“If you get the forecast right, you will be able to reduce inventory, improve customer service, foster more collaboration and energize your S&OP process, while increasing profitability and market share.” These were the binding threads of the recently held Supply Chain Planning Summit hosted by Global Supply Chain Council (GSCC). This online summit explored the latest best practices, tactics and strategies in the field of supply chain planning, S&OP, BIP and demand forecasting within the Asia-Pacific region and with plenty of real-life examples and end-users case studies. The game changers and experts in supply chain planning, forecasting, S&OP shared their insights on what successful companies are applying today that works. Excerpts…
What kind of trends or megatrends that you've witnessed in the last three years? It would also be interesting to know from you lessons that you have learned during this time…
Loknath Rao, Supply Chain Analyst & Technocrat: Raw material shortages have prompted the supply chain planners to come up with some creative ideas on how to allocate them profitably. The language of finance has become integral to supply chain in terms of how profitable the fulfilment is. Questions such as ‘Am I better off not fulfilling this demand now?” are being raised more often now than in the past. That's one big change that I have witnessed in the recent times. Existing enterprise systems and applications are also being redesigned to address it both from planning as well as fulfilment perspective. I am personally of the opinion that this is more a fulfilment problem and firms shouldn't be changing their planning paradigms.
Let me give you an instance of an American specialty polymer company who I was consulting. They were facing issues with the upstream supply of ethylene glycol, which is an important input to manufacturing of Spandex fibres. They had to figure out at least a dozen strategies – back-order processing, capable to match and generally running the network optimizer by applying heavy penalties for non-delivery more precisely based on customer and market priorities/ profitability. This partially addressed the objective of profitable allocation of raw materials.
Vijay Kasar, Regional Manager – S&OP and Business Planning IMEA, Henkel Adhesive Technologies: Over the last few years, we have witnessed demand patterns and supply chain solutions getting reinvented. I think the volatility of the demand and the supply chains’ ability to be agile are being brought up as boardroom agenda.
Supply chain, of course, has been the frontrunner always, but in the last two years, it's become the discussion of the boardrooms for most companies.
Prashant Shivsharan, Value Realisation Program Manager, Castrol: Over the past three years, there has been a pervasive focus on resilience, with our team shifting its discussion from a just-in-time approach to a justin-case mentality. The dialogue now centers on assessing our preparedness for potential scenarios, and whether we have the fortitude to respond accordingly. The entire supply chain has evolved to accommodate this shift, and decisions are made with this new perspective in mind. Rather than solely focusing on just-in-time delivery, we now thoroughly consider just-in-case scenarios, asking ourselves if we have the necessary resilience to navigate any potential obstacles that may arise.
Komal Ashu, Regional S&OP Lead- APAC, ANZ, SSA, Avery Dennison:In the last few years, the escalating geopolitical tensions have compelled companies to look out for opportunities to localize most of their supplies and not depend on geographies that are not friendly with the local manufacturing. These factors are also driving a different kind of strategy in terms of localization of supply chain, domestic self-sufficiency and government incentives. I do see a lot more focus on de-risking the supply chain with the help of local partners that offer more flexibility and sustainability with reduced total cost to the supply chain.
Sébastien Aubrey, Regional Vice President, Asia Pacific, OMP: During the Covid-19 period, we all have witnessed frequent plant shutdowns for a short period of time to quarantine workers or plants running on reduced capacity. All these deterrents compelled companies to quickly reallocate orders from other plants and manage final distribution. At some instances, we even witnessed demand dropping dramatically, resulting in huge production losses. All these scenarios reflect the restrictions of manual planning for companies and warrant the need for having an accurate data model so that they can make timely & accurate decisions and that's where automation takes the centerstage.
S&OP has always been used and practiced for over the last 10 years is still valid, we see a lot of changing and unpredictable consumer demand, which for a number of companies, will result in excessive stock and inventory. Would you think that S&OP can still keep up with what's happening in the current environment?
