Logistics Handling spoke with leading industry spokespeople from the analyst and vendor community about how modern Demand Forecasting & Planning/Sales & Operations Planning (S&OP) solutions and other planning-related technology can help to facilitate better supply chain, manufacturing and distribution practice.
Without a doubt, two of the most significant innovations currently driving Supply Chain technology are the advent of Machine Learning techniques combined with embedded advanced analytics. Indeed, as Shaun Phillips, global product director, QAD DynaSys, explained, the utilisation of Machine Learning techniques has moved from its embryonic stage to a phase of early maturity. “This is more evident in Demand Planning solutions forecasting products whose future sales behaviour is influenced by events other than historical sales,” he said. “These include periodic products, low-frequency high demand and products with no or little history. The same technology lends itself to the automation of exception resolution and the augmentation of daily decision making.”
So, what has driven these developments? Phillips maintains there are several forces behind this change that have fused into a powerful nexus. “The first is the unprecedented availability of real-time data that perpetually present risks and opportunities,” he said. “The risks challenge the feasibility of the plan while the opportunities question the efficiency of the plan.” Phillips believes the second is the mainstreaming of Artificial Intelligence (AI) and advanced analytics technology. “Although we have exponentially more data, we have more tools to extract fact-based insights,” he explained. The third is millennials. “This generation has entered the workforce questioning why enterprise software, such as Supply Chain Planning, does not look, feel, connect and respond the same way as the screens they have been raised on. This generation has an unprecedented comfort level with trusting AI decisions, deploying on Shared Platforms, and outsourcing business functions to Cloud Services.”
Horizonal and vertical alignment
Tim Payne, research vice president, Gartner, made the point that, whether it’s Demand Forecasting & Planning or S&OP, they are all planning at the end of the day. “It's all about decision-making in the supply chain,” he said, adding that there are currently three big megatrends in this field. The first one he calls horizontal alignment. “Horizontal alignment means companies don’t want to take planning decisions independently – they want them aligned across the supply chain horizontally including the customers and suppliers,” he said. “This trend has been happening for a while now, but that’s what drives the vendors to do what they do as well in terms of the evolution of their planning solutions.”
Similar to that, the second trend cited by Payne is vertical alignment. “When companies make planning decisions at a very granular level, short-term at the order level and make plans and decisions tactically and strategically, they need those layers of planning to be aligned as well,” he said. “The poster child for that is S&OP – companies don't want to have S&OP sitting in an ivory tower sucking data in and making decisions that never go anywhere. So, you’ve got to work the layers and you’ve got to be able to have the right level of alignment/communication between the different layers of planning that are going on.
“Ultimately, it's about your strategy – that informs what your supply chain is going to look like. It informs how you are using the resources you have available and informs how you are going to be able to respond to what's really happening in execution, and that's how you execute on your strategy. So, that vertical alignment is now really important and that's why we seen less distinction between the different planning solutions. Today, nobody really says I’ll think about a demand planning solution first, then I'll think about the supply planning solution and then I'll think about an S&OP solution – because you won't get the alignment in either direction.”
The third big trend highlighted by Payne – one that he believes has really taken off over the past 12 to 18 months – is automation; how to automate those decisions that you’re making. “This is where most of the main planning solutions vendors now do something around machine learning and AI,” said Payne. “It’s all about the desire from the market to reduce the amount of manual effort that goes into planning. To automate that, it doesn’t necessarily mean autonomous planning where no human is involved; it probably will in some cases where special decisions need to be made, but it certainly is about augmenting the decision-making that the planners do. So, if you look at what end-user companies are thinking about and the technology roadmaps they are working with, a lot now happens around those two dimensions of alignment in automation because there is now more desire in the market for what some people call digital planning.”
