- This article contains comments from the articles on DDMRP.
These comments are in response to the articles on DDMRP.
Article Questions & Comments
I had the following questions asked of me directly, so here are the answers.
This is a response to a question asked by Sanjeev Gupta on LinkedIn. I could not fit the answer into the LinkedIn comment box without breaking in into too many comments. I think Sanjeev’s questions are essential, so I am profiling it here.
“Potentially useful back and forth between Shaun and Carol/Chad!
I don’t think it matters whether DDMRP repackaging of JIT and Lean or not (sometimes truth is eternal, sometimes old truths are new fallacies), whether Orlicky would have agreed with DDMRP or not (he wasn’t God and even God can be wrong), and whether MRP is Pull or Push (that is just semantics).
My limited experience is that many supply chains struggle with high inventories and unavailability as well as firefighting, and I believe all that matters is, “Does DDMRP result in higher availability with lower inventories and less management effort?”
Carol/Chad: Is it possible to provide a mathematical proof of this, something that is not just qualitative cause-and-effect? Conversely, Shaun, can you provide a mathematical proof that DDMRP does not yield better results than old MRP?” – Sanjeev Gupta
Sanjeev is asking several questions. For clarity, I have created a topic heading for each answer topic. These topics are as follows:
- Does a Mathematical Proof Answer the Question of DDMRP?
- Admitting When Things Do Not Work
- A Mathematical Test for DDMRP Versus MRP?
- The Problem with Investment Prioritization
Does a Mathematical Proof Answer the Question of DDMRP?
You are proposing that there is a single way to determine whether DDMRP is an improvement over MRP. And that the way is a mathematical proof. I can provide several reasons why that is not true. (I am not trying to pivot away from your primary question as I have researched this topic, and I will refer you to a study that does what you describe.)
- Example 1: It is often the case that a new and complex forecasting method/mathematics is introduced, that it can outperform less complex forecasting methods in a laboratory environment. However, in actual practice, the more complicated forecasting method, which is “mathematically superior” will often lose against more simple straightforward techniques. Forecast papers often test small data sets and are willing to put a lot of effort into improving forecast accuracy. But real supply chain departments do not work like that. This is because, in real-world environments, there is a minimal amount of effort that can go into the maintenance of the forecasting system. I covered this topic in detail in the following article Complex Versus Simple Forecasting Methods. This is a concept promoted by strong evidence by Armstrong in his excellent book Principles of Forecasting 2001.
- Example 2: Inventory Optimization and Multi-echelon (MEIO) is more mathematically sophisticated than MRP. There is nothing in MRP beyond arithmetic (the procedure itself, not the parameters). MEIO can build inventory to match a service level (the inventory optimization mathematics), and it is aware of the stocking position of locations around it (the multi-echelon mathematics). Therefore MEIO is better right? Well, it depends. MEIO is too complicated for the level of investment that most companies are prepared to make into their supply planning systems.
Is the Criticism of DDMRP Being Repackaged JIT and Lean Relevant to its Probably Improvement Over MRP?
JIT and Lean proponents have a decades-long established history at this point of exaggerated claims. Of using hyperbole such as “forecasts are wrong” and Toyota “did this or that” and they have rebranded themselves once before, that is when they moved from JIT to a new name (with a new bunch of books called Lean). Do these proponents need me to trot out simple-minded books like “Zero Inventory” or the books that misrepresent what Toyota did, and to go over the terrible exaggerations and mistakes made by following JIT and Lean consultants? Is this some mystery at this point? It is indisputable that JIT/Lean has been littered with exaggerations as I cover in How the Toyota Production System Was Misrepresented to US Audiences. Deloitte JIT/Lean consultants told one of my accounts to break up their inventory into multiple stocking locations in the factory — which surprise lead to a major increase in inventory. They told them Toyota did this and that it was a “Supermarket.”
Strangely, I would be making this argument because I am a major proponent of turning off forecasting for product/locations that are deemed as non-forecasted — as the following article Forecastable Non-Forecastable Formula.
