s because some stuff are much more volatile than others. so that’s one thing is you need to assess this volatility for your for your supplies.
Then you need to think that if you’re doing those commodity processing high volume etc. Usually you can also change the composition of your batches. Correct. for example, if it’s food stuff, you can change a little bit the recipe within tolerances to still achieve I would say a taste that is within the bracket acceptable taste. and obviously the tolerances that you’re willing that you have depends on the branding and everything. But quality is a consideration.
So here in this sort of situation is quality can be something that is extremely I would say diffused because you see you can go higher on the quality of certain ingredients lower on the quality of other ingredients. It’s a mix and then you have to make sure that the taste overall stay within the boundaries that you consider acceptable. So where I say where you have this so here I would I would really differentiate the planning needs to doesn’t need to overcommit itself to decision until they are really needed if you purchase something one year in advance because it’s just one year lead time so be it for example for future harvest we want to get okay fine but then the for example the daily production batch that is going to consume your raw materials. This has no reason to be made one year in advance.
And even if you said, oh, one year in advance, oh, we should absolutely we decided to b to buy all those products this supply to do this. But meanwhile, markets move, shift and et. And now there is an opportunity to do something quite different with the exact same supply. So be it.
You know, you don’t have to rigidify the production. So what I would say is here when you have this situation is make sure that you’re always taking you’re not overcommitting your decisions to beyond what needs to be decided today. you see if today you need to decide for to buy one year in advance so be it. But you don’t need to commit yourself to production capacity to this production schedule this plenty of other things. Other things can wait and maybe for example you can even postpone that maybe this things that you’ve purchased will be shipped by boat or airplane depending on the conditions.
You know it’s and obviously if it’s food probably not airplanes but maybe trucks it can happens rather than boats.
Conor Doherty: Well in this in Steven’s context I know that within the ERP there are modules obviously as is of when we say ERP just parenthesis and or the planning modules that might be added close parenthesis. So with their ERP there is a planning module. The thing is the foundation of that as you can imagine is fixed values time series forecasting that is that is the basis of a lot of the decision-making now why does that perspective struggle I don’t want to say fail maybe I know I know that’s my take for I know it is I can I can hear it but for the sake of discussion why does that perspective struggle not in general in this very specific context where you’re dealing with you’re not just dealing with variability unlike let’s say fashion you’re dealing with enormous variability and yields like it’s not the clothes are not influenced by the yield so to speak unless you get really granular but what I’m saying is there’s so much variability there’s so much lead time there’s so much volatility why does the single value time series perspective struggle in this context
Joannes Vermorel: Just because it cannot represent this variability you see it’s again it’s a You have planning a planning module. Yes. You have something that vary, you know, with a stochastic aspect to it. So it’s really random and then you take something that is designed to eliminate randomness.
You know, again, typically the your accountant, you just ask an accountant for example, if you tell an accountant, they say, “Dear accountant, would you be okay if we have like, you know, a little bit like quantum style transactions?” you know the transaction it may have happened or it may not have you know at the same time you know that’s the Schrödinger transaction we are just going to say there is 50% chance that this transaction did happen and was indeed paid and 50% chance that it did not your accountant would be oh hell certainly not this is you eretic you don’t do that Schrödinger’s cat yes there you go exactly was nice French so the reality is that the ERP is absolutely not this at its it’s literally it’s in this DNA. You do not want to have anything that is like yeah v viable probabilistic all those things are heretical to the design of an ERP and it is also heretical to the mindset of an ERP. And that’s why I say the problem with planning modules designed by companies that are fundamentally ERP companies is that it is so adverse to their to their I would say core values to the way they approach engineering that it cannot be anything but crappy you see it’s just it’s just the way it is and that’s you see people underestimate the fact that’s something that I’ve seen very frequently that a software company no matter how large is still extremely driven by its core principles and it’s extremely difficult for a vendor even an extraordinary vendor to be not anything but abysmal when you go out of your I would say comfort zone just to give you an example let’s take for example Alphabet now Google now Alphabet it’s a fantastic company, one of the largest success, you know, of all time in term of financial returns. No question.
There is no question that Google what they achieved, they revolutionized the web search 25 years ago. Now all the attitude and all the success was on geared on the idea that they will automate anything and that Google can have like a billion customers without any customer support. Okay, that’s fine. And yet now Google is struggling immensely and still decades later they now have like an enormous array of B2B services and yet from what I hear from my circles is that whenever it comes to B2B service they are they are lost their quality of service in the sense of I am going to get you a human representative to hold your hand and be nice to you as it is usually expected when you do B2B.
You know, B2C you can say just deal with it. It’s self-service and if you’re not happy, just walk away. And but our stuff is cheap. That’s like the B2C attitude.