In terms of uptake of these solutions and strategies, Payne reflected that this is a maturity issue. “There are immature big companies and there are mature small companies, so it’s not necessarily about the size of the company. Indeed, it can be easier for smaller companies to embrace these trends. If you’re a big global multinational and you have operations in many different countries, it’s probably going to be more difficult to standardise and coordinate things; what Germany wants to do might be different to what Italy wants to do or the US wants to do. So, the size of a company really isn’t the issue – it’s all about maturity.”
Disruption in the value chain
Archana Vidyasekar, research director, Visionary Innovation Group, Frost & Sullivan, commented that technological revolutions in the logistics space have resulted in minimising the current complexities and creating new opportunities for value-chain participants. “Key future themes, including AI, autonomous technologies, digital platforms and Blockchain are expected to create much disruption in the value chain and give rise to new participants, business models, and disintermediation among others,” she said.
From self-running trucks to automated contracts, Vidyasekar believes the supply chain is ripe for innovation. “This is expected to create much disruption in the value chain and give rise to new business models and, potentially, even lead to disinter-mediation within the supply chain,” she said. “For instance, the industry is making a conscious shift from asset-centric models to more digital asset-light approaches, enabling new types of services such as on-demand, real-time, and agile last-mile delivery solutions suited for the uncertain nature of the urban supply chain.”
Richard Goluskin, director client services, Panorama Consulting Solutions, made the point that in the past traditional forecasting solutions tended to be built on historical information. “Now, we have better computer capabilities base around analytics, and we have also seen the development of machine learning and artificial intelligence (AI),” he said. “These concepts may still be in their infancy, but a number of vendors are trying to incorporate those types of technologies and trying to apply them to better forecasting algorithms and better supply chain visibility and management.
Together with that, there are just improvements in technology. Say you have something in the supply chain that is coming from China and it's currently on a ship in the middle of the Pacific Ocean. There's now software in the systems that can give you real-time visibility in terms of where that ship is. For example, it might be circumventing a storm in the middle of the Pacific that is going to require it to reroute and take a day or two longer to get to a port which, would have an impact on the supply chain. Knowing in advance that there will be a delay in delivery can give you time to re-plan your day’s or week’s supply chain activities more efficiently. By combining track and trace with GPS functionality all in real-time with graphical capabilities you can see that information rather than just rely on a tabular bunch of numbers.”
Bryan Ball, vice president, principal analyst supply chain management, Aberdeen Group, is a big believer in S&OP. “Indeed, I installed it twice myself as a practitioner,” he pointed out. “In the early days around the time ERP was first implemented S&OP used to be called production planning. The idea was to get a good supply and demand match and it's amazing to me how many people still underestimate the value of that. Think of a fast-moving organisation where the sales team is given an order, they pass it to operations and the operations personnel scramble to make it happen.
This is achieved, but at what cost to the organisation? If the organisation is in growth mode and keeps making money maybe there isn’t any major problem, but there is a point when the organisation – particularly as it matures – really needs to focus on disciplining itself and focus on refining the supply and demand match. Once the mechanics are in place and everyone understands that is the goal, then you get to the sophistication stage. This is where I think things have moved to and where things have actually happened over the past five years or so – connecting that S&OP and integrated business planning (IBP) into the financials, so that as soon as you put the operations plan into effect you can see the financial impact at that point. In other words, you are going to project what the results will be based on what your operational plan is.”
In terms of running S&OP scenarios, the ‘what-ifs’, Ball made the point that this isn’t something particularly new. However, he considers that where there has been improvement is where vendors such as Kinaxis can help companies run those scenarios almost as quickly as snapping their fingers. “Being able to run multiple scenarios quickly is probably one of the most valuable things you can do,” he said. “This is when you get really predictive. For example, on the demand side you might want to determine what is likely to happen if you change the price or run a promotion. If you want to improve something at market level, you might want to stop pushing some new items because they’re not making as much money as you would like or drop the price on something that could move along faster. The opportunity for this type of improved demand planning is there. The software and the tools are so much better now but not everybody has them. However, I think once you get the basic supply and demand match in place you can start to see things you never saw before. If you start changing the mix and you start doing demand shaping and you change the price all of a sudden the obvious solutions can come out.”
