The Seven Essential Capabilities of Demand Sensing

One crucial element in transitioning supply chains to be service-driven is enhancing Demand Sensing. This is a capability and technology for improving near-future forecasts using detailed short-term demand data. Near-future means hours or days, depending upon how dynamic your supply chain. Generally, the implementation of Demand Sensing results in reduced forecast errors, improved inventory accuracy, and optimal deployment of downstream inventory, such as in Distribution Centers and Sales Depots.

In a demand sensing environment, downstream data such as customer, point-of-sale (POS), or channel data is utilized to discern demand trends, offer early detection of issues, and eliminate the lag between planning and real-time supply chain activities. Rapid identification of deviations enables companies to respond swiftly and with greater intelligence.

Demand sensing can also use a much broader range of demand signals (including current data from the supply chain) and use different mathematics to create a more accurate forecast that responds to real-world events such as market shifts, weather changes, natural disasters, consumer buying behavior, etc.

Why have demand forecasts failed to improve in recent times despite technological advancements? One explanation lies in the reliance on "aggregate level planning" employing time-series methods, which frequently fall short of achieving forecast accuracy at item-location levels, even for fast-moving items. The use of aggregation, combined with slicing and dicing rules based on historical patterns, often obscures the latest trends and patterns at an item-ship-to-daily level. This aggregate approach followed by slicing/dicing introduces latency into plans and conceals the true demand signal. Demand sensing addresses precisely this issue by providing a solution that overcomes these limitations.

What essential capabilities must a supply chain solution possess to initiate demand sensing effectively?

The ability to model demand at the most atomic level, such as Item-Sold to-Daily basis. Ship-to locations can be key accounts, retail stores, sell-in channel partners, geographical territory etc. This granularity is very important because demand models need to be iterating using the latest supply chain demand data at this atomic level and to identify the statistical relevance of short-term spikes, outliers, trends, and patterns.

The ability to model demand variability – A demand confidence interval is needed to understand the latest data feeds and segregate noise from the demand signal. This is mandatory because noise has no statistical relevance, hence must be discarded, otherwise we will end up with a “nervous’ supply chain. And using a normalized variability (say plus or minus 10%) is not enough. Normal distribution might work for regular fast movers but for slow/intermittent, variability distribution could be anything but normal. So, with normal assumptions, it doesn’t understand true deviations well enough, causing numerous false positives and false negatives. This mandates variability distribution to be adaptable and not limited to a few known distributions.

The ability to use downstream data – This could be ship-to data, VMI feeds, POS data, collaborative planning, etc. Though actual demand data is available at the line order levels, aggregating to weekly/monthly level to generate statistical forecasts would mask precious insights into evolving demand patterns and line order distributions.

In advanced demand sensing, external variables like weather forecasts, economic conditions, and oil prices, among others, can be integrated into demand forecasting to predict short-term demand patterns. For instance, prolonged sunlight may boost beer sales, while extended periods of rain may increase washing machine sales. Employing Causal-based Demand Sensing enables a manufacturer to strategically reposition available inventory to locations expecting substantial rainfall, thus avoiding lost sales to competitors. Short-term weather data integrated into the forecasting system facilitates rapid inventory repositioning. However, merely receiving data feeds is insufficient; the solution must interpret it effectively and translate it into demand signals to drive deployment planning. To achieve such effective translation, the base statistical forecast models should operate at the most granular level, as previously mentioned.

Supply chain planning platforms must scale to process high volumes of data associated with hundreds and millions of items–location combinations every hour/day and at the line order level.

To gain potential network benefits, the platform must seamlessly integrate planning and execution processes while efficiently replenishing high-frequency demand signals with optimized execution. Crucially, there should be zero latency between planning functions, such as between the demand module and inventory and replenishment modules. Unfortunately, many supply chain planning solutions have evolved through acquisitions. Consequently, forecasting may utilize a different data structure from replenishment, with later additions like inventory planning modules introducing yet another data structure. Such discrepancies can severely hinder Demand Sensing capabilities.

Increased process automation is required to ensure that the resulting demand signal used to drive the execution environment does not require significant amounts of manual effort.

Do you possess the necessary data to initiate demand sensing? Most likely, yes. Ship-to information regarding distribution, replenishment, and sales constitutes key data feeds, along with corresponding line-order details. While many manufacturing and distribution companies lack retail Point of Sale (POS) data (sell-out), utilizing sell-in data at the granular level of line orders has been found to significantly enhance forecast accuracy.

To be concluded...