— Dirk Niederhäuser – Supply Chain Management Practice, IBM Consulting
— Vikram Murthi – VP Industry Strategy, o9 Solutions
The worldwide COVID-19 crisis brought many supply chain challenges that were percolating to a full boil. The shock to the supply chain that the pandemic created has led to a new environment that will determine tomorrow’s retail winners. However, the pandemic is only the latest disruptor in an already risky global retail environment.
Business challenges like sourcing from Asia with long and uncertain lead times now combine with demand uncertainty, which came to the forefront with the pandemic raging. And now demand is volatile and shifting again with consumers going back to restaurants and traveling, which will likely affect sales at grocery chains adversely but give a much-needed boost to services like hospitality and dining.
Over the past few decades, retail supply chains were built for operational excellence, not agility. The objective was to deliver products to consumers at the lowest possible costs, with standardized assortments produced at the cheapest locations and shipped in the most efficient quantities. However, consumers’ shopping behavior is radically changing in terms of what they buy, how they buy, and how they want products delivered to them. This combined with the COVID-19 impacts on demand and supply uncertainty means that retailers are dealing with a witch’s brew of factors.
To survive and thrive in this age of the never normal, retailers must increasingly focus on getting a better handle on their demand and place inventory optimally through efficient replenishment. In this first of two blogs, we will cover the need for demand forecasting excellence. In the second blog, we will discuss best practices in replenishment and inventory management.
Robust Demand Forecasting is Critical for an Agile Supply Chain Response
To respond to changing consumer buying patterns quickly, forecasting techniques that leverage demand sensing capabilities are being deployed. Demand sensing focuses on eliminating supply chain lags by continuously learning and reducing the time between demand signals (order frequency, order size, local events, DC/store inventory, POS, etc.) and the response to those signals. Sensing and forecasting demand at a lower level of granularity, by store or by consumer zip code, is emerging as the best way to truly understand and incorporate local consumer demographics and drive inventory placement to fulfill that demand.
Market knowledge can improve forecast accuracy
Retailer planning teams are finding a treasure trove of external market data like COVID-19 infection rates, mobility indices (Google, Apple), demographics and macroeconomic information, that could be used as drivers to explain demand patterns, and also improve the forecast accuracy.
Retail forecasting competencies and systems are rapidly changing to incorporate market knowledge through publicly available data on consumer demographics, macroeconomic indicators such as Gross Domestic Product (GDP) and interest rates, social media buzz, and global trade. Other leading indicators of demand like news, product reviews, search engine statistics, and website glance views are becoming increasingly important as demand sensing levers.
No longer a secret sauce – AI/ML methods improve forecast accuracy
Apart from traditional time-series statistical methods, newer Machine Learning (ML) techniques enable demand sensing with pattern recognition and improve the accuracy of forecasts across all channels in several ways with the use of newer algorithms like Gradient Boosting, Support Vector Machines, Tournament and Ensemble methods and leveraging internal drivers like everyday shelf price, sale price, product placement, offers, digital coupons, and incorporating a host of external casuals like weather, GDP, new housing starts, mobility indices, local school and sports events, interest rates, inflation, and debt to income ratios.
Pricing and promotions
Retailers are constantly trying to shape demand through promotions and campaigns at both the store and online channels. They resort to various in-store promotions with temporary price discounts, displays, and feature inserts in local publications. There are also omnichannel demand shaping activities such as placement on the website, special offers like free shipping, price reductions, email offers, digital coupons, and social media campaigns which drive incremental sales.
Robust modeling of these demand shaping activities can greatly benefit from ML techniques and there has been considerable success in using ARIMAX based algorithms. Category managers can run “what-ifs” to look at the impact of changing the timing and duration of promotions, try different product placement strategies on the feature insert or the website, different levels of price discounts, free shipping, and understand the impact on in-store sales or online orders. What-if scenario planning is a critical capability that retailers use to evaluate multiple options for single or sets of promotions, to understand and select those that achieve category and regional goals.
Uncertainty and scenario analysis
Promotions are not the only reason to do scenario analysis. For example, retailers may want to know what sales will look like in the event of another lockdown or how to manage inventory if there is a severe supply disruption due to port congestion. Unforeseen events can trigger higher or lower demand for certain products and be modeled with scenarios for the probability of likelihood.
o9 Solutions offers a proven platform for retailers to leverage rich market and consumer knowledge and apply ML models to create accurate and granular forecasts, to drive inventory placement decisions.
Download our white paper Digital Transformation of Retail Inventory Management to learn how retailers:
- Need to adapt and upgrade their supply chain competencies in demand forecasting and replenishment to stay competitive.
- Drive their supply chains with forecasts that consider category nuances driven by market knowledge.
- Make inventory placement decisions based on store and DC labor/transportation/storage capacities, operational schedules, and promotions.
- Enable store order decision-making with automation at scale and explainable recommendations.