The power of social media
Ball believes social media can also be of major benefit from a Demand Forecasting and Planning perspective. “As an example, maybe sales for a particular clothing product aren’t going as well as you expected, but people on social media say they really like it except the button or zipper is in the wrong place. This type of feedback from people who have actually worn the clothing can provide you with just the right nuances in terms of what could make the product better and therefore more attractive to customers. Let’s say I'm looking at a particular product on the planning screen and I’m trying to understand why our forecast was overly optimistic. Without leaving the screen I can look at social media input in a sidebar and this could give me the answers I’m looking for. So, social media doesn't become a separate project; it's part of the stream of data that I can look at while I’m evaluating what's happening with the forecast. It could be that things are flying off the shelves and the analytics aren’t providing all the answers as to why this is the case. Again, social media feedback could be a useful indicator of the level of interest. Over the past two or three years I’ve seen more companies incorporating social media within their demand forecasting and planning activities.”
The omnichannel model
Goluskin believes the omnichannel supply chain model is one of the main drivers for developments in the world of Demand Forecasting & Planning and S&OP. “If you think of B2B (business to business) and B2C (business to consumer), in the B2C world there is a major focus on satisfying the customer; that’s really what’s driving this market,” he said. “So, there is a shift from the predominance of the traditional bricks and mortar store to more online ordering, and Amazon in particular is setting the pace. I ordered something from Amazon today and was asked whether I would like to receive it by 9 PM tonight. Imagine the logistics required to be able to guarantee that product a can be delivered to a customer the same day. You can’t just have 10 warehouses across the country; you need a large number of forward depots where you can have the product available for the customers. And you have to forecast each product’s life; you can't stock everything all the time so you need to be very intelligent about what you should stock. So, the driver is customer expectations but the complexity behind that is logistical, knowing what customers are going to buy. The whole concept of trending products and placing those products in locations where the turnaround can be very quick is key. So, the Amazon model is putting a lot of pressure on other organisations to produce excellence in terms of delivery and customer satisfaction.”
In the B2B world, Goluskin reflected that the pressure is probably not quite as high; at least when he talks to clients about what their expectations are as far as the supply chain is concerned. “They don't have the massive numbers that are involved in the consumer business and mainly deal with the more predictable goods,” explained Goluskin. “However, they are starting to be influenced by what they see in their daily lives when they make B2C purchases, so they are beginning to think why can’t I get better information from my suppliers in terms of where that shipment is, and why can’t I be notified by my supplier if there is going to be an exception to my shipment. So, the pressure isn’t normally so high in B2B, but I think they are starting to trail the B2C world in being able to have real-time visibility of their supply chain and benefit from real-time notifications. One of the big drivers in the B2B world has always been cost, but more and more companies are now basing their buying decisions on service reliability and accuracy; those kinds of things. So, this whole topic of demand forecasting & planning and S&OP is important not only from the buyer’s point of view, but also incredibly important for the supplier as well.”
Digital supply chain twin
Payne believes the omnichannel model can also benefit from the horizonal and vertical alignment principles he previously highlighted. “Don't run all your online activity as something completely separate, like a lot of manufacturers do because that’s the way their systems were built,” he said, adding that this need for better planning is driving a move towards what Gartner calls the digital supply chain twin. “This is the notion that with all this granularity and near real-time data that’s available every planning solution no matter where it is in the world has two major components; it has to have a model of the part of the supply chain it’s trying to plan and then it has to have analytics that run on that; mainly predictive analytics because you’re predicting the schedule, predicting the demand plan or predicting the replenishment plan.
