Ryan Somers, Sr Manager, Product Strategy Retail Planning and Supply Chain | January 12, 2023
To break through the noise of unpredictable supply and the challenges of new demand patterns, retailers are turning to machine learning (ML) and other applications of artificial intelligence (AI) to improve forecast accuracy with as few touches as possible. Demand forecasting, traditionally managed by multiple teams, may be not only inefficient but now also increasingly harder to maintain. Talent is hard to come by, and myriad factors—from the way consumers shop to what they are buying—have diverged from historical patterns. ML helps retailers predict changes in sales based on previous trends and customer preferences. Machine learning constantly analyzes data to find these patterns, and other AI tools translate this data into actionable insights with a larger impact.
Consider a day in the life of a typical retail planner that manages a wide range of products across hundreds of locations. Now imagine that with today’s supply chain mess and consumers’ need for instant gratification. AI can help resolve these mounting challenges. With an intelligent demand forecasting solution, such as Oracle Retail Demand Forecasting Cloud Service, a single retail planner can now manage multiple categories. For example, they can manage fast-moving, small, or seasonal items such as glasses alongside slow-moving, high-ticket items such as couches. The planner gains a single view of all inventory which they can revise and optimize with the full assortment in mind.
Forecast-driven inventory starts with a data-driven demand forecast, which serves as the foundation and basis of inventory planning. With the right forecast, inventory can be optimized for the right channel at the right time, based on predicted demand. Working off the demand forecast, a retailer can incorporate sales, receipts, and returns forecasts based on customer behavior and transaction-level insights. This allows the retailer to optimize the product lifecycle, manage safety stock, and rebalance inventory. Finally, the retailer’s time-phased inventory plan should be directly linked to how customers choose fulfillment services – and driven by optimized profit and service levels.
Considering how much of a planner’s time each day is currently spent revising forecasts and coordinating with planners for other departments, an optimized, cohesive forecast offers a major change, allowing planners to manage by exception, freeing up time for the planner and requiring fewer touches than before.
The right demand forecasting solution enables retailers to take actionable steps in analyzing demand transference. Machine learning collects and analyzes customer preferences to produce insights that inform assortment and availability of stock.
Streamlining hardline inventory is crucial to maximizing space and profits. An appliance and electronics retailer previously saw planners touching 50% of forecasts, but when using Oracle Retail Demand Forecasting, with no touches, the retailer noticed:
In trying times of staff shortages and evolving consumer preferences, predictive and prescriptive models can help retailers make small changes that yield major results. With the right technology, retailers will find even small tweaks, like a reduction in safety stock, can have a big impact on improved inventory management and store operability with little to no changes in sales—a huge cost-saving measure.