B2B is when your clients are spending, you know, hundreds of thousands of dollars or euros and contractual SLAs, there is contracts and whatever is you hold the hand of your of your customer. You have a nice sales representative. they make sure that everything is right et and here my take is that all in all from what I see Google is still extremely crappy on this front and that’s a problem it’s not their DNA it there they are very strong on other things but on this sort of things that’s not their DNA they are very crappy and it’s even now I would say a decade and a half of efforts of Google being like a public company having all the resources has solve that and I don’t think they will ever solve and same thing SAP is the largest software I would say probably like enterprise independent enterprise software vendor around and yet I’ve never seen SAP manage to they have pushed a few things that were for SMB segment but it was all in all it was mostly a failure and they are very good at holding the hand of their large enterprise customers But SMBs are not their sweet spot and they can’t manage to really have anything that emerge on this segment. They tried, they invested a lot and it has been constantly trying. Again, that’s the problem of you have a DNA and that prevents you from being good out of your comfort zone.
And if you are able to do an ERP, then usually by construction, you’re not going to be good at something that is dealing with those fuzzy situation where you need to look at all the possible futures and things that are there or not there with probability.
Conor Doherty: Well, that’s the thing because again typically when we say like there’s one value in the ERP, we’re generally talking about average. So we’re talking about like average yields, etc. Yes. Okay.
But the question then is and spell it out for people. What is fundament from a risk perspective? We were talking about volatility. We’re talking about value at risk var.
What exactly is the risk profile with basing enormously expensive decisions with very long lead times on average scenarios? Be it be it with a pen and paper, be it with a spreadsheet, be with an ERP, be it whatever planning based on one outcome, one potential scenario.
Joannes Vermorel: But again the point is that averages are just a mathematical construct. It is it is lunacy. You see it’s it do not reflect anything real. I mean ju just think just that’s the old joke you know how many children will you have two 2.1 it’s you it’s lunacy it’s a if you if you think of is this girl pregnant I would say 0.3 the average is 0.3 it’s a binary thing it’s either there or not there and you need you see the averages make no sense and planning based on the averages you.
Yes. But also most of the decisions make no sense on based on average. You see an average is nice because you’re collapsing multiple futures into one number. Yeah.
It’s cool. It’s a construct but it is not it is not appropriate and mostly if you were to base I mean just think of your of your daily life and mo if you were to base your decisions on averages you would get insane decisions and it’s it makes it makes no sense. Seriously, just think of it. You have a garden and on average any given day you’re not using the lawnmower.
So should you conclude that the amount of lawnmower that you need to have in your shed is zero? No. Because when you need it, you need it.
Conor Doherty: So you see I think it’s I think I have a again a very recent I would say continentwide example. So, if you look at the heat, we will come back to commodities, but just if you look at the heat wave last month, and again, we’re recording this in the third heat wave, and we’re not even in the middle of July yet, but if you look at the like what was the average temperature for June? So, if you were planning a holiday, what was the average temperature for June or in general? What is the average temperature?
So, it’s the end of May. Oh, I’m going to go to Paris in June. What’s the average temperature? Let me be planned.
Do I want to go to Paris next month? Oh, the average temperature is 22. I’m picking up 22 23 Celsius, whatever. Then you arrive and you have a in the month alone, I think somewhere in the order of magnitude of like 18 to 20 of the days of June were astronomically hot.
There were consecutive back-to-back record-breaking days of sweltering heat. The thermal island effect in Paris is lethal. So, you planned based on the average. You show up in Paris.
You’ve committed a whole bunch of resources to non-refundable hotels and it’s a horrible time to come to you brought kids. It’s a horribly dangerous time to bring kids to Paris. Maybe that’s an example of like you can base you can plan based on the averages, but no one would do that. Like very few people would do that personally.
Joannes Vermorel: Yes. They wouldn’t say that, but in a company they will. in everyday life. That’s the interesting thing about planning is that when people think for their daily life, they make this intuitively decision very intuitively and they don’t even I would say explicit the probabilities. They would just say this can happen.
Conor Doherty: Exactly. Some of the time I need to take that into account. You know, you go for ski and you say, “Okay, I have warm clothes, but there is a chance that it’s like extra cold like blizzard cold. So, I will take I will take a little bit that I will most likely not use, but if you see and just in case just in case.” And this is the sort of you see probabilistic mindset where you are not collapsing the future into one number.
Joannes Vermorel: And the crazy thing about the this modern mainstream practice of supply chain is that it has established as you know as something that is supposedly acceptable to collapse the future into like one average and then have the entire applicative landscape so ERP but all the things that are all the business systems that gravitates around ERP that adopt this completely nonsensical take on the future and just and just operate everything from there. It makes zero sense. And for commodities just like the rest, it doesn’t make sense.
Conor Doherty: Well, of course, again, I’m going to pull random numbers just to illustrate a point. But again, if you plan based on average yields and average it turns out that the yield is 20% lower than you than you considered and you’ve planned based on the average, that’s going to cost you a lot of money. And again, what contingencies were in place? Some people would say, well, you know, we buy a little bit extra.
That’s our safety stock that we bolt on to do to handle this. Which of course then kind of surfaces the tension that you described a moment ago that intuitively in their private life, people will think probabilistically. Once you take those same people and put them in the same office, it will be well, we’re going to we’re going to collapse all of that uncertainty into one value. We might be wrong, so we’ll just add 15% on top of that.