“There’s a lot of focus on the analytics these days where companies say they are going to apply machine learning and it's going to figure all this out and we going to see all the causality. OK, so this will give them better analytics, but what's the model like? The models are often out of date, stale, static, when did they ever last look at their lead times, did their lead times change during the year – well actually they do, but the models don't say that, they say whatever number was put in the ERP system whenever it was implemented. So, the argument here is analytics are great but if your model isn’t very good it’s not going to be a reliable representation of your physical supply chain. It doesn’t matter how good your analytics are you're still running it on a bad model which means you're going to get a bad plan. This is why companies should focus more on the model so that it offers a better representation of the physical supply chain. Your physical supply chain is a living breathing thing, it's changing all the time, so your model needs to track that. Therefore, you get to the realisation that what you need is a better digital representation of the physical supply chain – hence the name digital supply chain twin – that is more living and breathing. It needs to be able to track much more closely what's actually happening in the supply chain and will happen in the supply chain because it can become time phased as well.
“We are seeing a lot more interest in that as a notion coming up from the market because with a lot of the newer technologies coming through there are opportunities now to be able to create a much better model of the supply chain upon which you can then apply your predictive and prescriptive analytics. That then helps with your different business models, whether you they are online or traditional. If you have a good digital twin of the supply chain, you can get that updated very quickly in near real-time. Then, you're in a really good place to be able to see, respond and plan and make decisions about you what you really want to do in the supply chain. This tends to be something that the more mature companies are really starting to think about now. It's the natural endpoint of that horizontal and vertical alignment – the need for a model that represents your supply chain more accurately. There are some really interesting solutions being introduced by some of the newer entrants in the market in terms of how they create those models from the data – which is all Cloud, Big Data, analytics, machine learning driven.”
Picking up on the theme of the Cloud, has the Software as a Service (SaaS) model had any notable level of impact on the planning-related software solutions market so far? Phillips made the point that QAD DynaSys customers want systems and the associated costs that scale in a linear manner with their growth. “Our customers do not want to be experts in cybersecurity or disaster recovery, our customers want to focus on their core business,” he said. “The Cloud is no longer a deployment option; it is a unique market segment. Tech vendors must offer a true Cloud SaaS solution that leverages the benefits of a shared infrastructure platform. The on-premise SCM solutions will deservedly soon become a relic.”
Phillips believes the commercial case for SaaS/Cloud will deplete the on-premise market. “The efficiencies gained with shared infrastructure and shared services will offer compelling cost savings over the on-premise model,” he said. “Furthermore, in addition to the lower cost, the reduced risk from cyber-attack, the improved performance from 24/7 monitoring, and the increased up-time due to hardware redundancy will make the decision for the Cloud more powerful. Tech vendors are investing heavily to provide the lowest cost shared platform with reduced royalties via open source components and pay per use licensing models. At QAD DynaSys, we foresee the single version platform utopia with pushed software patches and releases. This will, in turn, further reduce the providers costs-to-serve. We believe hybrid SaaS/on-premise architectures will exist only as a transitional state on the journey to full SaaS, or in specific situations or industries such as aerospace and defence.”
Is security an issue with regard to Cloud-based solutions? Payne commented that the issue of security still comes up occasionally, but nowhere near as often as it used to. “Now, a lot of the software is Cloud only anyway,” he said. “Occasionally, particularly with aerospace and defence, they can get more sensitive about security, but there are big Cloud platforms running the US Department of Defence, so it can be done. And as long as the Cloud platforms have all the latest security certifications then it doesn't seem to be a problem. Sometimes it can be an issue in terms of where the data is held. If a vendor is using their own Cloud and they haven't data centres in Europe that could be an issue. But if you’re SAP, Oracle on Microsoft, or if you’re Amazon or Google you have data centres everywhere so it's not really an issue.”
Goluskin thinks Europe leads the way as far as privacy protection is concerned. “The US seems to be considering adopting more stringent data privacy and security type regulations, and when you hear of the scandals surrounding companies that have suffered data breaches I think invariably it’s going to have an impact on the whole landscape of data privacy. I don't think it’s going to be limited just to consumer data, so this could consequentially have an impact on the type of data management involved in the demand planning world. My view is that the US is trailing Europe at the moment from a data security perspective, but I foresee that in time similar kinds of regulations are likely to be adopted in the US as well.”
