The survivor bias applied to DDMRP projects.
First, let’s recall what survivor bias is:
Survivor bias is a selection bias. It consists of overestimating the effectiveness of a method by focusing on those who used the method successfully…. (and ignoring those who used the same method and failed).
Then, we can ask ourselves whether this is relevant and sufficient to form an opinion on a particular approach?
“During a visit to a sanctuary dedicated to the gods, Diagoras, a Greek philosopher famous for his atheism, was presented with numerous tablets depicting shipwrecked people saved from the waves by their prayers. The priest then asked, “You, Diagoras, who are convinced that the gods do not interfere in any way with the affairs of Men, how then do you explain that they have saved so many?” Diagoras replied, “It’s simple. Those who prayed and drowned are not portrayed anywhere.”
(The Black Swan: the impact of the highly improbable, Nassim Nicholas Taleb).
What does this excerpt teach us? You have to learn to see what cannot be seen.
In other words, to learn, it is not enough to grasp the successes and believe in all the marketing promises proposed by some or others. Learning a new method is also built on failures and difficulties.
It’s true that right away, it’s much less sellable and much more difficult to romanticize or promote. But this is the most beautiful school, the school of learning.
When we read “Divide inventory by 3 ‘thanks to DDMRP’ in 4 months”. Should we read a promise? A slogan?
Obviously, for those who know the approach, we know how to qualify. But the target of this slogan is not the experts. The target population is made up of curious people who are interested in or wonder about the method. When we look at the enthusiasm of people for the approach for several years now, we see that it works. The slogan attracts and stirs.
Based on our projects and failed experiences, we wish to bring an objective, realistic and pragmatic view on DDMRP project deployments.
Let’s try to answer a first question: what does “a successful DDMRP deployment” mean?
Some of our customers used to have a MOQ equivalent of 1,5 years of production. By setting up the model with launches covering 60 days of requests in a DDMRP tool, we quickly get relevant recommendations that lead to impressive results. However, have we succeeded in deploying the DDMRP method? What is the share of success linked to the DDMRP method? If we had simply questioned the historical settings and strategic choices; strategic decoupling point, batch size, choice of suppliers…, would we have had such different results?
For the majority of the examples we have seen, if we were a little provocative… (obviously, this remains an assumption to be confirmed!)… To announce that you “have successfully deployed DDMRP in a few weeks” is like saying that you “have successfully completed a LEAN transformation” because you have just finished your 5S project. You would then retort that we are acting in bad faith… maybe!
Beyond the deployment timeframe, which depends on the company’s context (Internal: maturity of MRP2 models, health/reliability of master data, BOM, article base, range…. / External: market, supplier, integration, positioning), it is important to take into account the notions of levels and maturity of deployment.
The deployment of a DDMRP approach and tool takes teams to a new level. The approach is easy to integrate and ergonomic. Planners and users are often won over by the simplicity of interpreting priorities. We regularly observe a faster appropriation of the tools. This is the visible side of the iceberg.
However, the challenges of change management and management development remain very important. They must be understood and anticipated in order to ‘succeed*‘ in the deployment.
*to succeed would then be to set up a Demand-Driven system, guided by relevant indicators of flow and which are kept up to date over time.
Two important issues to be well integrated:
1. The evolution of the planners’ profession
Contrary to what we may have read or heard, planners are not turned into button pushers; “It’s in the green, I wait. It’s in the yellow, I order”. The job remains just as sophisticated and complex.
One of the challenges of DDMRP is to enable mass validation of recommendations so that managers can focus on:
- Find alternative solutions to ‘normal’ routes for emergency items (red).
- Handle execution alerts. And therefore follow the flow on the supplier and internal sides and propose solutions or prioritizations if needed. For some of them, who are real experts in their fields to anticipate, interpret and manage the nervousness of the system before deployment, this can sometimes be experienced as a ‘loss of expertise’. It is then necessary to explain and remind them of the core of their target activities, and their real added value.
- Support and improve the system.This topic is often underestimated or misunderstood by planners. Planners should be expected to take a critical look at the fine-tuning of items. They should report their concerns on a daily basis: new items not configured, buffers that seem to be oversized, batch sizes that are not up to date, items that goes from green to red too quickly… So many points that need to be dealt with via the control rituals. And that intelligent automatons can help to deal with.
It is important to understand that the planners are the experts who will be able to assess the consistency of the settings.
Thus, they must be actors by validating the recommendations, but above all co-constructors of the model by having a critical and constructive eye on a daily basis. And there is no secret! It will be necessary to train them, support them, and explain what is expected of them to give meaning to all these actions.
If this is not done properly, then one falls into the same pangs as with any planning system; loss of confidence in the recommendations and tinkering with the recommendation system in Excel to compensate for this.