And that’s us. for covering ourselves which is a very to be charitable it’s a low-resolution approximation of an attempt to deal with uncertainty. It’s like we might be wrong. It says we might be wrong, but you know, plus 15%, we’ll just go with that rather than analyzing all the a million or 500,000 scenarios and trying to assign economic impacts to each possible one which would be the next evolutionary step certainly in a commodities context where you are dealing with I think Steven is the name dealing with million-dollar these are million-dollar decisions taken 12 18 months in advance. Yes.
So I get I realize we’ve been going for a while. So just what in Steven’s case? So we’re in commodities. They’re making a lot of planning based on spreadsheets.
A lot of expertise is being applied to manual overrides as well as is always the case in this context. What is the advice?
Joannes Vermorel: The advice is you need to have a numerical recipe that runs outside and where you have no manual overrides. Again manual overrides are just a symptom that of improper design. It’s really not. If people are again we are back to another topic which is do you treat your employees as consumable or are you treating your supply chain practice as something that is capitalistic and accretive?
Whenever I see I hear people doing overrides, doing daily tweaking of numbers and whatnot, this is not capitalistic. This is not accretive. This is just treating people like consumable. And this is this is essentially a waste of resource.
And you need to get out of this paradigm. It since I would say since the late ’90s, we have computers that are powerful enough to do that. You see, it’s not it’s not even new. It’s it was something that was technically that is technically feasible that has been technically feasible for at least three decades and now it is extra easy due to the fact that we have much better it became possible three decades ago but it’s it is now relatively straightforward unless you’re dealing with an incredibly incompetent vendor.
So and unfortunately there is like quite a few of those but if we put the incompetent enterprise software vendor aside problem it’s relatively straightforward. This is the game should be how good can you make the can you make the numerical recipe and then that’s like do becoming like a very good chess player as opposed to an average chess player. But if you have you see the incompetent vendor is just like a pigeon you know crapping over the board whatever and not even playing chess and then you should your bar should be you want someone who is at least able to play the game and ideally you want to want to have a vendor that is able to play the game very well. That’s what we should strive for.
But you see that’s a little bit the spectrum. And when it comes to planning, you have to think of vision on a chessboard. You know, that’s the level of nonsense that you’re getting.
Conor Doherty: I want to close with just a very constructive thought and then you can give me your closing thoughts across the examples that we’ve given. I want to make sure that I’m representing our position clearly, which is no one is arguing that the ERP your ERP is bad. You’ve been very clear about that. under what you would call resource management. That’s what you would call in your system of record whatever where you wherever you store your transactions as long as it doesn’t represent more than probably 10% nowadays of your IT budget that’s we’ll come back to that conversation we’ve had that conversation as long as it as long as the budget for it is like extremely narrow and tight it’s fine it’s fine so with all those caveats yes no one’s criticizing the concept of your ERP whatever that might be the statement the thesis here is to make better decisions look forward to make better decisions to allocate resources more effectively and with a greater chance of financial reward you have to look beyond that
Joannes Vermorel: Yes I mean or rather you need to look at things that by design will never fit into your you see that’s just as it is it just it will never manage to fit because as soon as you want to look at probabilities and this sort of thing. It’s literally has nowhere you cannot even fit those things into your and so you’re stuck. You see it’s just and again if you cannot even look at those things then you can’t even start doing an economic calculation and without economic calculation you’re just blind. We are back to the fact that you are just taking a decision without any economical assessment and then you’re calling it a day and the reality is that if you if you reassess afterward this decision with a rate of return you would realize you’re leaving a lot of money on the table.
You know that’s the thing is that supply chain will work will flow but much less efficiently if you if you just don’t do any kind of economic optimization. That’s it. It will just you will just lose probably several percent absolute of profits yearly just because of that.
Conor Doherty: Yeah. Again, it’s a relative argument that’s being made here. Money on the table which can grow. That’s the whole point.
Yeah.
Joannes Vermorel: And again, if you have a fantastic brand and you’re selling stuff with 80% gross margin and you’re growing and nobody cares about your supply chain and you can waste a lot of money there and it will still not make any impact, that’s fine.
Conor Doherty: That’s not the majority.
Joannes Vermorel: Yeah, exactly. It depends. It depends. So you see if your company’s fantastically successful and you can be you can have like order overheads that you that it doesn’t even register on your supply chain.
It’s fine. There is no point in optimiz maybe no points in optimizing supply chain. But if you’re not in a Louis Vuitton situation then probably you should you should pay a little bit of attention to your supply chain.
Conor Doherty: All right. Well, Joannes, I’m out of questions. We’ve been going for almost 80 minutes. So, I’ll take no more of your time.
Thank you as always for joining me. And to everyone else, thank you for watching. As I always say, if you want to continue the conversation, feel free to reach out to Joannes and me on LinkedIn connect. We’re always happy to talk or you can send us an email at contact@lokad.com.
And with that, we’ll see you next time. And yeah, get back to work.
