“They aren’t perfect, but they facilitate collaboration by everyone to accomplish the goal.”
- Demand Forecasting,
- Inventory Optimization, and
- Price Elasticity
Supply chain planning relies heavily on science and algorithms for better demand forecasting and inventory management. This includes forecasting independent customer demand, which changes with customer preferences. It also includes dependent demand that aggregates demand streams at a stocking point upstream to other echelons of the supply chain.
Let’s look at how science and machine learning in Sales & Operations Planning (S&OP) is streamlining demand forecasting, inventory optimization and price elasticity to improve long-/short-term planning for complex distribution supply chains.
Why is the Distribution Supply Chain So Complex?
Many factors are contributing to increasing complexity in these forecasts, including outbound customer demand volatility, managing multiple suppliers with long lead times (especially overseas), constrained transportation capacity, and increasing inbound supply line volatility.
Distributors face even greater complexity in managing their supply chains. Large assortments to serve existing customers has become key to differentiation, but introduces an ever larger assortment with intermittent, lumpy and sparse demand in the “long tail.”
Multiple distribution centers serving a demanding customer base with increased expectations of rapid and reliable fulfillment introduces further complexity to the multiple echelons of inventory. Finally, increasing sales on mobile, ecommerce, and other direct channels is growing and competing with traditional channels for inventory and adds to inventory volatility across channels and echelons.
Sophisticated Science and Analytics
One solution to address this complexity is advanced analytics that can handle large data sets down to transaction-level details. Demand classification is key to identifying the type of demand for each product on each channel to apply the best forecasting techniques.
Advanced seasonality techniques are critical for identifying recurring seasonal trends, de-seasonalizing demand, and determining baseline demand. This supports flagging peaks in demand above baseline that can be associated with events like promotional activities, non-recurring natural incidents, and other factors driving demand forecast anomalies.
Machine Learning Tools for Inventory Optimization
Sophisticated science also improves inventory optimization. Machine learning is becoming a more significant addition to the demand forecasting toolkit, and is being widely used to optimize inventory forecasting accuracy.
Price Elasticity and Price Optimization
A crucial piece of the supply chain planning puzzle is price optimization. Since distributors face increasing transparency and competition, there is growing price sensitivity across channels and customer segments. Price elasticity must use similar techniques to demand forecasting in de-seasonalizing demand to determine baseline and measure response to price changes.
A proven practice for using elasticity of demand to measure willingness-to-pay is to measure price sensitivity for products sold on all channels and locations down to the customer level.
Another critical practice is using product attributes to infer price elasticity deeper into the assortment using Bayesian inference and other techniques. Remember analyzing large data sets down to the transaction level? Invoice-level data and analysis is part of this approach.
“Information is the oil of the 21st century, and analytics is the combustion engine.”
Long- and Short-Term Supply Chain Planning
Short Term: Sales & Operations Execution (S&OE)
Demand forecasting, Multi-Echelon Inventory Optimization (MEIO), and Price Optimization science certainly complement each other in tactical, shorter term (12-18 month) supply chain planning. I refer to the concept of short-term SCP + Pricing = Demand Shaping as “Sales & Operations Execution (S&OE)”. But what about using these advanced techniques in longer-term S&OP?
Longer-Term: Sales & Operations Planning (S&OP)
There is indeed a role for science in S&OP more long-term. Pricing scenarios can consider projected costs, including potential tariffs for future periods, to compare financial impacts and choose the right approach by pricing segment and customer segment. Price elasticity is a key component in recommending better prices in different scenarios to test and select the best way to achieve desired financial objectives for future periods. These science-driven decisions feed directly into S&OP processes as alternatives for evaluation.
Inventory Optimization for Alternative Trade-Offs
Inventory optimization can also be applied to consider alternative supply chain configurations, such as alternate distribution center locations, evaluating forward buying opportunities in advance of supplier cost changes, and other long-term inventory and service level trade-offs.
An unconstrained forecast from pricing systems isn’t enough. Supply chain and inventory constraints must be considered to evaluate alternate plans. Strategic planning organizations spend more time evaluating planning decisions to understand the optimal way to invest in inventory to maximize customer service and free working capital.
IBP: The Next-Gen S&OP Solution
Demand forecasting also plays a key role in long-term S&OP processes. Integrated Business Planning (IBP), the next generation of S&OP solutions, extends S&OP processes from within a company’s four walls to include downstream customers. These customer-level demand forecasts provide additional inputs to drive to a consensus forecast.
Effective IBP also includes upstream suppliers to provide better insight into planned demand changes to feed rough-cut capacity planning; production scheduling and other sourcing; assembly; and manufacturing processes. Once a consensus long-term forecast is chosen, direct integration with tactical demand forecasts is critical to ensure demand planning and purchase order recommendations are updated. This is especially critical for communications to suppliers with extremely long lead times to avoid inventory shortfalls.
Modern Tools Support Collaborative Planning and Forecasting
These three key scientific techniques combine to improve IBP process outcomes and supplement collaborative and effective sales operations planning across a company, downstream to customers, and upstream to suppliers. This is especially important in the face of increasing supply chain complexity. Distributors face an even greater challenge as they increase assortment sizes while combating the “Amazon effect” rapidly entering distribution.
Data science in S&OP is never perfect, but it supplements strong collaborative planning and forecasting, and arms the entire organization with modern tools to give them the best fighting chance.