The majority of companies I have consulted with are forecasting items and putting effort into improving the forecast for items with no discernable pattern. (that is they are assigned to a Level forecast by a best-fit algorithm). Turning off forecasting for products that cannot be forecasted efficiently is an essential strategy in both improving stocking positions, but also in allocating the scare time available in overburdened forecasting departments.
But JIT/Lean proponents move far beyond the judicious use of consumption based techniques, and into diminishing all supply planning procedures, and sell companies on the idea of moving to make to order environments that can’t I refer to this as The Make to Order Illusion.
It has been proposed by many JIT/Lean consultants to many soft target executives with predictable results. JIT/Lean proponents should be held to account for these inaccuracies. But they will have none of it, they are interested in hiding these failures, and going to make more projections. It would undoubtedly be the ultimate dream of every JIT/Lean proponent to undermine procedure based planning. And DDMRP gives them a cover to do this. So we should be suspicious of their intentions.
If people misrepresent their history of success and then rename their methods into something that is the opposite of this type of planning, then this is quite relevant to the discussion of the validity of what they propose. And this gets to the topic of honesty in recognizing the failures or shortcomings of your approach.
Admitting When Things Do Not Work
I come from a background of advanced planning. I was very excited in my mid to late 20s in working in advanced planning, and I drank some (although not all) of the Kool-Aid served by my employer, i2 Technologies. However, I observed from working on projects that the optimization projects I worked on did not match the sales that we presented to customers. The media gave i2 Technologies the typical unquestioning coverage, and for several years everyone wanted to “do optimization.” How i2 Technologies was able to “corner the market” on optimization is another story because optimization had been kicking around in academics for years, and no one “owned” anything all that proprietary when it came to optimization.
With my background, I am one of the only, and probably most well-published critics of how badly cost optimization failed to add value to companies.
While i2 Technologies eventually fell, major companies are still trying to get a flawed optimization approach to work in SAP that will never work, which you can read about in detail in the article The Problem with SNP Optimizer Flow Control.
I say this as a longtime consultant in SAP, but the SAP is ripping off accounts and lying to them about the benefits they can obtain from a system I have extensively tested, and primarily does not work. SAP charges somewhere around $450 per hour to a group of Ph.D.s in Waldorf that must be seriously laughing at these customers who can’t put two and two together that the applications need to be deactivated.
And these are big name brand companies using this solution you would recognize. If you like, you can hire SAP consultants to “improve your SAP optimization,” and you will not hear a peep from these consultants about the history of cost optimization.
Many SAP consultants have reached out to me in private, not speak, and write about these things. They have asked that I air my issues in private, that to do so in public creates “bad feelings” and reduces the optimism about cost optimization. And you should have confidence about something if a major entity proposes that it is true, and if you can bill hours for selling the illusion and hiding the truth.
So my point being, I have a track record of being honest when things that I have worked in don’t work. But I can’t see where I have witnessed JIT/Lean proponents do that. They shy away from telling the truth of what happens on projects. There are certificates to be attained, various colored belts to receive, and different trendy items to be placed on resumes, but little time to determine if things work.
So hopefully that explains why I do think the issues I have brought up are relevant to whether or not DDMRP will produce a real benefit. I have used the same techniques of skepticism to make previous predictions on SAP, and have so far batted 1000 on my SAP predictions. Of course, one’s accuracy improves if you don’t get paid by the entity for which you make projections.
Now, let us switch gears.
A Mathematical Test for DDMRP Versus MRP?
There is a single published study I was able to find, which tests DDMRP. It is titled MRP vs. Demand Driven MRP: Towards and Objective Comparison and published by the Institute of Electrical and Electronics Engineers in 2015.
I don’t want to give away everything in the study, as the researchers deserve compensation for their work. Furthermore, the conclusion is a bit of a mixed bag, so one has to read it for oneself. In conclusion, the authors stated that DDMRP performed a little better in one area and a little worse in another. And the study points out that if the demand had been made more variable, the results might have been better for DDMRP.