2. The piloting rituals and alignment of the operating model :
The idea is not to detail all the rituals to set up. We have listed them and described them in dedicated sheets, but to be adapted to the specific contexts of the companies. We wish to focus on the needs of:
- Developing management indicators
- The behavior of the teams changes depending on the indicators.
- Okay, you measure inventory value. And how about looking on the cover?
- Do you measure the number of breaks with need? The number of complete kits?
- Do you focus on the OTD? On the respect of the schedule? On the evolution of average lead times?
The choice of indicators defines the rules of the game with your teams. It sends a strong message of commitment. So you do not only need to integrate these Demand-Driven metrics into your existing reports, but you also need to remove any metrics that contradict them.
- Measure the number of buffers in the red. And at the same time keep an indicator measuring the OA or OF ‘without needs’ on items we want to stock.
- Ask to produce according to the DD recommendations. But maintain a productivity bonus pushing the groups into production.
- Structure the process around rituals. They are the backbone of the system in the long term. They should enable managers to:
- Collect feedback and comments from planners,
- Feedback on the evolution of the operating model,
- Follow up on the necessary actions to improve the recommendations and maintain the system coherent and relevant over time.
It is therefore a question of giving your managers the keys to reading and acting; What to look at, how to interpret, and what actions to take? So many reflexes to change. Let’s avoid putting them in a position of failure. They must be one step ahead of their teams and be comfortable with the appropriation and animation of these daily and weekly rituals.
These two business issues are key to the robustness of the system.
Companies that restart a project 1 year after the pilot, that do not manage to go beyond the first scope in their deployment, or that do not obtain the expected results have, a priori, not seriously taken into account the two points above.
Just as we denounce DDMRP projects based on Excel files, which all too rarely lead to successful projects, we would like to emphasize that the real success of DDMRP projects lies more in the evolution of mentalities, daily reflexes of teams, and paradigm changes at the highest level of the company than in the installation of a software solution.
« The one more thing » – What was also missed by those who failed – Adopting the helicopter strategy.
This consists of providing a high level of visibility and legitimacy to the teams while dealing with problems in the field at the same time. You have to know how to navigate between the two at all times. This is to put day-to-day actions into perspective with a global strategy and orientation that legitimizes the actions of our field teams.
This is why we should not deploy locally, as a pilot, but take these projects as a whole, starting by defining the company’s strategy; MAKE, BUY, potential order winner, positioning on its market, and expected timeframe.
- Is it a business issue in my market if I offer these products tomorrow in 4 weeks instead of 6?
- Should I outsource these activities?
- Would paying more for my subcontracting, but reducing its timeframe, be more relevant for my reactivity and for my business?
On the basis of these MACRO ‘global supply chain design’ workshops validated by the management committee, you will be able to go down one level in order to deploy the approach by unit.
From then on, you can launch your first perimeter. It will be used to build your team and your deployment kit; communication document, training support, list of operating rules, deployment checklist. It will also allow you to build feedback to benefit other units/sites/BUs.
For each of these deployments, you will have to deal with very specific problems, linked to the professions and to the habits that need to evolve… It is important to deal with them, to respond to them, and to make reflexes evolve. But it’s much easier to do this with the legitimacy of a global project and macro operating rules. This will be the direction of history. It is necessary to give legitimacy to the action and serenity in the realization. How many times have we faced suppliers afraid to place orders due to historical inventory and ongoing pressure…
What to keep in mind?
Our experience allows us to believe in the success of DDMRP deployments when companies have understood that “success” means betting and investing (money, but above all time); on its teams, on training, and on the effort to accompany the changes in mindset that this induces.
Our past and current experiences with Demand-Driven approaches suggest that without a software solution, deployment is not possible. And without a profound change in the company, deployment is not sustainable.
The objective of this article is to enlighten companies wishing to start a DDMRP approach. They should not tackle a pilot only but prepare a journey, a transformation that will require the full attention and involvement of the management. It will obviously be more difficult and/or longer than promised, but the return on investment will be worth it.
Are you looking for an ROI on your DDMRP projects? Ask yourself these questions:
- Are my teams doing better now? Getting better results than if they had operated as before? Are they spending more time dealing with problems than recalculating the CBN in Excel?
- Have they acquired habits and reflexes that are positive for my flows and my WCR?
- Are my production and planning departments functioning well by being aligned with customer needs rather than siloed?
- Do my teams not blindly follow the settings made during the deployment, but have developed their critical thinking?
- Are they able to develop the DDS&OP model independently?
If the answers are positive, then the numerical results on OTD and WCR will follow. The gains will be sustainable and you will have truly worked on the resilience of your supply chain. You will have successfully completed or started your Demand Driven deployment.
Obviously, this can be done without deploying Demand Driven methods. But let’s recognize that in addition to the relevant innovations offered by the method and respecting the recommended approach, it sets the right tone to review a lot of parameters and habits that can only help to orient your supply chain models in the right direction.
Thomas JANSSEN, Thomas COUTERON, Damien BROCHARD