Loknath Rao: The underlying reality is your raw materials availability can be more constrained than your ability to make finished goods. Former can have much larger lead times and much larger variability in the lead times. A beer company whose network optimization I implemented recently; had issues procuring the specialty corks from Belgium, having a lead time of six months. The company’s S&OP plan was all about doing packaging material availability check and then accordingly decide on the finished goods production plan. Surprisingly, the liquid beer was not a constraint as that could get matured in 15 days and then bottling would just require two hours. But for the procured materials such as corks, they had to wait for six months. That was incorporated in their S&OP as a constraint, particularly availability of bottles, corks and certain labels such as special imprints and embossed labels.
The plan was totally disconnected by design. I advised them that the finished products’ demand forecast should not determine when they should order the corks. Rather they should plan them separately ahead of time using completely different set of planning strategies. These are certain business realities, that sometimes don’t easily transcend on solution design consultants.
Vijay Kasar: S&OP, with its power of decentralized decision making, can still survive given the volatility. With the rising complexities, I believe collaboration is a great focus and given that as a background, S&OP is a perfect process model where collaboration is the center point. I believe from the IBP or the other powerful tools perspective, it is more important to offer companies a longer term horizon while the S&OP has a limit to 12 to 18 months of projections, staying true to the process, but there are other IBPs that companies may be interested to look beyond 18 months in volatile situations.
Prashant Shivsharan: At its core, S&OP (Sales and Operations Planning) functions as a vital tactical layer that bridges the gap between planning and execution. While technology may evolve over time, this crucial layer remains essential, and its importance will not diminish regardless of the external environment. S&OP serves as the primary platform for monitoring the progress of planning and execution, identifying gaps and areas for improvement, and taking corrective action as needed. It is through this process that sales data can be examined and leveraged to ensure that planning and execution are closely aligned, enabling businesses to optimize their operations and achieve their goals.
Komal Ashu: While there is definitely an angle of strategic planning where we’re looking far into the future and collaborating with finance and business to make the right investment decisions given the changing demand map, I also want to focus a lot, at this point, on the whole process of S&OE where we are executing what we are planning for S&OP process every month. We are making the S&OP cycles more agile and representative of the latest scenarios. If there are sudden demand volatilities and we know that it could help us optimize capacities and resources during the month even before our next S&OP cycle, we tap and convert such opportunities at the earliest through smaller tactical decisions taken together with related stakeholders that are brought together by the S&OE process. With S&OP, we talk about one month demand and supply balancing and then drill it down to a weekly balancing of demand & supply, making sure we execute those on-time decisions sooner than later. This obviously wouldn’t have been possible if we did not have S&OP and S&OE processes for short term planning and very short term tactical execution decisions.
Sébastien Aubrey: I think some decisions that are taken during the S&OP reviews cannot be handled in a short time, for example are we going to source material from different suppliers? Are we going to add additional shifts in a plant to increase production, let's say shift on the weekends? Companies have to give it enough lead time. Those decisions have to be still made and S&OP remains probably the best time for the best occasion. For the technology angle, I strongly believe that there needs to be agility in terms of streamlining a company's operational processes before even focusing on their ability to supply replanning. In simplest terms, there's a need to drill down till the operational level and streamline Master Production Schedule (MPS). If you don't have those processes and the data available, the reality of the plan is easily questioned and that can be a great problem. You need to have the trust in the plan, which is based on reality, and you need to have the data that offers you the short-term horizon as well.
What are the main tenets of a successful S&OP implementation?
Loknath Rao: The language of finance is increasingly getting integrated into planning applications now, such as RoI, profitability, price sensitivity analysis, yield, and revenue management, etc. There are indications of product roadmaps from major software companies that they're going to dollarize the whole S&OP process so that companies can simulate various profitability scenarios with various demand supply versions. The mission then is to analyse how expensive and how profitable is the S&OP Plan. Pick best.
Secondly you need to segregate your constraints as internal or external constraints. Things you can control and things you cannot. if you're into the business of mining coal, you will always have a situation where trucks will be in short supply when there is huge incoming cargo at the ports. There are seasons when a lot of cargo arrives. Shipping agents allocate them to highly profitable cargo. So that's an important external constraint. Then you have all these subcontracting or contract manufacturing companies where a small to significant portion of value add done outside of the organization and you may not have visibility into their constraints.