So the results would seem inconclusive. However, I found something that the study did not highlight. And that is that DDMRP requires a lot of investment and a lot of change regarding working in the DDMRP paradigm. New inventory buffers must be calculated and maintained; there are new seasonal factors, planning adjustment factors to be incorporated, and so on.
In reading the study, which did an excellent job of explaining what had to be done, I began to wonder where the funding for all of this effort is going to come from. The MRP systems that I have seen at companies are held together with glue and tape. Multinationals with billions in revenues tend to want to invest as little as possible in their MRP systems. The planners tend to be poorly paid. Many of the directors of the supply chain don’t even understand how MRP works (I have also seen people come over from sales to head supply chain departments — how can that end well?). Many speak in platitudes about high service levels and come up with ideas like “how about we move to daily forecasting!” At one company where there was virtually no budget for MRP improvement, I was shown the stock sales of company insiders, which ranged from $2 million to up to $36 million. So let’s face it, MRP investment is a distant second to exercising stock options.
The Problem with Investment Prioritization
I came up with an inexpensive way to optimize supply planning/MRP parameters. I am often told there is no budget, and is there something their planners can do to get the same benefit, like maybe reading some of the books or reading some articles or going to APICS. At the same time, some analytics project that often has little output ends up with plenty of funding. People have frequently told me, “it’s too complex.” Read it for yourself at 3s – Safety Stock and Parameter Setter.
Well, if parameter optimization — which does not change the MRP process in any way — is too complicated and too time-consuming — where is the funding and motivation for a great effort like DDMRP going to come from? I recently worked as a sub-contractor for a consulting company that had won a contract in my area but had no reason to have received the contract for this work. I realized they had received the work because they exaggerated and “sold” the account.
It makes me wonder if this is where we are with DDMRP? Does one have to oversell, to propose amazing outcomes to get investment for MRP improvement? Is this why money is flowing to DDMRP projects?
So to answer your question, I only know of one study on DDMRP, and please read the study for yourself, but the study appears inconclusive. Chad Smith has been very vigorously promoting the case studies that show DDMRP success. And I am not trying to be difficult or contrary, which seems to be the view of a few commenters, but the indisputable fact is that companies don’t publish their failures. If the customer provided published information was reliable, then we would have to accept that every single case study published on every vendor and every consulting website is true. Secondly, it is scarce for companies to have good metrics on their supply chain. Companies routinely overestimate their forecast accuracy.
This is because most do not know that the only relevant forecast error measurement has to be at the product location (and not at a higher level of aggregation). Research shows that they think they deliver a seven percentage point service level than they do, and there is virtually no funding for such measurement. The business usually is just trying to keep up with their regular planning work. If I could personally measure these benefits, it would mean something. But self-reported benefits from companies that aren’t very good at measuring and don’t follow a scientific approach does not mean much to me.
Good day Mr. Snapp, my name is (XYZ), I am responsible for the group planning activities of an Italian mid-sized company which manufactures door hardware (locks, cylinders) and provide for access solutions.
That is curious because I talked to a door lock company a week ago about supporting them in forecasting and supply planning. I have never run into a door lock company before in my career.
Some time ago, I read with greatest interest an article of yours about DDMRP (How to understand DDMRP as yet another repackaging of JIT) and I must say I fully agreed with your opinions, even if they seemed critical. Now, I am sorry to bother you, but as some years have gone by and the methodology spread out or it seems so, I was wondering if you have some new suggestion about it.
Answering First Part of The Question #2
I have heard less about DDMRP than I did in the past. There was a time where it seemed like a lot of software vendors were “adopting it.” That is, vendors announce many things, I don’t know how much was really adopted, and how it was just about marketing.
In my company we use a sound and well managed MPC system, customized during the years to satisfy our business needs, manufacturing processes and complexity, but as we are always curious to innovative things, I have since explored DDMRP deeply. I had attended the first course in Italy (2018), then read a couple of books and recently visited a company near ours that implemented it, but still I have many of the doubts I had before and that you expressed in your article. I am now wondering if after those years and all the “hype” around it, you still have the same doubts (as I still have) and if you can share yr opinion about the topic.
