The shelf is a dynamic environment, where shoppers select items purposefully and on impulse, where store owners showcase products to entice customers in the store and online.
Inventory management does not just entail having the right stock but also ensuring said stock was effectively sourced, stored, and sold at the right price, at the right time.
A customers first impression of a store determines whether they will make a purchase. A retailer must therefore optimize shelf space planning and product placement.
Space planning with machine learning (ML) enables retailers and customer packaged goods (CPG) manufacturers to easily blend statistical modeling with customer behavior and key business drivers for shelf layouts to satisfy your most valuable customers. Retailers can do this while adhering to inventory rules, physical constraints, and merchandising strategy. With a good model, the store owner can minimize the risk of ineffective inventory management and allow their business to continue to thrive.
ML gathers real-time data to optimize store layout, improve stock predictions, and reduce the risk of loss.
For a good ML model, you should have data on historical sales including demand (number of sales) and item promotions. In addition you should have data of stock costs such as holding costs and stock-out costs. Moreover, personal data on your customers, such as customer demographics, is essential.
External data like signals about competitor prices and public events (like holidays) should appear in your ML solution. This solution can account for stock-outs, inventory turns, and holding costs to properly balance under-buy and over-buy risks. Furthermore, macroeconomic factors like the stock market can impact the consumption habits in the short term as well as the demand of a product, therefore affecting shelf planning.
Weather is a major determinant to what consumers buy, as well as where and when they do their shopping, so you should know take it under consideration in your shelf planning.
There are a fair number of challenges in effective shelf planning. Firstly, consumer online behavior doesn’t match their offline behavior and, especially during Coronavirus lockdowns, that online behavior is changing. Secondly, the frequently changing issue of international tariffs can effect shelves before business owners can prepare.
Further, predicting shelf demand for seasonal items has always been especially difficult.
Meanwhile, there are challenges with the AI programs themselves. These systems require large amounts of data to be effective and uploading inventory data—which is not universal—is very difficult.
There is also the problem of needing to weigh inventory data differently depending on the type. For example, some items sell predictably while others do not but which stores must have a constant supply in store.
Opex Analytics: Machine Learning for Inventory Optimization
Remi AI: Artificial Intelligence for Inventory Management
A McKinsey report found that AI-enabled [Shelf Planning] management can reduce forecasting errors by up to 50%, while helping businesses scale back inventory by up to 50% and slash lost sales by up to 65%.
Practical Machine Learning: How Data Can Optimize Inventory Forecasting