Strategic role of data and analytics in modern supply chains

Supply chains today face significant challenges, including fluctuating demand, supplier uncertainties, and increasing complexity in global logistics. Global disruptions, from geopolitical tensions to natural disasters, further add unpredictability. Supply chain leaders today face an array of challenges and the pressure to balance cost efficiency with sustainability, all while managing the complexities of global disruptions, volatile demand, and increasingly stringent regulatory requirements. Navigating these issues require adaptability and strategic foresight, pushing organisations to rethink their approaches and leverage new technologies. Data and analytics in supply chain – the ability to use data and leverage AI, Gen AI, and other advanced analytics techniques such as simulation and optimisation – is proving critical to overcoming these challenges by transforming data into actionable insights and driving innovation in an ever-evolving landscape.

Building robust data and analytics capabilities, however, presents several challenges. Burden of legacy applications, data silos, fragmented architectures, shortage of skilled talent, identification of right analytics use-cases, and unclear mapping of these use-cases to business objectives are just some of the many reasons that are limiting the supply chain leaders to deliver the impact that they envisage to. In-order-to match the heightened stakeholders’ expectations, supply chain leaders need to invest in and build certain supply chain data and analytics capabilities. These are:

Supply chain data management
Supply chain control tower
Digital twin of supply chain
AI enabled supply chain planning
Gen AI in supply chain.

Supply chain data management

Effective data management is one of the key capabilities needed to build a data-driven supply chain. But managing data can be very complex for end-to-end supply chain. There is little debate that generating actionable insights, often in real time, is of high strategic importance to build more resilient and sustainable supply chains and organisations are investing in advanced analytics tools and technologies to achieve the same. The value that these tools will deliver, however, will be of limited value without the ability to process large amounts of high-quality data at high speeds. Supply chain leaders need to work closely with organisation’s technology leaders to build a scalable, flexible, and secure data architecture that can deliver abundant, high quality, and easily accessible data. As per KPMG International’s latest Future of Supply Chain report, where 300 global supply chain professionals across industries such as retail, industrial/manufacturing, healthcare/life sciences, technology, energy, power and utilities and telecommunications were surveyed, 39 per cent respondents plan to invest in digital technologies to bolster their data synthesis and analysis.

Supply chain control tower

There are certain misconceptions associated with supply chain control towers, such as a single software implementation project will enable all control tower functionalities or control towers are only in context of logistics and no other supply chain functions. Beyond logistics control tower, inventory control towers, distribution control towers, planning control towers are all being conceptualised and implemented by organisations today, many of them leveraging organisations’ existing data and analytics technology architecture. Supply chain control tower is a set of capabilities built over a period of time layering on increasingly advanced capabilities. A good way to think about control towers is around the four levers of visibility, insights, integration, and automation. Further, each of these four levers could have different maturity levels. For example, maturity levels of insights lever could range from descriptive to diagnostic to more advanced predictive, prescriptive and simulation capabilities. Supply chain leaders could assess their current maturity levels along the four levers and then identify use-cases that can help them move from the current level to the next. There is no single definition of what a supply chain control tower is, or is not, so every organisation will have to think through what differentiated capabilities it wants to build, and which use-cases can help it achieve the same.

Digital twin of a supply chain

Supply chain modelling and simulation can help us in managing disruptions and uncertainties better. A simulation model of a supply chain provides a digital environment to test out multiple forward-looking scenarios and identify an optimal course of action. This is especially useful when we are faced with an unknown-unknown kind of risk and there is no precedence of how such an event should be handled. For example, one can understand the impact of upstream logistics disruption in a data-driven fashion on the overall supply chain and figure out how long it will take before the customer orders start getting impacted. Once the impact is understood, subsequent mitigation strategies and next steps could be pursued much more confidently. Supply chain leaders are increasingly facing such events and hence building a digital twin of a supply chain becomes a critical capability to invest in. A digital twin is not only helpful for managing disruptions and uncertainties but can also help in testing supply chain policies in a risk-free environment before actually executing them on the ground.

AI enabled supply chain planning

Traditionally, organisations have relied on what we call consensus planning for their demand planning process. Different organisational functions, such as operations, finance, sales and trade use standard statistical techniques, historical sales data, and some external data to generate their own forecast. Then all functions get together to arrive at a consensus forecast. The process takes a few weeks of time and by the time consensus is reached sales data used is already old. A much better way could be to use both internal and external data and apply Machine Learning (ML) techniques to arrive at a single forecast. For example, an FMCG manufacturer can combine historical secondary sales data, prices, discounts, and other promotional data, with external data such as macroeconomic indicators, social media mentions, temperature, holidays, and competitors’ data to generate a more accurate secondary forecast at distributor level. As per bullwhip effect, predicting demand down the supply chain should be easier than predicting primary orders and hence a better secondary forecast translates into a better primary forecast and subsequently into a more effective supply plan.

Gen AI in supply chain

A lot of attention around Gen AI has been on large language model (LLM) or the foundation model layer, which is generally a very capital-intensive exercise. For a market like India, a much larger value can be created and captured by building applications on top of these foundation models. Three useful archetypes to help one think through applications of Gen AI could be writing, reading, and chatting. Example of writing tasks could include translation related tasks or a web user-interface of a Gen AI application supporting in brainstorming or code generation to help developers/programmers improve productivity. Similarly, Gen AI models can read complex documents and can summarise key points of a document. Some applications of this archetype could be text proofreading or customer feedback analysis or summarisation of contracts/legal documents. For example, an LLM based software application can help you scan through a wide array of subscribed and unsubscribed data sources from the country where you have the largest supplier base and can help identify any instances that need attention. These early warning signals are otherwise difficult to catch. This example also falls under the reading archetype. Gen AI models can perform conversations with users by generating human-like text responses in real time. For example, organisations can look at building a conversational supply chain application where users can query in their natural language and get responses that aid in real-time decision making. While some of these applications have been built earlier using AI techniques, Gen AI significantly improves the quality and speed of development.

Conclusion

Supply chain teams are generally busy in their day-to-day operations and do not get time to think much about how they can leverage data and analytics to improve health of their supply chain. Similarly, if technologists and data scientists are left to their own, they also cannot bring about the kind of wholesome innovation supply chain needs. Hence, we need to build what we say an analytics mindset that allows supply chain teams and leaders to operate at the intersection of functional knowledge and data and analytics technologies know-how. An analytics mindset essentially means that when you are encountered with a business problem, you can think through what data and analytics tools and technologies can help you solve this problem; similarly, when you are learning about some new tool or technology, you can identify which business issues this new data and analytics technology can help you solve. Hence, investing in supply chain data and analytics capabilities is no more an option but an urgent imperative that needs to be undertaken.