In our latest Insight Hour: Demand Planning using AI/ML in APAC, the aim10x Innovators Network members were informed on Digital Transformation and how an AI/ML Planning Twin can improve agility, increase demand, and drive top-line growth for APAC businesses. The session was presented by Vikram Agarwal (Former CSCO at Avon & Former EVP at Unilever) and chaired by Schalk de Klerk (o9).
Interaction is a core aspect of aim10x sessions, and once again, the conversation between members was open and engaging. Some of the questions posed and answered included:
- What is the best way for Supply Chain professionals who are not programmers to learn AI/ML?
- How does AI/ML use inherently volatile information such as weather data?
- How to decide which supply chain scenarios to build from the limitless options?
- Are Demand Planning roles & responsibilities changing with AI/ML solutions?
- What is the right Return-on-Investment calculation for an AI/ML solution?
The Digital Shift
Vikram Agarwal opened the session with a keynote evaluation of crucial elements for the use of Artificial Intelligence and Machine Learning within APAC Supply Chains. Vikram explained how he preferred to use the term Digital Shift rather than Digital Transformation since the elements being converted by businesses are rarely completed seamlessly.
Supply Chain optimization has undergone dramatic changes over the last 25 years. In the past, the only demand data was customer orders or shipments, and supply chain flexibility was limited. Compare that to the current situation; with highly demanding customers, the rise of e-commerce channels, and countless signals driving demand. The need for immediate and agile delivery has intensified the impact of supply chain disruptions when they occur.
There is a clear digital maturity journey for the variables that Demand Planners can use. The entry points are customer order signal, Point-of-Sale data, and distribution and channel inventory. These historical signals can be enhanced with forward-looking signals such as new product introductions, marketing, and promotional activities, and then external drivers such as weather, political, macro-economic, product ratings, and competitor activity.
Serve the Customer Faster and Enable Growth
Dynamic lane switching is the ability to get a quick result when you want. Should you move to supply your customer from lane A or lane B? Network Design, which used to be based on cost alone, is now a trade-off between cost and agility. The supply chain world is all about customer service and agility, about how fast you can serve rather than how cost-efficient you are. The solution is to construct a flexible, extendable Digital Twin with a Control Tower that is fed and operates with real-time data.
Markets are expanding, the factors controlling them are expanding, distribution management is becoming more complex because you want to reach deeper penetration. There are warehouses, satellite warehouses, fulfillment centers, and shipping points. The order sizes are becoming smaller, the transports are deploying the concepts of ‘milk runs’, ‘honeybee’ shuttles with small vehicles performing replenishments. Achieving greater, deeper and faster distribution is becoming a very significant driver of volume growth.
Additionally, there is the increase of online and direct to consumer trade but it’s not just the multi-nationals, the traditional ‘Mom & Pop’ stores are selling on common digital platforms and can deliver to your home. These new formats of shopping are leading to increased consumer penetration and competition and are a very important driver to capture and measure.
Consumption and marketing patterns are also changing. In earlier days, more advertising equated to more growth, but this calculation is no longer so simple or linear. If you order an item on Amazon, it immediately shows you: ‘Do you also want this?’ and ‘The previous buyers also purchased this – you should buy it too’. Customer fulfillment and commercial growth have become very complex, and this is where AI/ML solutions can read the driver signals and refine demand forecasts more efficiently than planners can.
Challenges and Opportunities Across APAC
The moderator opened the floor for reactions to Vikram’s keynote insights and Schalk de Klerk chaired the interactions of members from Indonesia, Thailand, Singapore, Australia, and South Africa. There were discussions on the impact of COVID and AI/ML solutions on industries and businesses that ranged from tea to pharmaceuticals, and from airlines to surgical procedures.
Members talked about demand fluctuations impacting supply chains in the APAC region. Indonesia was suffering from “a nightmare” of demand plummeting and rising in dramatic waves, while China was talked of as currently being in “a critical state” for one industry.
A participant described issues at Shanghai Pudong airport where 220 consignment lanes were restricted, leading to significant disruption with reduced incoming supply, onward load reductions, and subsequent customs issues. The member asked: “How do you build these scenarios? Because there are too many of them. Is there sophisticated intelligence or a set of things that we can do to make it smarter and less resource-intensive?”
The question was discussed, and an answer was provided; to dynamically update master data in transactional terms, to enable learning to extend the lead times through Pudong and run inventory calculations overnight. The member’s response was enthusiastic: “So even if you sub-modularise it and put a loop around it, that still has a huge impact.”
Another participant described to the peer group about how initial demand sensing using pandemic rates and hospital capacities had provided useful learning if not outright success. The member then explained how these learnings had then helped to define pilot schemes in Vietnam and Korea using channel inventory, secondary sales, and distributor inventory as multi-variables into an AI model.
Learning and Change Management
How to learn about AI/ML, the impact it has on Demand Planning in an organization, and managing that change was a popular topic in the peer session. Participants recounted how their own learning journeys are progressing and it was agreed that the role of Data Scientist was becoming a significant additional skill set.
One member from Australia explained how their company goal had been to get closer to customers and to share information faster throughout their supply chain but that project success depended upon a broad change management program. The change journey required the time to train, learn and grow, with clear guidance on why and how it was being done and to be specific to each region. “To move from a very old system and an old way of working to a modern, different approach to demand and to the way you view information is quite a big step and unless we give people the time and space within their workload to grow and change and do these courses, we will never get there.”
A Business Case Calculation for AI/ML
Just before the Insight Hour ended a final question on Return on Investment was posed by a participant; “In Supply Planning, improvement calculations are standardized and accepted in S&OP but how do you put a value in front of agility and lost sales that might be proposed by AI/ML?” Vikram quickly provided an answer before the networking breakout began; “The best way to measure the business case for an AI/ML planning solution is to construct it around working capital and improved customer service leading to more revenue.”