Answering Second Part of The Question #2
You may or may not know that a lot of what I do is evaluate hype by software vendors and consulting firms. I noticed about DDMRP proponents that they made extensive claims without being able to substantiate them. Secondly, DDMRP is “ephemeral.” As soon as I would debate a DDMRPer and answer a topic of DDMRP, it would change. The entire approach is to minimize forecasting, but they say you don’t understand as they still use the forecast. So they do, but not for driving the supply plan. You can see my article here on this misdirection on forecasting titled How Accurate is DDMRP’s Explanation of Forecasting?
I find the DDMRP proponents dishonest, and at this point, I am not really interested in debating DDMRP proponents as I caught them playing what amounts to word games. Their renaming of safety stock as buffer stock is an example of what I consider a word game.
I don’t know if “word game” translates in Italian or if you frequently speak English. I was in Italy for a few months, and most people I ran into who were white-collar workers spoke English, but the term is I am using is not necessarily standard English. Anyway, I think the article explains what I mean.
And they also make a lot of negative statements about MRP, but many of the claims are untrue — such as the claim that things are so rapid now that you can’t forecast. It depends. Some things are not forecastable, but others (that is, other sections of the production location database are). This is well known, and I cover it here. https://www.brightworkresearch.com/lumpy-demand-forecasting/ This is the logic for putting some product locations on a constant forecast and others on a “real forecast.” MRP requires effort, and it is not helped if the MRP system is run from an ERP system — which most are. I don’t think this is a good idea, as ERP vendors are not who you want managing supply or production planning functionality. I cover How to Repair Your MRP systems. And discuss related topics in How to Resolve MRP Problems.
If someone has a problem with MRP systems, at least take the time to properly set up the systems. That is part of what I do. And just one item, supply planning parameters not being set invariably is a major opportunity. I cover this in this article where we have a product that does this in the Brightwork Explorer.
So the steps have to be taken to tune the MRP system and get a good MRP system. And really, most companies don’t want to do this. And they don’t want to do it in the US, where IT spending is highest. To play on the unhappiness with MRP systems without observing how the system is maintained is a very inaccurate way of approaching the problem. Suppose DDMRP were lower in maintenance than MRP, then they would have a point. But it’s hard to see how DDMRP is lower in maintenance.
This is the reason why I ask you to connect via Ld, so I know it could sound strange but I can assure you my aim is just to try to get an impartial and clear opinion about the topic. Mr Snapp, I thank you in advance, no matter if you would like to answer or not, in the latter case I bag yr pardon to have bothered you.
Answering Third Part of The Question #2
Nothing about your questions sounds strange. You want to figure out what is best for your company. And you hear that DDMRP is popular or gaining in popularity. However, I see no validity in DDMRP, and what it states around “traditional planning” is not true or exaggerated. If DDMRP made sense, I would have jumped in. I have tried cost optimization, allocation, inventory optimization, I am open to new ideas. I believe in rethinking our assumptions. For example, after I spent my career using the standard forecast error measurements, I finally stopped using all of them around 4 years ago. I made my own and have not looked back. I made an effort to understand DDMRP, and I did not find it a legitimate method. You will be much better off by investing in analyzing your current system and just figuring out how it can be better maintained. There are very good options to do some custom development to reconfigure things. As I said, MRP engines in ERP systems are just not very adjustable or of good quality. But it means more alterations than replacing MRP with an entirely different method. And it does not require a bunch of new software selection or bringing in expensive consulting firms. Actually, consulting firms are a major part of the problem as they only seem to know how to repeat whatever the vendor says as if it was written in the Bible. I say this as I am coming off of an Oracle advisory project, and it was just impossible to break the client out of the “trance” of Oracle salespeople and Oracle consulting.
Would You Like to Comment and Have it Added to This Thread?
Just provide your comment in the chatbox in the lower left of this screen.