Depending on how your systems and applications are integrated, they may not be in a position to share their currently available capacity or the capacity they committed to you over phone but actually doesn't exist. So, these are constraints that you need to validate in real time.
Vijay Kasar: I just want to add two things – one is the right metrics. I think identifying the matrix and now with the changes that we have seen in the supply chain over the last two years, defining your own metrics for your own portfolios apart from what those standard metrics will help you actually monitor the performance and I think on top of it, the focus of continuous improvement if you keep that lens on, I believe you would just add on to the value of your S&OP process.
Prashant Shivsharan: S&OP is fundamentally a comprehensive approach that relies heavily on collaboration. To ensure that accurate forecasts are developed, it is critical to have the right stakeholders involved in the process. This involves a thorough assessment of manufacturing capacity and supply, as well as effective capacity planning and scenario planning. It is important to measure the performance of the team using relevant metrics and to have executive sponsorship in order to drive agility and ensure the ongoing success of the S&OP process. Ultimately, this approach enables businesses to make informed decisions and to respond to changing market conditions in a timely and effective manner.
Komal Ashu: I think one fundamental thing that I have experienced while setting up the S&OP process is that we need very active stakeholder participation in S&OP. S&OP is, quite a few times, underestimated in terms of the potential that business sees because we’re mostly talking supply chain and not enough business in terms of the cost benefit we bring to the table. I think in order for a successful S&OP, we need to identify the list of stakeholders we want in the process and get their buy-in on the process deliverables. We know that finance is a critical stakeholder today given the dynamic market scenarios. I also see product development or R&D being a very important stakeholder of the S&OP process because time and again, we are talking about supply disruptions and the need to identify alternate products or alternate sources.
The third one is the transportation and logistics team given the rapidly changing trucking and shoring lead times and transportation lane costs. I think it’s pertinent to bring in the transportation & logistics partners in the system so that they can help the supply chain take relevant sourcing decisions and plan better.
Lastly, the sales or the commercial team must be dedicatedly involved in making sure that we are attaining the forecast accuracy levels as an organization and not just as a supply chain measure. In a nutshell, identify your stakeholders for the S&OP process, all those stakeholders that can add value as part of the process and make sure that their representation is well stated and the association is clearly balanced in terms of the process we have every month.
Sébastien Aubrey: Executives need to believe in the data when they are offered scenarios for making decision. They need to feel that the scenarios come from reliable data. First & foremost, you have to have the process. Second and the most crucial aspect is to get full attention from the senior executives. They have to believe what they are seeing and that the decisions that they take will be right based on reality and will be executed later on. We need to show them the source of the data and make sure that people believe it's an accurate data.
With Chat GPT in play, do we still see commercial SNMP continue or do you think there might be a reform?
Loknath Rao: I just learned about this Microsoft's visual analytics platform where you just have to speak up. Ask a question. It converts that into a query and then shows e.g., what was the forecast last year and what was the actual sales and what was the forecast accuracy. You can dig all those queries from a database. By applying this logic, you can gauge the root causes of problems in your company and understand the most frequent issues. If you keep training it by questions and by collaboration in the intranet within your company by creating your own ChatGPT API, companies can understand the reasons for late delivery late or low fulfilment rate, for instance.
Maybe nine out of 10 cases, you'll find the answers are similar. Maybe there is a credit block and maybe one particular set of customers in particular region is more prone to credit defaults. The worst thing about working on a large software like SAP is error/information messages on the screen. An error message can have hundreds of contexts depending on what exactly you're doing. What programs you are executing now. No consultant actually knows the root cause of an error/information/metric. So, you maintain reason codes for non-delivery or reason codes for not fulfilling an order.
I think new age technology companies are curating all these root causes and are making it intelligent for other users to ask questions so that it can give a reasonably good answer next time about possible reasons behind non-fulfilment or a quality defect.
