Atilla the Hun, Supply Chain

From the webinar “How End-to-End Optimization Reveals Supply Chain Drivers” Q&A with:

  • Dan Gilmore, Supply Chain Digest
  • David Simchi-Levi, Chairman of OPS Rules, MIT Professor of Engineering Systems
  • Laszlo Molnar, Sr. Director, Supply Chain PepsiCo Worldwide Flavours

1.    Laszlo, what was the catalyst for this inventory project?

Laszlo: In our case, the catalyst was an organizational change in operations that we introduced a year and a half ago which demolished the barriers between the plants and the distribution network. Now that the plants belonged to the same supply chain, that change in itself provided the opportunity to look at our supply chain from an end-to-end perspective and determine how we can optimize it and still produce the same service level, but with less inventory.

2.    David, as many projects as you’ve done and as many companies you’ve worked with, there are always some surprising insights, were there any surprises on this project?

David: As you pointed out, I have done many inventory optimization projects, and in every one of them there is a surprise. In this case, the surprise was what we were able to identify as the key driver of inventory and service level in the supply chain. Importantly, it was all about supplier performance, but not supplier lead-time. In one of my slides, I showed that the impact of reducing supplier lead time on inventory level in this supply chain is relatively small. However, variability in lead time and supplier delivery performance have had a huge impact on inventory. Thus, managing supplier performance is critical in this supply chain to improving performance.

3. How about you Laszlo? Any surprising views from the project?

Laszlo: Yes, this was the first time that we were exposed to multi-echelon optimization methodology. Before, we used a single echelon which can be compared to how Louis the XIV looked at of the universe. He only spoke to the people of the court, which is like looking one step to the left at the vendors and one step to the right at the customers and optimizing that way. You can go this way through the entire supply chain, but still the results are optimized for a single echelon.  Versus what we did here which is multi-echelon.  I compare this approach to Attila the Hun’s methodology. He was a great warrior and overall very devastating to the Roman Empire. He interacted with his troops around the campfire and anyone could come there and provide their viewpoint. The analogy is that multi echelon optimization can provide new numbers and a different way of optimizing the supply chain. For me, it was a learning experience both personally and professionally and I am glad that I was exposed to it.

4. Why can’t regular companies move that curve? Is it some special technology? What is the barrier for companies that keeps them from moving the curve as you illustrated in those first couple of slides?

David: I think part of the problem is that most companies either do not focus on inventory or if they do, it is mostly local optimization. That is, in local optimization, every facility in the supply chain tries to optimize its own decisions with very little regard to the impact of its decision on other parties within the same supply chain.  In this project, we replaced the local optimization strategy with a global optimization strategy where we focused on the supply chain end to end. This allowed us to understand what drives inventory and service level in this specific supply chain and in this case, positioning inventory correctly was critical. In fact, in this supply chain, positioning inventory correctly allows Pepsi to cut inventory by between 24-29 percent depending on service level. The analysis also suggests that if you manage suppliers better, you can achieve an even better performance that will reduce inventory even further.

5. How far off does the model have to be before you have the need to redo it? And how do you know where to make the fixes?

David: These are important questions and there is no simple answer. We have developed a methodology that allows you to make sure that the models you are using correctly represent your supply chain. There are number of steps. I described here the two-phase validation process. In phase one, you compare the fill rate in the model to the historical fill rate and in phase two, we provide the model with inventory and fill rate information and we treat them as constraints and try to identify locations where there is inconsistency between the two. But, this two-phase approach is only part of the validation process.

If the insight generated by the model is not consistent with your understanding of the supply chain, this suggests that either there is a problem with the model or the model generated a new insight. This is the power of the model and the new insight. Here we emphasize that the process of validating the model includes the two steps that we described during the webinar as well as generating different scenarios and looking at what these results are telling you. If they are not consistent, we need to understand why this is happening and what mistakes were made that generated these types of results.

6. Do you always do that validation the same way? Are there other ways?

David: The validation includes the 2-phase process plus generating the different scenarios and the art here is to know what type of scenarios to generate to test the models. That is, where it is company and supply chain specific.  Once you identify the appropriate scenarios, you have the ability to identify modeling and data problems.

7. Did you spend time educating your inventory and other team members on what this process was going to be like and how this was a different approach from what they may have thought about inventory optimization in the past?

Laszlo: Yes, it was part of our change management process. We communicated the organizational and operations model and it was well understood. Therefore, the need for end-to-end optimization came up as a natural consequence. And then, we kept the communication throughout the organization as it is very important that everyone is a part of the change and there is ownership of the organization after the project.

8. There is a little bit of black box aspect. What’s going on behind the scenes isn’t clear to people. I would assume the education process has got to be key in most projects?

David: Absolutely right, in fact even after PepsiCo convinced the team that they needed to move from local to global optimization. We ran three different workshops: one at the beginning of the project to show case studies and the types of trade-offs, one after creating the model where we show what we have built to see if the model correctly represents PepsiCo’s supply chain, and then the last one at the end of the project to illustrate the results and to provide guidelines on how the company can use the model and the technology on a regular basis. This will allow them to run the modeling independently and they don’t need OPS Rules to come in every six months to re-optimize their supply chain.

9. Once this model is built and you’ve executed the strategies, what kind of tuning has to be done and how does that work?

David: Many things change, new products are introduced, some products are phased out, there can be seasonality, new trends and forecast changes as well. What we found is that many companies go through inventory analysis of the type we did for Pepsi every quarter or as a part of their Sales and Operations (S&OP) process. In fact, some companies call it SIOP, where “I” refers to Inventory, to manage supply effectively and use global optimization and multi-echelon inventory technology to match supply with demand.

To watch the full version of the webinar, click below. 

Written by Stephanie Stein, Marketing Coordinator at OPS Rules Management Consultants