Legislation wise, Ball believes Brexit will be one development that will result in considerable market volatility. “For example, maybe there was a supply agreement in place between companies, and now there's not because of different rules and regulations that are put in place post-Brexit,” he said. “I don't want to overstate things; I'm just anticipating that there will be some level of disruption that will need to be settled. Perhaps this will be resolved quickly but I think there is going to be some legislative agreements that get erased and will need to be re-established in accordance with new rules.”
Ball also said that where Trump has changed the US trade agreements with China this could introduce new suppliers into the supply chain equation that weren't there before because of the new tariffs or reduction in tariffs. He added that the converse could also be true. “Some companies that were big players because they were taking advantage of the old tariffs could find themselves less competitive. So, I absolutely think those kinds of things are going to have an impact. Exactly how and where it’s going to hit I don't know, but that’s a ‘stay tuned’.”
Regarding the pharmaceutical industry, Ball thinks there could be further legislative impacts affecting companies in that sector. “Opioid-related issues have resulted in some tough laws that are now in place, and companies now have to report orders that are anomalies – if they don't they could be fined or taken to court.” He added that these types of things need to be understood and addressed in order for supply chains to remain legally compliant.
Best of Breed
Is there still an argument for best of breed? In the case of Planning & Scheduling systems, Phillips believes the argument for best of breed has never been stronger. “Consumer-grade software like Google and Facebook have homogenised the user experience,” he said. “With the advent of the Cloud, all solutions appear to ‘reside’ on the same platform. The delineation between disparate systems has never been so nebulous. Furthermore, some of the larger ERP vendors are struggling to deliver supply chain value with monolithic transaction-centric systems. We observe that in many supply chain sales opportunities, ERP vendors have been omitted from participation or have disqualified themselves.”
Have ways of best integrating these types of best of breed systems with other solutions developed to any notable degree over the past year or two? From a technical level, Phillips believes the capability of connectivity has increased. “Using RESTAPIs and loosely coupled messaging, dissimilar systems can give the perception of using a common data model,” he said. “However, the increased complexity lies in the additional number of disparate data systems required to present an end-to-end supply chain picture. Whereas in yesteryear supply chains integrated with one or more transactional (ERP) systems, a typical supply chain planning product requires interfaces into ERP, CRM, PLM, MES, TPM, TMS, and WMS. And these are just the in-house systems. When connecting the end to end supply chain we look to harness external data sources such point of sale data, and opportunistic freight and procurement opportunities.”
Goluskin made the point that the topic of Demand Forecasting & Planning and S&OP all revolves around the supply chain, adding that, within this context, one of the big trends that has been developing for a couple years is greater complexity. “Depending upon the product that is being produced, supply chains can involve many different suppliers in many different locations in many countries,” he explained. “So, the complexity of the supply chain is what’s driving the need for better planning, and that’s really what the software vendors have been trying to address. They’ve been addressing this by trying to provide best of breed forecasting and planning tools where you can see a lot of that complexity in the supply chain through having better tools and better algorithms for preparing forecasts.”
And what of the role mobile devices are playing in the world of planning and forecasting? Goluskin considers that mobility is such a major expectation in just about every walk of life these days. “That applies just as much to the area of Supply Chain Management and Demand Forecasting and S&OP as it does to any other business activity,” he said. “If there is some disruption in the supply chain you want to know about it right away; you don't want to have to go back to your desktop computer – you want to receive notification on your phone informing you that an exception has happened so you can let your customers know that there may be a delay in delivery etc. In today’s supply chain environment, you want exception recording and real-time notifications and alerts for things that are happening.”