Prashant Shivsharan: Let me broaden this particular question, not limit it to only chat GPT and the generative AI. Let me break this question into three components – What the future can be? What is the overall journey? What we can achieve as on today and how we can build upon it and what are the limitations coming to the part of what can be the future?
Imagine a factory of the future operated by only your two employees – a man and a dog. The man will be there to feed the dog. The dog will be there to keep the man from touching the equipment and we, as planners, are spending quality time on thinking how we can add further value to the customer. I think this is what the future can be, but it's important to understand this is a continuous journey and not just a one-stop solution. If you want to understand the entire concept of the AI, it's important to understand the journey how it has evolved. It was the year 1956 when the concept of AI was born when people started thinking how they can build a machine that can replicate the human intelligence or in some cases exceed the human intelligence. In 1997, the concept of the machine learning came into the picture, which is again a subset of AI. Machine learning essentially learns the existing data and attempts to improve on that data and try to get a prediction out of it. This Machine learning is further improved and then the new concept got evolved in 2017, which we know as Deep Learning. It learns about the overall the predictions coming out of that particular machine learning process. In 2021, the existence of generative AI came into being, which is again a subset of AI. With this evolution, we have moved from the MS DOS concept to the front UI, now having access to the screen and we can ask the question and generate the usual content marketing.
Now coming back to the question – how it is going to impact the supply planning or currently impacting the supply planning. We will always use the technology in combination and try to correlate with real world scenario. Supply planners analyse the data, predict probable scenarios, and based on their experience & judgement planner comes up with most optimized plan. With this process in conjunction, we can use the generative AI to automate our routing task and the process freeing of the remote resource time for the most strategic work, for example, AI powered chatbot can handle the customer queries or order processing, which can eventually the reduce the workload for the supply planning team and empower them to focus on the complex task. I think, it has a huge potential and then it's up to us how we can use it and over the time.
But there is again a caveat over here – this technology still uses the historical data, neural network decisions, we don't know how that decisions are made and what risk is being counted. Planner’s job is to understand the risk coming out of this decision and managing those risks rather than making the decisions or preparing the backend data. That's how the generative AI can play a role.
There is a buzz around chat GPT. Let's try to go to the basic Chat GPT is just one application of using the generative AI principle of technology. Essentially when it comes to a supply planning for the respective organization, we need to feed that organization-specific data to train model. Now the question is – Is there anything which is readily available in the market or other place anybody working on it. Over the period, companies will build that capability, so it is never going to happen that you take a subscription of Chat GPT, which is going to solve your problem. You need to take generative AI application to train the tool and then build specific application for your organization.
Komal Ashu: It’s not about Chat GPT as an App, there are multiple other platforms that are already available. We talk about generative AI. I think I’ll just talk from the experience I have had. In fact, my team has been using chat GPT or similar open AI platforms for their regular day-to-day work and I must say that it is no longer about a specific skill set of programming. The Open AI platforms help us with resources that are beyond the existing skill set of the team. So you no longer have to wait for a programmer to code an information, we can look up the SQL code and curate analytical models depending on the business requirements. That’s where an open AI platform like Chat GPT or others help us bring more agility to our daily work and decision making. We are definitely looking forward to much more evolution of open AI on each of our partner systems, making them more adaptive in nature and helping teams become smarter day by day.
Sébastien Aubrey: Chat GPT is not a supply chain optimization AI tool, rather it's a natural language tool. It helps to put into natural language a huge amount of data. I don't think it replaces the need for the current technology for supply chain automation or optimization because those are specialized algorithms. Many of them are AI-based using the same machine learning concept, but they might enable the planner to understand better why the system made that particular decision or that particular plan, how it was explained, how the algorithm, planning, or optimization solver came up with that particular result. That could be a very powerful application of chat GPT or a similar natural language tool in the future because right now there are a lot of cases where you don't really know how the plan came out from it and that’s how you can build the trust.
Will that technology impact jobs? Do you think that people might be losing jobs because of that technology?
