HITECH BYTES

How Next-Gen Digital Supply Chains Boost Forecasting Accuracy

 
August 20, 2019

Today, digital is viewed as a strategic driving force. It’s easy to see why. Thanks to digitalization, the world’s most valuable retailer carries no inventory, while the world’s largest taxi company owns no vehicles. In this new paradigm, supply chains cannot remain untouched. Concepts like master data, missing data, spreadsheets, electronic data interchange (EDI), and so on, could soon be things of the past. For instance, imagine order management using Amazon’s Alexa. The e-tailing giant has always been a front runner in tech-enabled supply chains. Some of its patents include machine-readable codes to replenish its inventory and predict safety stock levels.

As most tasks in organizations are handled by sophisticated algorithms, accurate intelligent forecasting is critical to their success in today’s complex and competitive supply chain ecosystem. A precise forecasting strategy helps optimize manpower utilization (resourcing, analysts, and subject-matter experts), strengthen automation, facilitate on-time and error-free delivery of goods and services, reduce variability, and enable better coordination between systems and strategies. The result: the ability to meet customer needs in ways that enrich customer experience.

Intelligent Supply Chain Forecasting for the Digital Era

Augmenting intelligence from multiple data sources is the key to driving effective forecasting systems, and in turn, it curtails operating expenditure. For instance, in the high tech industry, a leading software and hardware manufacturer uses 43 different factors, combinations of scenarios, and tools to arrive at a forecast. These factors include product demand, cash flows, transportation time/cost, manpower, capacity, availability, down time, lead time, seasonal effects, history, and more. Besides these, manufacturers have to factor in managing demand and stock keeping units (SKUs) in their forecasts. A number of elements affect demand, including discontinuous and wild fluctuations in demand, final product demand uncertainty, sudden demand slack, the negative impact of the economic slowdown, and others. New SKUs, variable aggregate past benchmark, and overstocking by channel members impact SKU management in forecasting. Additionally, manufacturers must account for competing SKUs, such as the iPhone 8 vs iPhone X, as they can cannibalize each other’s sales. Some of these parameters are stochastic in nature, resulting in homoscedasticity. Parameters having a random probability distribution or pattern can be analyzed statistically, but may not be predicted precisely. This leads to inaccurate forecasts.

The good news: the growing number of connected devices and advancements in technologies such as Big Data, machine learning, and Blockchain are enabling high tech manufacturers to leverage data-driven, smart automation for supply chain optimization to ensure increased forecasting accuracy. Harnessing the abundance of data and applying advanced statistical methods to address the most likely demand scenarios, can help high tech companies leverage the right information at the right time to enhance profitable decision-making. Such a data-driven digital solution must incorporate the following aspects to drive superior forecasting accuracy in the supply chain:

  • In addition to traditional algorithms, use advanced statistical theories like nonlinear forecasting, stochastic differential equations, or bass diffusion model to accurately predict demand patterns that are generally difficult to model employing standard algorithms (i.e. statistical forecasting, econometric methods).
  • In case of wild fluctuations in demand, employ extreme value theory using large deviations and simulate the probability of events.
  • In case of sudden demand slack, predict rare events based on historical behavior, calculate rare event probability, and simulate scenarios to provide forecasts (also known as anomaly detection). 
  • In case of overstocking by channel members, develop feedback loop to monitor end-product sales with SKU demand.
  • Instead of completely filtering out the ‘chaos’ in the data, use it effectively for data modelling. Leveraging in-use data increases the velocity of doing business. 
     

Supply Chain 4.0

Evolving customer expectations combined with emerging trends such as adaptive intelligence, a digital workforce, and Business 4.0 compel companies to reimagine the future of supply chain management. Organizations that design supply chain forecasting systems that are faster, granular, and more precise will build the agility to respond to changing market conditions, thereby minimizing the impact of external factors and fostering stable growth. 

Suhas J is a Managing Partner with the Enterprise Transformation group within TCS’ HiTech business unit. He works on next-generation digital transformation engagements and has been part of several strategic solutions consulting and implementation projects for TCS’ global clients. He holds a bachelor’s degree in Engineering (Computer Science) from BMSCE, Bengaluru, India and a post-graduate degree in management (strategy and information systems) from Católica Lisbon School of Business and Economics, Europe. He has also completed an executive education program on ‘Strategy in the Digital Era’ from the Indian School of Business, Hyderabad, India.