Big Data and IoT
What effect are methodologies such as Big Data and the Internet of Things (IoT) having within the Demand Forecasting & Planning solutions space? Phillips made the point that a large and growing amount of data is now required to drive fact-based supply chain decisions that doesn't reside within the four walls of the enterprise. “This is a significant evolution from 10 years ago where most supply chain data existed in-house,” he said. “Consequently, Big Data and IoT connectivity are having a huge impact on the world of SCM technologies. However, not so much because of the connectivity difficulty. Accessing external data is becoming simpler via RSS feeds and RESTAPIs. The challenge of IoT is how to cleanse the data, how to structure it, and finally determine if and how it adds value to the plan. To overcome this, QAD DynaSys introduced intelligent analytical techniques to make sense of all the data. Our target was to find actionable insights from within volumes of data. We used this technique in Demand Planning to identify trends and causation and determine suggested actions. We have only just started to realise the potential value of data.”
What do our commentators think will be some of the key developments to lookout for in the world of Demand Forecasting & Planning/S&OP solutions over the next year or two? Phillips believes the role of digital twins will become more tangible. “This will greatly impact Demand and Supply Planning solutions as parameters such as price, cost, lead-time, batch-size, run-rate etc. will be derived from a physical object or transaction rather than a master data management system. Business Process Automation will start to gain momentum and the hands-off planning environment will become to look like a reality although it will take some time. This will re-focus planners to engage in value-add tasks such as analytics and scenario planning. Like many supply chain leaders, I believe the emergence of Blockchain will be temporarily hindered by the lack of scale and security.”
Goluskin thinks systems will continue to get better with integration of unstructured data and various sources of data. “For example, when you are planning the supply chain maybe there is a weather forecast out there that could affect the way you do your planning – it could inform you that there is a risk of winter storms in the Upper Midwest in the US. So, going forward it’s not just about further developments involving AI and machine learning – which are just better ways with coming up with the numbers – I think it’s also about the integration of many different kinds of data. And in terms of unstructured data systems are going to get better at interpreting those things and using AI-type techniques to put all that together to help organisations to plan better. I think it’s all about integrating data and having the algorithms to do something with that data that was never done before in a structured way – that is the frontier.”
Vidyasekar anticipates that autonomous technologies such as drones and autonomous trucks are expected to plough the road, as they drive efficiency through cost savings from less fuel consumption. “Approximately 200,000 trucks are expected to be making autonomous deliveries by 2030,” she said. Vidyasekar also believes Blockchains will introduce more trust and transparency into the supply chain, improving business process agility. Finally, Vidyasekar maintains that artificial intelligence approaches will lend new cognitive capabilities to moving and thinking assets in the supply chain.
Payne believes it’s a question of further evolution regarding many of the areas highlighted above. “For example, we will see more developments in terms of AI and machine learning coming to the fore,” he said. “At Gartner, we paint a picture of what we think that will look like eventually – something we call Algorithmic Supply Chain Planning. So, there will be much more automation in place, and more autonomous things happening from a decision-making perspective. It won’t be everywhere as somewhere people think. It will be significantly automated but there are still decisions that require human input, and this need won’t go away. That said, a lot of the basic activities – particularly the short-term tasks – can be very highly automated and that will continue.”
Ball thinks the trend towards greater platform-to-platform connectivity will also have a greater impact on Demand Forecasting & Planning and S&OP. “It is now possible to look way beyond where we have been able to do in the past. In terms of the wider supply chain – you can access valuable supply chain data not just from your customers, but from your customers’ customers. Also, you might already be working closely with your suppliers' suppliers, but it could be possible to go back three or four further levels. Then, moving upstream and looking at the demand side the platform-to-platform model could mean you can see from the distributor into the retailer, and from the retailer to the end customer in the store. So, you can start to look at things that haven’t been within the normal scope of your planning tools up to now. I think we’ll be seeing more of that type of connectivity and data access, which could help you to make better forecasting and planning decisions.”
Additonal content on Demand Forecasting & Planning and S&OP can be seen at Manufacturing & Logistics IT magazine