Making it possible with Demand Sensing

The volatility of businesses in the current day and time has made it almost impossible to predict and meet current demands. The demand planning models like demand forecasting crashed when the world battled unprecedented events one after another, and businesses realised the importance of predicting demand in real-time. The concept of demand sensing took the front seat. In this feature, we draw a comparison between the traditional and the new method of demand planning and explore the possibilities opened by demand sensing.

The past two years were characterised by material shortages and stockouts, leaving organisations almost handicapped when it came to producing and meeting their customers’ demands. In these times of volatility and changing demand dynamics, every supply chain manager dreamt of attaining complete visibility of their supply chain. They wished to know which product is demanded by the customers, when will it be sold, and in what time and quantity. Converting this dream into reality was, however, no longer possible with the traditional techniques of demand forecasting.

The traditional concept of demand forecasting, although fit for mid to long-term demand planning, fails for short-term planning, and this is where demand sensing steps in to bring granular visibility and actionable insights.

Demand Sensing:

The Concept and Goal Demand sensing is a combination of methodology and technology and promises of doing a better job at anticipating and planning for short[1]term changes in demand. Demand sensing uses real-time data and advanced analytics to understand and predict customer demand, allowing organisations to optimise their supply chain and inventory. It is a concept that has evolved from 1822 to 2006.

It is to be noted that historically data, by nature, is disconnected from current events, and this affects the demand. Events including financial downturns, a spike in energy prices, an outbreak of diseases, unrest or disasters and even changing weather alters demand. Thus, the inclusion of data throughout the supply chain is required to determine accurate predictability.

The concept focuses on a customer-centric supply chain and aims to help planners to make short-term decisions based on what just happened hours or days ago and not what happened last year. It looks to eliminate the latency issues associated with the traditional methods.

Sandeep Chatterjee, Supply Chain and Sustainability Leader, IBM Consulting says, “Demand sensing aims to reduce variation in the system. The more accurate the demand, there are less hiccups in the supply chain.”

The goal of demand sensing is to eliminate supply chain lags and build growth paths for businesses through precedented and unprecedented events.

Explaining the goal of demand sensing, Shammi Dua, Vice President, Kearney highlighted the rapidly changing macro business and geo[1]political environment, the unpredictability of customers, the shift of manufacturers and retailers to omnichannel and D2C to delight customers, and the undeterred focus on cost-cutting said that demand sensing helps respond faster to market dynamics.

According to him, “Supply Chain challenges have aggravated in the areas of high inventory, low customer service level, warehouse & transport capacity constraints, and month[1]end push for targets, supply risk, and the drive for sustainability agenda. In current times, demand sensing bullet-proofs the planning process through an enhanced ability: to predict disruptions and pivot plans. As part of the digitalization journey, companies have benefited through end-to-end visibility, integrated planning, and scenario-planning for better & faster decision making.”

 Designed to be more agile and responsive, demand sensing is becoming increasingly crucial for companies to fight back against the challenges that are making the traditional method inefficient.

Demand Sensing over Forecasting

Adapting a new way of demand planning naturally makes one look at the flaws in the traditional system and wonder if throwing more horsepower into the traditional approach could fetch us the desired results. It turns out that the accuracy of time-series forecasts eventually hits a wall, and meets with fundamental limitations imposed by information theory. As per the theory, increasing a model’s sophistication to pursue a ‘perfect fit’ will ultimately reach a point where the outcome is decreased forecast accuracy. Even high-volume consumer packaged products with well-understood seasonality established over decades continue to experience high near-term forecast error rates.

To move beyond this, it becomes imperative to adopt a new planning method that is not based on historical data and focuses on reducing supply chain challenges.

Highlighting why demand sensing the modern way of planning triumphs the traditional demand forecasting, Chatterjee said, “Demand Forecasting looks at the history and other factors and throws a probable demand which is never accurate while demand sensing is actual firm demand similar to an actual sales order.”

Supporting the statement, Dua cites the example of COVID-19. He shares that the pre[1]covid and post-covid environments for supply and demand dynamics are very different; historical data with simple time-series forecast methods may not yield the desired accuracy level.

“Most of the demand planner’s time has been spent in cleansing historic data (manually), reviewing ongoing SKU/category segmentation rules, and statistical forecast review. Little time is left for collaboration of the forecast output and managing input factors (demand influencers),” he said, adding that in demand sensing (with ML tools) the historic data preparation, segmentation, and base forecast generation takes lesser time; releasing quality time for curating the external/ internal input factors, reliable sourcing of such information, and subsequent collaboration with sales & finance functions.

 Traditional demand forecasting which is integrated with ERP takes longer lead times for adjustments while demand sensing which can be integrated with a wide range of systems can have short lead times for adjustments.

Adding another point to why demand sensing is better than traditional demand forecasting, Dua shares, “In the modern set-up of demand sensing, one can plan for prescribed service levels based on profitability and cost-to-serve for various channel, SKU, category, geography segments and also predict service levels at various nodes.”

This is an abridged version of the feature story published in the June edition of the Logistics Insider Magazine. To read the full story click here.

Leave a Reply

Your email address will not be published. Required fields are marked *