Prashant Shivsharan: I don't think so. I think learning is a continuous process. We need to continuously look out for new technology and trends in the market. If we don't learn relevant changes in our field, we will definitely be out of the market. If you learn the technology, you will have the potential. It's a never-ending debate whether technology is going to replace humans and eventually lead to job loss. The technology revolution has been happening over the last 150 years and people are still having jobs. There will be more jobs in the market with more and more technology. It's all about who is going to learn the new trends and technology and who is not.
Do you think that maybe one day the S&OP process could be replaced by bio and AI, biorobot and they will make decisions for you?
Vijay Kasar: We are looking at the current abilities and the question really hints towards the future. I do not want to undermine the capabilities of the generative AI for now. I believe right now it's already capable to enhance your S&OP process to a certain level and can automate certain operational decisions to a certain level. If you want the chat
GPT and the current capabilities to perform operational decision making for you, I think largely it's there with the available information. If I talk about a probable futuristic scenario that it will completely replace the S&OP process for me, it seems too far-fetched given the human intervention in the process wherein the strategic decision making when it boils down to the judgment and the experience of our own industry and your previous project success, project failures. I believe that human judgment and the empathy towards your own business unit and understanding how volatile and uncertain the world is going to be in due course of time, the historic data doesn't stand tall for any of the future uncertainties. It can definitely make it very smooth to be able to reach the executive S&OP and make the decisions quicker. I think the chat GPT or the generative AI will definitely harness that kind of power.
What would be your main focus moving forward when it comes to supply planning?
Loknath Rao: Intelligent enterprise or an Integrated Enterprise is the idea that a well-known software company is promoting these days, which essentially implies integration within and beyond the borders of the company. A lot of analytics projects are happening right now. People are taking data out of their internal ERP systems and generating certain actionable insights not just reports. Now the question is how those actionable insights will be brought into the day-to-day business as usual transactions? Simple questions like - Shall I increase the delivery quantity? Shall I increase the credit limit? Should I reschedule this delivery? Shall I increase the price? These things are quite static in ERP systems that are mere systems of record. How do you make it intelligent?
Planning systems were again traditionally not systems of intelligence. They were just planning systems. Now with all these intelligences that is getting generated outside of your enterprise systems, how do you bring it back to your execution systems? I think that's going to be a big thing in the coming days.
No matter how much ever intelligence you throw into your expensive software, if you don't have a systematic training program for your employees, it’s not going to work. They may pretend to know a lot, but they actually don't. It's getting harder and harder to keep up with the creativity and use cases of Big Data driven Analytics. I guess training the people, helping them understand the technical nuances and making them feel comfortable is going to lead the way into the future.
Vijay Kasar: From the planning perspective, I believe for 2023 and beyond, the focus will definitely shift to scenario planning. I think the capability of the teams to build more scenarios and come up with simulations is what we are focusing on. Secondly, we are looking out for having interfaces that nurture a lot of collaboration within the organization. I think these two are the main fundamental pillars where we are moving forward.
Prashant Shivsharan: The top priority is ensuring and building the right capability of the team and second is making the basic right. If these two fundamentals are there, I think the rest all can be managed. For instance, if you want to build an alternate supplier but if you don't give order to one supplier, he may not help you at the right time. How are you making the basics right and making sure that it works when it is needed, is the priority agenda for us.
Komal Ashu: It’s important to collaborate with the right partners to foresee challenges and opportunities and de-risk our supply chain -- procurement as a partner, product management as a partner. We need to give enough end-to-end visibility and transparency to the team and make sure that we have the right S&OP process set up for stakeholder collaboration to allow us to act on any risks and opportunities in the short and long term, in the right time & direction of cost and operational feasibility.
Sébastien Aubrey: Being in Asia, I want to focus my energy on, in the next couple of years, refining how we help companies from emerging markets get better prepared for implementing and using those technologies. A lot of companies have tried and many of them have failed, and I think that's real challenge we have to deal with in the APAC region. We need to find the right methodology to help companies get up to speed as soon as possible and improving their RoI. Once they automate it all, the customer service will ultimately improve.